MEDICON’23 and CMBEBIH’23: Proceedings of the Mediterranean Conference on Medical and Biological Engineering and Computing (MEDICON) and International ... in Healthcare (IFMBE Proceedings, 94) 3031490673, 9783031490675

Citation preview

IFMBE Proceedings 94

Almir Badnjević Lejla Gurbeta Pokvić Editors

MEDICON’23 and CMBEBIH’23 Proceedings of the Mediterranean Conference on Medical and Biological Engineering and Computing (MEDICON) and International Conference on Medical and Biological Engineering (CMBEBIH), September 14–16, 2023, Sarajevo, Bosnia and Herzegovina—Volume 2: Bio-innovations, Sustainable Practices, and Multidisciplinary Applications in Healthcare

IFMBE Proceedings Series Editor Ratko Magjarevi´c, Faculty of Electrical Engineering and Computing, ZESOI, University of Zagreb, Zagreb, Croatia

Associate Editors Piotr Łady˙zy´nski, Warsaw, Poland Fatimah Ibrahim, Department of Biomedical Engineering, Faculty of Engineering, Universiti Malaya, Kuala Lumpur, Malaysia Igor Lackovic, Faculty of Electrical Engineering and Computing, University of Zagreb, Zagreb, Croatia Emilio Sacristan Rock, Mexico DF, Mexico

94

The IFMBE Proceedings Book Series is an official publication of the International Federation for Medical and Biological Engineering (IFMBE). The series gathers the proceedings of various international conferences, which are either organized or endorsed by the Federation. Books published in this series report on cutting-edge findings and provide an informative survey on the most challenging topics and advances in the fields of medicine, biology, clinical engineering, and biophysics. The series aims at disseminating high quality scientific information, encouraging both basic and applied research, and promoting world-wide collaboration between researchers and practitioners in the field of Medical and Biological Engineering. Topics include, but are not limited to: • • • • • • • • • •

Diagnostic Imaging, Image Processing, Biomedical Signal Processing Modeling and Simulation, Biomechanics Biomaterials, Cellular and Tissue Engineering Information and Communication in Medicine, Telemedicine and e-Health Instrumentation and Clinical Engineering Surgery, Minimal Invasive Interventions, Endoscopy and Image Guided Therapy Audiology, Ophthalmology, Emergency and Dental Medicine Applications Radiology, Radiation Oncology and Biological Effects of Radiation Drug Delivery and Pharmaceutical Engineering Neuroengineering, and Artificial Intelligence in Healthcare

IFMBE proceedings are indexed by SCOPUS, EI Compendex, Japanese Science and Technology Agency (JST), SCImago. They are also submitted for consideration by WoS. Proposals can be submitted by contacting the Springer responsible editor shown on the series webpage (see “Contacts”), or by getting in touch with the series editor Ratko Magjarevic.

Almir Badnjevi´c · Lejla Gurbeta Pokvi´c Editors

MEDICON’23 and CMBEBIH’23 Proceedings of the Mediterranean Conference on Medical and Biological Engineering and Computing (MEDICON) and International Conference on Medical and Biological Engineering (CMBEBIH), September 14–16, 2023, Sarajevo, Bosnia and Herzegovina—Volume 2: Bio-innovations, Sustainable Practices, and Multidisciplinary Applications in Healthcare

Editors Almir Badnjevi´c Verlab Research Institute for Biomedical Engineering Medical Devices, and Artificial Intelligence Sarajevo, Bosnia and Herzegovina

Lejla Gurbeta Pokvi´c Verlab Research Institute for Biomedical Engineering Medical Devices, and Artificial Intelligence Sarajevo, Bosnia and Herzegovina

ISSN 1680-0737 ISSN 1433-9277 (electronic) IFMBE Proceedings ISBN 978-3-031-49067-5 ISBN 978-3-031-49068-2 (eBook) https://doi.org/10.1007/978-3-031-49068-2 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Paper in this product is recyclable.

Preface

Dear colleagues, It is with immense gratitude and delight that we look back on the successful joint event of the 16th Mediterranean Conference on Medical and Biological Engineering and Computing (MEDICON) and the 5th International Conference on Medical and Biological Engineering (CMBEBIH), which took place in the city of Sarajevo, Bosnia and Herzegovina, from September 14th to 16th, 2023. During those three unforgettable days, we had the privilege of witnessing engaging keynote speeches by esteemed experts in the field. We were thrilled to see attendees actively participating in scientific and industry sessions, passionately contributing to interactive panel discussions, and connecting with peers from diverse backgrounds and expertise. This conference served as an exceptional platform, uniting researchers, scientists, engineers, professionals, and public sector representatives from various disciplines within the medical and biological engineering field. The meticulously curated program of the conference encompassed a wide range of captivating topics, focusing on the transformative power of digitalization in healthcare and the environment. The introduction of digital innovation hubs proved to be a resounding success, enabling interactive demonstrations and fostering collaboration among attendees. Additionally, the “Meet the Editor” session provided valuable insights into publishing research, leaving a lasting impact on the attendees. We are truly humbled by the active participation and enthusiastic engagement of all those who attended, making this conference a resounding success. The networking opportunities that were availed allowed for fruitful connections, idea exchanges, and the forging of new collaborations that have the potential to shape the future of medical and biological engineering. As we reflect on the conference, we are filled with profound gratitude for the shared experiences, the sparks of curiosity ignited, and the passion for innovation that was palpable throughout the event. It is our sincere hope that the knowledge gained and the connections made will continue to foster progress in the field of medical and biological engineering. We hope that this book, which compiles the valuable contributions and insights presented during the conference, will be a valuable addition to the advancement of biomedical engineering. Through sharing these ideas and research findings, we aspire to further propel the field’s growth and pave the way for future innovations and breakthroughs.

vi

Preface

Once again, we express our heartfelt thanks to all participants, speakers, sponsors, and the organizing team for making this conference an unforgettable experience. We are honored to have had each of you contribute to its success. Prof. Dr. Lejla Gurbeta Pokvi´c Prof. Dr. Almir Badnjevi´c Conference Chairs

Contents

Molecular, Cellular and Tissue Engineering Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine on Nerve Tissue During Peripheral Nerve Blockades . . . . . . . . . . . . Zurifa Ajanovi´c, Lejla Derviševi´c, Ilvana Hasanbegovi´c, Amela Derviševi´c, Almira Lujinovi´c, Maida Šahinovi´c, Jasmina Bišˇcevi´c-Toki´c, Džan Ahmed Jesenkovi´c, and Emina Derviševi´c Skin Substitutes: An Overview of Current State of the Art . . . . . . . . . . . . . . . . . . . Nina Kocivnik Evaluation of Platelet-Enriched Plasma Antimicrobial Effect: In Vitro Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tea Be´cirevi´c, Izet Eminovi´c, Nadira Ibrišimovi´c Mehmedinovi´c, Edin Omeragi´c, Edin Falan, Ermin Papraˇcanin, and Mirza Ibrišimovi´c The Influence of Popular Antibiotics in Culture Medium on the Cell Motility and Superoxide Production of Mesenchymal Stem Cells . . . . . . . . . . . . . Inna Zumberg, Larisa Chmelikova, and Vratislav Cmiel Treatment Planning for Electrochemotherapy of Spinal Metastases . . . . . . . . . . . Helena Cindriˇc, Damijan Miklavˇciˇc, and Bor Kos

3

14

22

29

37

Bio-micro/nano Technologies Development of Polymer-Based Nanoparticles for the Reduction of Melittin Toxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Berrin Chatzi Memet, Eren Demirpolat, Turgay Yildirim, and Omer Aydin

45

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both Biosynthesizing Silver Nanoparticles and Anti-microbial Activity . . . . . . . . . . . . Berna Oyku Ozbey and Gulizar Caliskan

51

Metrology in Medical Measurements Lessons Learned from External Audits in Medical Device Testing Laboratories: Best Practices and Recommendations for Quality Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Baki Karaböce

65

viii

Contents

Implementation of Legal Metrology Framework for Medical Devices to Healthcare Sector in the Republic of Uzbekistan . . . . . . . . . . . . . . . . . . . . . . . . . Vohobjon Nishonov, Lemana Spahi´c, Amar Deumi´c, Ammar Traki´c, Najmiddin Muminov, Sheroz Ismatullev, and Lejla Gurbeta Pokvi´c Performance Test and Calibration in End Tidal Carbon Dioxide Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ˙ Mana Sezdi and Nazif Ilker Sezdi A New and Fast Approach for Antimicrobial Resistance Detection: Combination of Artificial Intelligence and Surface-Enhanced Raman Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Omer Aydin, Zakarya Al-Shaebi, Munevver Akdeniz, ˙ Gizem Kursunluoglu, Gokmen Zarasız, Serra Ilayda Yerlitas, Ahmet Sezgin, Mustafa Altay Atalay, and Pınar Sagiroglu

74

91

98

Metrology Versus Medical Metrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Baki Karaböce The Effect of Different Final Irrigant Activation Techniques on the Bond Strength of an MTA-Based Endodontic Sealer- An in Vitro Study . . . . . . . . . . . . . 112 Sagar Borse, Anita Sanap Tandale, Sanjyot Mulay, Vini Mehta, and Karishma Krishnakumar Verification of Ultrasound Imaging Phantoms: An Evaluation Study . . . . . . . . . . 120 Baki Karaböce and Hüseyin Okan Durmu¸s Clinical Engineering and Health Technology Assessment Development of a Failure Prediction Strategy for Imaging Systems . . . . . . . . . . . 135 Dario Léon Merten, Dubravka Maljevic, and Markus Buchgeister OHIO - Odin Hospital Indoor Compass for Empowering the Management of Hospitals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Alessio Luschi and Ernesto Iadanza Calculating the Required Mammography Machines for Breast Cancer Screening in Mexican States with High Incidence Rates: A Proposed Model . . . 150 Fabiola M. Martinez-Licona Establishing the Optimal Standard for Preprocessing Head CT Data in Diagnostic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 Petra Nemcekova, Tomas Holecek, Jiri Chmelik, Petr Ourednicek, Katerina Valis, and Roman Jakubicek

Contents

ix

Standardization of Failure Codes and Nomenclature of Medical Devices for Evidence-Based Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 Ernesto Iadanza and Alessio Luschi Advanced Diabetes Technology for Better Glucoregulation, Opportunities and Cost Benefit (“Review on the Reality of a Developing Country”) . . . . . . . . . . 178 ˇ ˇ Alma Badnjevi´c-Cengi´ c, Amila Cerim-Aldobaši´ c, Mubina Hodži´c, and Davorka Dautbegovi´c-Stevanovi´c Prevalence of Electronic Gadgets Usage and Its Impact on Sleep Patterns Amongst 7–10-Year-Old Children During Covid-19 Pandemic: A Cross-Sectional Observational Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Sejal S. Shah, Nilesh Rathi, Y. Nankar Meenakshi, and Vini Mehta Gender Impact Assessment for Medical Devices: A Compass to Find the Way in the Gender-Technology Reciprocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Manuela Appendino, Maria Agnese Pirozzi, Rossella Tomaiuolo, Veronica Moi, and Luca Radice Open-Source Medical Device for in Vitro Diagnosis of Malaria . . . . . . . . . . . . . . 208 Florinda Coro, Andrea Arcangeli, Carmelo De Maria, Valentina Mangano, and Arti Ahluwalia Assessing the Baseline Impact of Agile Human Resource Management in the Healthcare Systems of Western Balkan Countries . . . . . . . . . . . . . . . . . . . . . 215 Sandra Tinaj, Milica Vukotic, Bojana Malisic, and Lidija Lukovac Genetic Engineering Review Paper: Autism Spectrum Disorder—Molecular Mechanisms and Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Selma Cifri´c Mujezinovi´c and Dado Latinovi´c Analysis of MicroRNAs in Correlation to Astrocytoma . . . . . . . . . . . . . . . . . . . . . 236 Lejla Kadri´c, Dina Neiroukh, Johannes Wagner, and Aida Hajdarpaši´c How Inherited Thrombophilia Affects Success Rate of IVF Treatment in Women . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 Damilola M. Ajayi and Emmanuel Ajayi Examining the Therapeutic Potential of Stem Cells in Treatment of Infertility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 Lejla Hadži´c, Sara Sejdi´c, and Faruk Guti´c

x

Contents

An Overview of Personalized Medicine Development Through Recent Advances in Genome-Wide Association Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Dženita Omerki´c and Adna Aši´c Neural and Rehabilitation Engineering Association Between Variations in Kinematic Indexes of Manual Dexterity and Mu Rhythm Desynchronization Changes After Action Observation and Motor Imagery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Federico Temporiti, Alessandra Calcagno, Stefania Coelli, Giorgia Marino, Roberto Gatti, Anna Maria Bianchi, and Manuela Galli Handedness-Dependent Brain Networks Re-organization During Visuo-Motor Task Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 Alessandra Calcagno, Stefania Coelli, Federico Temporiti, Roberto Gatti, Manuela Galli, and Anna Maria Bianchi Brain-Computer Interface Through the Prism of Modern Age . . . . . . . . . . . . . . . . 292 Amina Radonˇci´c, Semina Hadži´c, and Jasmina Lakovi´c IFMBE Young Investigator Competition for Biomedical Engineering Restoring the Arterial Tree in Sepsis: A Neglected Therapy Target . . . . . . . . . . . . 327 Marta Carrara, Diletta Guberti, Stephan Jakob, and Manuela Ferrario A New Algorithm for Arterial Inflection Point Identification in Critically Ill Patients: A Preliminary Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 Marta Carrara, Riccardo Campitelli, Antoine Herpain, and Manuela Ferrario Hemodynamic Cardiovascular Indices to Predict the Response to Angiotensin-II in Septic Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 Marta Carrara, Bruno Garcia, Antoine Herpain, and Manuela Ferrario Assessing Left Ventricle Radiomic Features Robustness by Segmentation Perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356 Francesca Lo Iacono, Gianluca Pontone, and Valentina D. A. Corino Assessing the Performance of MRI-Radiomic Prognostic Signatures in Head and Neck Cancer Patients: A Comparative Analysis . . . . . . . . . . . . . . . . . 363 Anna Corti, Luca Mainardi, and Valentina D. A. Corino

Contents

xi

In Vitro Electrochemotherapy Experiments to Quantify the Number of Cisplatin Molecules Needed for a Cytotoxic Effect When Different Types of Pulses Are Delivered . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 ˇ Maria Scuderi, Janja Dermol-Cerne, Janez Šˇcanˇcar, Stefan Markovi´c, Lea Rems, and Damijan Miklavˇciˇc Spectrogram-Driven Convolutional Neural Network for Real-Time Non-invasive Hyperglycaemia Detection in Paediatric Type-1 Diabetes via Wearable Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376 Owain Cisuelo, Muhammad Salman Haleem, John Hattersley, and Leandro Pecchia Spectrogram-Based Approach with Convolutional Neural Network for Human Activity Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387 Martina Sassi, Muhammad Salman Haleem, and Leandro Pecchia Biosensors and Bioinstrumentation Effect of Cardiolite on Enzyme Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405 Ajla Selimovi´c, Safija Herenda, Edhem Haskovi´c, Mirjana Ðermanovi´c, and Emina Opankovi´c Evaluation of Printed Coplanar Capacitive Sensors for Reliable Quantification of Fluids in Adult Diaper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414 Muhammad Tanweer, Liam Gillan, Raimo Sepponen, Ihsan Oguz Tanzer, and Kari A. Halonen Biosensor for Rapid Methods for the Detection of Viruses . . . . . . . . . . . . . . . . . . . 423 Sara Deumi´c, Aida Lavi´c, Neira Crnˇcevi´c, Emina Pramenkovi´c, and Amar Deumi´c Implantable and Ingestible Biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430 Neira Crnˇcevi´c, Damilola M. Ajayi, Tarik Zubˇcevi´c, Sara Deumi´c, and Haris Koli´c Characteristics of Ocular Following Responses (OFRs) in Children with Stereodeficiencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438 Aleksandar Miladinovi´c, Christian Quaia, Miloš Ajˇcevi´c, Laura Diplotti, Simone Kresevic, Stefano Pensiero, and Agostino Accardo Influence of Tilt Angle and Probe-Sample Distance on Tissue Diagnosis by Diffuse Reflection Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447 Sonia Buendia-Aviles, Margarita Cunill-Rodríguez, José A. Delgado-Atencio, José L. Arce-Diego, and Félix Fanjul-Vélez

xii

Contents

Biomechanics, Robotics and Minimally Invasive Surgery ARTHROprint: A System for the Immediate Restoration of Cartilage Lesions by Implantation of Printable Autologous Cell Scaffolds . . . . . . . . . . . . . . 455 Georgia Peleka, Ioannis Mariolis, Ioannis Kostavelis, Trifon Totlis, Efthymios Papasoulis, Aristotelis Sideridis, and Dimitrios Tzovaras Mysteries of Phytoplanktonic Robotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462 Sumeja Buljugija and Mevludin Maliˇcevi´c Creation of Anthropomorphic Bone Phantoms With Customized Fused Filament Fabrication 3D Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468 Petar Valchanov, Nikolay Dukov, Nikiforos Okkalidis, and Zhivko Bliznakov Bioinformatics and Computational Biology Using Service Robots as the Base Technology of Industry 4.0 for Global Management of the Covid-19 Pandemic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479 Isak Karabegovi´c, Lejla Banjanovi´c-Mehmedovi´c, and Ermin Husak The Utility of RNA Triplex Formation in Autoimmune Disease: Identification of Therapeutic Dual Synergistic MicroRNAs in Systemic Lupus Erythematosus—A Bioinformatics Approach . . . . . . . . . . . . . . . . . . . . . . . . 493 Adna Salihovi´c Identification of the Arrhythmogenic Substrate in Brugada Syndrome: A Computational Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513 Paolo Seghetti, Niccolò Biasi, Matteo Mercati, Valentina Hartwig, Andrea Rossi, Marco Laurino, and Alessandro Tognetti In Silico Closed-Loop System for the Assessment of Cardiac Pacing Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521 Matteo Mercati, Niccolò Biasi, Paolo Seghetti, and Alessandro Tognetti Exploring Psilocybin as a Tool for Studying Parkinsonism-Related Psychosis: A Narrative Review Supplemented with a Computational Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530 ˇ Asja Campara and Dženan Kovaˇci´c

Contents

xiii

New Approaches, Technologies, Materials and Telemedicine in Dentistry Orthodontic Correction of Traumatic Luxations, Disadvantages of Continuous Arch Techniques: A Clinical Protocol . . . . . . . . . . . . . . . . . . . . . . . 551 Benedetta Vaienti, Marco Di Blasio, Marzia Segù, and Alberto Di Blasio A New Model of Herbst Appliance for Young O.S.A.S. Patients . . . . . . . . . . . . . . 559 Marco Di Blasio, Benedetta Vaienti, Diana Cassi, Marco Melegari, and Alberto Di Blasio Redefine the Anterior Limit of the Dentition in Case of Functional Genioplasty: A Clinical Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569 Marco Di Blasio, Benedetta Vaienti, Chiara Di Blasio, Marzia Segù, and Alberto Di Blasio Treatment of an Odontogenic Maxillary Sinusitis due to an Oroantral Opening with Low-Level Laser Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579 Ali Temelci, Erdo˘gan Kıbçak, Gürkan Ünsal, and Giuseppe Minervini Comparative Evaluation of Efficacy of Antibiotics Incorporated Platelet Rich Fibrin Versus Platelet Rich Fibrin Alone in the Treatment of Intrabony Defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586 Shweta Kadam, Anita Kulloli, Sharath K. Shetty, Santosh S. Martande, Gopika Nair, Mariam Poulose, Meghana Guruprasad, and Vini Mehta Evaluation of the Impact of Educational Status on the Anxiety Levels of Patients Undergoing Root Canal Therapy Using Modified Corah Dental Anxiety Scale—A Cross-Sectional study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598 Hmoud Ali Algarni, Meshal Aber Al Onazi, Amjad Obaid Aljohani, Kumar Chandan Srivastava, Deepti Shrivastava, Merin Mathew, and Mohammed Ghazi Sghaireen Prevalence of Occlusal Interferences Among the Saudi Sub-population: A Cross-Sectional Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609 Amjad Obaid Aljohani, Mohammed Ghazi Sghaireen, Deepti Survistava, Mohammed Assayed Mousa, Thamir Ahmed Bahattab, Mohammed Abdulhakim Bafaraj, and Kumar Chandan Srivastava Masticatory System—A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616 Zainab A. Alkhalaf, Mohammed Ghazi Sghaireen, Deepti Survistava, Mohammed Assayed Mousa, Amjad Obaid Aljohani, Vinod Bandela, and Kumar Chandan Srivastava

xiv

Contents

The Role of Telemedicine in Oral and Head Cancer Management . . . . . . . . . . . . . 628 Rocco Franco, Giuseppe Minervini, Maria Maddalena Marrapodi, Salvatore Crimi, Alberto Bianchi, and Marco Cicciù Digital Workflow in the Management of Patients with Temporomandibular Disorders and/or Bruxism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636 Giuseppe Minervini, Salvatore Crimi, Maria Maddalena Marrapodi, Alberto Bianchi, Marco Cicciù, and Rocco Franco The Role of Social Media on Dental Education and Oral Health: A Focus on Instagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645 Rocco Franco, Giuseppe Minervini, Maria Maddalena Marrapodi, Maurizio D’Amario, and Gabriele Cervino Technologies in Temporomandibular Disorders and/or Bruxism Patients Management: Occlusal Splint Construction Performed with Digital Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654 Giuseppe Minervini, Rocco Franco, Francesco Catalano, and Marco Cicciù Technologies and Innovations in Oral Health: The Role of Telemedicine in Orthodontic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661 Rocco Franco, Giuseppe Minervini, Maria Maddalena Marrapodi, Gabriele Cervino, and Marco Cicciù Combined Surgical and Orthodontic Treatment for Eruption of the Impacted Premolar Due to a Dentigerous Cyst . . . . . . . . . . . . . . . . . . . . . . . 668 ˙ Ahmet Özant, Ismet Ersalıcı, Ali Temelci, and Giuseppe Minervini Occlusal Splints in Athletes: A Useful Appliance to Prevent Dental and Temporomandibular Joint Traumas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677 Marco Cicciù, Rocco Franco, Maria Maddalena Marrapodi, Aida Meto, and Giuseppe Minervini New Technologies in Oral Health Education, Patients Motivation, and Patient/Dentist Communication in the Covid-19 Era: The Role of WhatsApp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686 Rocco Franco, Gabriele Cervino, Maria Maddalena Marrapodi, Andreea M. Musat, Marco Cicciù, and Giuseppe Minervini Influence of Temporomandibular Disorders and Bruxism on Prosthodontics Rehabilitations Survival: A Focus on Veneers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695 Gabriele Cervino, Giuseppe Minervini, Saurabh Chaturvedi, Marco Cicciù, and Rocco Franco

Contents

xv

Advancements and Applications in Digital Dentistry: A Scoping Review . . . . . . 702 Pio Genovese, Pietro Giambò, Fulvio Abramo, Maura Mancini, Mariana Pastore, and Cesare D’Amico Standardized Tool for the Assessment of Bruxism (STAB): A New Method to Assess the Temporomandibular Disorder Patients . . . . . . . . . . . . . . . . . . . . . . . . 710 Giuseppe Minervini, Rocco Franco, Mario Capogreco, Vincenzo Ronsivalle, and Marco Cicciù Evaluation of Temporomandibular Disorders Patients Through T-scan System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718 Giuseppe Minervini, Rocco Franco, Maria Maddalena Marrapodi, Marco Cicciù, and Vincenzo Ronsivalle Application of Stem Cells in Dentistry: A Review Article . . . . . . . . . . . . . . . . . . . 726 Dženita Omerki´c Dautovi´c, Belkisa Hodži´c, and Selam Omerki´c Cardiovascular, Respiratory and Endocrine Systems Engineering Evaluation of Electrocardiographic Leads Using Conventional ST-Segment Depression and ST-Segment Depression/Heart Rate Hysteresis for Diagnosing of Coronary Artery Disease in Women Population: The Finnish Cardiovascular Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749 Serkalem D. Beyene and Jari J. Viik Application of Machine Learning in the Prediction of Hypothyreoidism . . . . . . . 756 Hanna Hela´c, Edina Kamenjaš, and Nejira Hodži´c Predictors of Mortality After Index Hospitalisation for Acute Heart Failure–Difference in HFREF and HFPEF Group . . . . . . . . . . . . . . . . . . . . . . . . . . 762 Azra Durak-Nalbanti´c, Edin Begi´c, Alen Džubur, Alden Begi´c, Almir Badnjevi´c, Damir Rebi´c, Aida Hamzi´c-Mehmedbaši´c, Nafija Serdarevi´c, Mirela Halilˇcevi´c, Amer Iglica, Nerma Resi´c, Orhan Lepara, Nermina Bešli´c, Refet Gojak, and Ena Gogi´c Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773

Molecular, Cellular and Tissue Engineering

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine on Nerve Tissue During Peripheral Nerve Blockades Zurifa Ajanovi´c1(B) , Lejla Derviševi´c1 , Ilvana Hasanbegovi´c1 , Amela Derviševi´c2 , Almira Lujinovi´c1 , Maida Šahinovi´c3 , Jasmina Bišˇcevi´c-Toki´c4 , Džan Ahmed Jesenkovi´c5 , and Emina Derviševi´c6 1 Department of Anatomy, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and

Herzegovina [email protected] 2 Department of Physiology, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and Herzegovina 3 Department of Histology, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and Herzegovina 4 Department for Occupational Medicine, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and Herzegovina 5 Department of Epidemiology, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and Herzegovina 6 Department of Forensic Medicine, Medical Faculty University of Sarajevo, 71000 Sarajevo, Bosnia and Herzegovina

Abstract. Peripheral nerve blocks play an important role in achieving analgesia in patients undergoing various surgical procedures. Knowledge of the functional histology of nerves is essential for understanding the mode of action of local anesthetics and the mechanism of nerve damage. Geometric morphometrics is a method which we can used to analyse the morphological changes of examined structures. The main objective of our research was to explore the potential nerve tissue damage after intraneural and perineural application of liposomal bupivacaine using geometric morphometrics methods, by assessing the change in the shape of the axons and nerve fibers. Material and method: The sample in this study consisted of sixty sciatic nerves, randomized into four groups to receive intraneural 4mL of liposomal bupivacaine or physiological solution or perineural 4mL of liposomal bupivacaine or physiological solution. Nerve samples were taken and stained with the Azan method. Using a light microscope connected to a digital camera, pictures were made, after which geometric morphometrics was applied. Five nerve fibers from five parts of the image of the microscopic preparation were used for analysis (center, upperleft, upper-right, lower-left, and lower-right parts of the image), and eight specific points (landmarks) were marked. The file was imported into the MorphoJ program for further analysis of the shape of the two-dimensional models. Results: A statistically significant difference in nerve fiber shape was observed after the perineural application of liposomal bupivacaine and the intraneural application of physiological solution (p = 0.014). A statistically significant difference © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 3–13, 2024. https://doi.org/10.1007/978-3-031-49068-2_1

4

Z. Ajanovi´c et al. in axon shape was recorded after the intraneural application of the liposomal bupivacaine and intraneural application of the physiological solution (p = 0.0061). Conclusion: When applied in a clinically controlled environment and in appropriate doses liposomal bupivacaine will not pose significant changes in the shape of nerve fibers or axons. Keywords: Geometric morphometrics · Liposomal bupivacaine · Neurotoxicity · Nerve fibers

1 Introduction Peripheral nerve block is widely used in surgery in patients undergoing various surgical procedures [1]. The latest approach to achieving a long-lasting effect of local anesthesia is the encapsulation of local anesthetics, which enables the use of large doses of local anesthetics that are released gradually, and prolonged analgesia is ensured. Liposomal bupivacaine is a formulation of bupivacaine with extended release, that enables drug diffusion up to 72 h after a single application. However, in everyday clinical practice, the use of liposomal bupivacaine for peripheral nerve blockades is still less popular due to insufficiently researched potential toxic effects on nervous tissue[2]. A nerve fibre is a complex neurophysiological structure that includes the axon of any neuron together with the glial sheath. Knowledge of the functional histology of nerves is essential for understanding the mode of action of local anesthetics and the mechanism of nerve damage. The anatomical specificities of the nerve and how its protective structure against external injuries is organized is perhaps the most important factor that determines the probability of injury [3]. Histopathological diagnosis and grading of cellular changes have important implications for designing improved therapeutic strategies. Despite the existence of advanced genetic, biochemical, radiological, and nuclear imaging tools, the gold standard for typing and grading is the evaluation of histological specimens using classical micro-scopes, which relies on subjective assessment of the degree of cellular change[4]. Geometric morphometry is a new methodology used to analyze the shape of the examined structures. Based on specific points (landmarks) that are marked on the examined structures in a precisely defined order, it is possible to analyze even the smallest morphological changes in the examined structure. The position of specific points on the surface of the examined structure is defined by the x and y-axis values in the coordinate system for two-dimensional models of the examined structures, or by the x, y, and z axis values for three-dimensional models. Based on the above values, in geo-metric morphometry, it is possible to analyze the differences in the position of the same specific points on all the structures of the tested sample, as well as the mutual relationship of different specific points on the same structure of the tested sample. Ultimately, the geometric morphometry tests included in the specially created pro-grams analyze whether there is a statistically significant difference in the morphological characteristics of the examined structures. Precisely because of the above, geo-metric morphometry can be used as a new methodology in examining the qualitative characteristics of nerve fibers after performing peripheral nerve blocks with liposomal bupivacaine [5–7].

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine

5

The present study aims to explore the potential nerve tissue damage after intraneural and perineural application of liposomal bupivacaine using geometric morphometric methods. Specifically, we addressed the potential morphological changes in the shape of the axons and nerve fibers.

2 Material and Methods The study was conducted at the Veterinary Faculty of the University of Sarajevo, Bosnia and Herzegovina, after obtaining approval from the Ethical Committee of the Veterinary Faculty University of Sarajevo. Animal handling complied with the Principles of Care and Preservation of Laboratory animals [8]. The sample in this study consisted of sixty sciatic nerves, that were taken from thirty Wistar rats. Nerves were randomized into four groups, each containing fifteen nerves: PSI–0.9% intraneural physiological (saline) solution; PSP–0.9% perineural physiological solution; LBI–1.33% intraneural liposomal bupivacaine; LBP–1.33% perineural liposomal bupivacaine. For each application 4 mL of tested solution was used. Before the nerve extraction procedure itself, rats were put to sleep by intraperitoneal general anesthesia, as in previous research [9]. After that, sciatic nerves on both sides (right and left) were exposed, and the corresponding tested liquid (1.33% liposomal bupivacaine or 0.9% physiological solution) was applied under eye control, using a 26 G needle (Terumo Europe NV, Leuven, Belgium). An automatic pump (PHD 2000; Harvard Apparatus, Holliston, MA, USA) was used for the application. After the applications of the tested liquids were completed, the wounds were sutured in the gluteal area, and animals were followed for recovery. After monitoring the animals for three days, the animals were euthanized. Nerve samples were taken and processed by standard histological procedures [10] and stained with the Azan method. After obtaining the appropriate sections and preparations, using a light microscope connected with a digital camera, pictures were made. All pictures are taken at a magnification of 40 × and saved as JPG format. A geometric morphometric methodology was applied on digital pictures of microscopic preparations of nerves to examine possible changes in shape. On each examined nerve, five nerve fibers from five parts of the image of the microscopic preparation were used for analysis (center, upper-left, upper-right, lower-left and lower-right part of the image). On each nerve fiber was marked eight specific points (landmarks) using the tpsDig program (four on the outer edge of myelin sheat, four on the axon, according to the principle of 12-6, 9-3 clockwise) in the same order, which is a condition for the validity of the data used. The entire sample was combined using the tpsUtil program. The obtained data in the form of a tps. File was imported into the MorphoJ program for further analysis of the shape of the two-dimensional models. The MorphoJ program contains geometric morphometry tests that perform scaling, centering, and rotation of the examined structures to make it possible to compare them in shape. After generating the covariance matrix and entering the classifier, drug or physiological solution as well as the method of application (intraneural or perineural), a discriminant functional analysis was performed. The results were exported from the MorphoJ program in the form of graphs, tables, and text.

6

Z. Ajanovi´c et al.

3 Results Analysis of the shape variation between groups of the examined sample was performed using the geometric morphometric method and statistical tests contained in the MorphoJ program. First, we analyzed shape differences between groups of nerve fibers with different applications of the liposomal bupivacaine and physiological solution. In total, we analyzed differences of the nerve fibers shape of four groups: LBP-liposomal bupivacaine, perineural application; LBI-liposomal bupivacaine, intraneural application; PSP-physiological solution, perineural application, and PSI-physiological solution, intraneural application. For analysis was used eight landmarks, four were marked on the border of the nerve fiber, and four were marked on the border of its axon. In Table 1 was presents eigenvalues and percent of the variances of the shape variation of the nerve fibers of the examined simple. First six principal components described total variability of the shape variation of the nerve fibers (Table 1). Table 1. Eigenvalues and percent of the variances of the nerve fibers shape explained by the PCs Eigenvalues

% Variance

Cumulative %

PC 1

0,00381504

33,517

33,517

PC 2

0,00326555

28,690

62,207

PC 3

0,00154203

13,548

75,754

PC 4

0,00135031

11,863

87,618

PC 5

0,00087278

7,668

95,286

PC 6

0,00053662

4,714

100,000

Changes in the position of the landmarks used for the analysis of the shape variation of the nerve fibers were showed in Fig. 1. Figure 2 showed the position of the nerve fibers in morphological space defined by the PC1 and PC1. Discriminant functional analysis was used for the analysis shape differences between groups of nerve fibers in examined simple. Statistically, significant difference was observed only between two groups: group of the nerve fibers with perineural application of the liposomal bupivacaine and group of the nerve fibers with intraneural application of the physiological solution (LBP–PSI). Procrustes distance was 0,02558834 and P-value was 0,0144 (Fig. 3). For analysis of the shape variation of the axons between four examined groups was used four landmarks marked on the border of the axons. In MorphoJ program was performed tests of geometric morphometry. Two first principal component described total variability of the axons shape in examined simple (Table 2).

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine

7

Fig. 1. Changes in the position of the landmarks used for the analysis of the shape variation of the nerve fibers. Blue points represent the middle position and blue lines represent the direction of its changes in examined sample.

Fig. 2. Shape variation of all nerve fibers of examined sample defined by the PC1 and PC2.

Changes in the position of the landmarks used for analysis of the shape variation of the axons was showed on Fig. 4. Figure 5 showed position of the axons in morphological space defined by the PC1 and PC1. Discriminant functional analysis was used for the analysis of shape differences between groups of the examined sample. Statistically significant difference was observed only between two groups: the group of the axons with intraneural application of the

8

Z. Ajanovi´c et al.

Fig. 3. Discriminant analysis of shape variation between nerve fibers with peripheral application of liposomal bupivacaine (LBP) and nerve fibers intraneural application of physiological solution.

Table 2. Eigenvalues and percent of the variances of the axons shape explained by the PCs PCs

Eigenvalues

% Variance

Cumulative %

PC1

0,01069181

71,110

71,110

PC2

0,00434386

28,890

100,000

Fig. 4. Changes in the position of the landmarks used for analysis of the shape variation of the axons in examined sample. Blue points represent middle position and blue lines represent direction of its changes in examined simple.

liposomal bupivacaine and group of the axons with intraneural application of the physiological solution (LBI–PSI). The Procrustes distance was 0,05499485 and P-value was 0,0061 (Fig. 6).

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine

9

Fig. 5. Shape variation of all axons of examined sample defined by the PC1 and PC2.

Fig. 6. Discriminant analysis of shape variation between axons with the intraneural application of liposomal bupivacaine (LBI) and axons with the intraneural application of the physiological solution.

4 Discussion The neurotoxicity of local anesthetics is manifested when they change the normal function and appearance of nerve cells [11]. Neurotoxicity caused by local anesthetics has been documented through previous experimental studies [12]. However, in most of these

10

Z. Ajanovi´c et al.

studies, neurotoxicity assessment methods were based on neurological symptoms and assessment of changes in sensibility or motor function. Despite the use of standardized tests to examine the neurological functions of experimental animals, in practice, it is quite difficult to be objective about the evaluation of neurological disorders, especially in small experimental animals such as rats. Even the use of classic qualitative histological methods for describing changes can be somewhat difficult, due to the subjectivity of the pathologist who examines the specimens, even in a case when he/ she is blind to the research groups. Therefore, in our research, we offered a new concept in the assessment of neurotoxicity, of a new, long-acting local anesthetic, liposomal bupivacaine. Using the method of geometric morphometry, we determined whether there were changes in the shape of nerve fibers, excluding the assessment of fiber size, after different applications of liposomal bupivacaine. The results showed that there was no change in the shape of nerve fibers after perineural application of liposomal bupivacaine compared to the control group, where physiological solutions were applied perineurally. However, statistically significant differences in nerve fiber shape were found between perineurally applied liposomal bupivacaine and intraneurally applied physiological solution. When any tested solution is applied intraneurally, the absorption of it into the surrounding tissue is slower because the perineurium prevents rapid dilution of the tested solution. Because of this, the nerve fibers under the perineurium are longer exposed to the higher pressure of the solution, which can result in a change in its structure and shape due to the mechanical injury of nerve fibers. Previous studies showed no significant histological evidence of neurotoxicity following sciatic nerve block with liposomal bupivacaine, compared with physiological solution [13]. Unlike the classic qualitative histological analysis, in our study, using geometric morphometry on digital images of histological preparations, we established precisely and objectively that there was a difference in the shape of nerve fibers. Our study’s geometric morphometric analysis of digital images was more reliable than assessment by visual inspection, which uses poorly defined criteria that cannot be reliably reproduced. The myelin sheath surrounds the axon and electrically isolates it and enables the rapid conduction of the nerve impulse [14]. The thickness of the myelin sheath can be influenced by numerous factors, both physiological and pathological. The thickness itself can dictate the appearance and shape of the nerve fiber on histological preparations [15]. Due to this, in our study, we used four Landmark points on the axons, to determine whether there are changes in the axons, excluding the myelin sheath, which can dictate the shape of the nerve fiber on the histological preparation. The geometric morphometric method showed that there are no statistically significant differences in s shape of axons after perineural and intraneural application of liposomal bupivacaine, although statistically significant changes were found after intraneural application of liposomal bupivacaine and intraneural application of the physiological solution. It is widely known that any local anesthetic under perineurium, due to high mechanical pressure, can cause shape changes. Previous studies on pigs showed no detectable changes on electron microscopy after perineural application of liposomal bupivacaine [13].

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine

11

In contrast to our results, a lack of histological changes was also recorded after subarachnoid administration of liposomal bupivacaine in a pig model [16]. While Joshi et al. found certain histological modifications after subarachnoid administration in dog models, albeit with higher doses of liposomal bupivacaine than is usual [17]. Since previous studies had different methodological approaches in assessing histological changes, it is difficult to directly compared comparison of the results. The use of geometric morphometry in our study for assessing changes in the shape of axons undoubtedly brought an element of accuracy that helped us make accurate diagnostic decisions. It is possible that liposomal bupivacaine when applied intraneural manifested a selective toxic effect on axons, which is also demonstrated in previous studies with the usage of traditional local anesthetics such as lidocaine [18]. These studies reported signs of active denervation by using various tests on rabbits. Others found no signs of histological damage on the spinal cord or surrounding tissues after the usage of liposomal bupivacaine [19]. Traditional histomorphometric analysis of the nerve fiber is a quantitative approach based on the assessment of the thickness and density of the nerve fiber and its substructure while ignoring the shape as a separate parameter [20]. In traditional histomorphometry, it is not possible to obtain a representation of the shape of the nerve fiber using the usual data matrices and distance measurements [21]. With traditional histomorphometry, it is usually possible to obtain data on the length and width of the structure and the distance between individual landmarks [22, 23]. In these terms, geometric morphometric methods are more valid in studying shape variation than traditional morphometric methods. In our study, by using geometric morphometry, we excluded the subjectivity factor that may be present during classical histological analysis. Also, it is possible to observe minimal changes on the nerve fiber and its substructures more precisely, because each point marked on the nerve fiber in the coordinate system has x, y-axis values, which enabled us to compare them with each other as well as compare their position on different nerve fibers. In this way, we obtained precise data on the shapes of nerve fibers and their substructures. With geometric morphometry tests that are contained in specially created programs, we analyzed whether there is a statistically significant difference in the morphological characteristics, in this study shape, of nerve fibers and axons. Considering the above, we believe that geometric morphometry can be introduced as a new method in assessing the effect of not only local anesthetics on nerve fibers, but also for testing other drugs on individual tissues. The study of individual structures’ shape and shape changes is the basis of biological sciences. The use of biochemical and nuclear tools requires much higher financial expenses compared to the method of geometric morphometry, which provides an objective insight into the change in the shape of the observed cellular elements. Computeraided image analysis can improve the analysis of histological entities by generating continuous variables that can be used for statistical comparisons. Using geometric morphometry as a new, objective diagnostic method, it was proven that the use of liposomal bupivacaine will not lead to a statistically significant change in the shape of nerve waves or axons when administered under clinically controlled

12

Z. Ajanovi´c et al.

conditions and in approved doses, which makes it safe for performing peripheral nerve blockades. Acknowledgments. The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Federal Ministry for Education and Science, Bosnia and Herzegovina [grant number 05–35-2174–1/22].

References 1. Chang, A., Dua, A., Singh, K., et al.: Peripheral nerve blocks. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing (2022) 2. McCann, M.E.: Liposomal Bupivacaine: effective, cost-effective, or (just) costly? Anesthesiology 134, 139–142 (2021) 3. Carriel, V., Garzón, I., Alaminos, M., Cornelissen, M.: Histological assessment in peripheral nerve tissue engineering. Neural Regen. Res. 9, 1657–1660 (2014) 4. Rastogi, A.: Changing role of histopathology in the diagnosis and management of hepatocellular carcinoma. World J. Gastroenterol. 24, 4000–4013 (2018) 5. Liuti, T., Dixon, P.M.: The use of the geometric morphometric method to illustrate shape difference in the skulls of different-aged horses. Vet. Res. Commun. 44, 137–145 (2020) 6. Klingenberg, C.P.: MorphoJ: an integrated software package for geometric morphometrics. Mol. Ecolog. Res. 11, 353–357 (2011) 7. Ivanovi´c, A., Aljanˇciˇc, G., Arntzen, J.W.: Skull shape differentiation of black and white olms (Proteus anguinus and Proteus a. parkelj): an exploratory analysis with micro-CT scanning. 82, 107–114 (2013) 8. National Institutes of Health: Principles of laboratory animal care. National Institutes of Health publication No. 86–23. Bethesda, MD: National Institutes of Health (1985) 9. Hasanbegovic, I., Kulenovic, A., Hasanovic, S.: Effects of intraneural and perineural injection and concentration of Ropivacaine on nerve injury during peripheral nerve block in Wistar rats. J. Health Sci. 3, 243–249 (2013) 10. Hasanbegovi´c, I., Derviševi´c, L., Valjevac, A., Hasanovi´c, S.: Potential neurotoxic effect of dexamethasone used as adjuvant to local anesthetics during peripheral nerve blockade. Folia Med. Fac. Med. Univ. Saraeviensis 52, 16–22 (2017) 11. Rodionova, N.N., Allakhverdiev, E.S., Maksimov, G.V.: Study of myelin structure changes during the nerve fibers demyelination. PLoS ONE 12, e0185170 (2017) 12. Verlinde, M., Hollmann, M.W., Stevens, M.F., Hermanns, H., Werdehausen, R., Lirk, P.: Local anesthetic-induced neurotoxicity. Int. J. Mol. Sci. 17, 339 (2016) 13. Damjanovska, M., et al.: Neurotoxicity of perineural vs intraneural-extrafascicular injection of liposomal bupivacaine in the porcine model of sciatic nerve block. Anaesthesia 70, 1418–1426 (2015) 14. Snaidero, N., Schifferer, M., Mezydlo, A., Zalc, B., Kerschensteiner, M., Misgeld, T.: Myelin replacement triggered by single-cell demyelination in mouse cortex. Nat. Commun. 11, 4901 (2020) 15. Graf von Keyserlingk, D., Schramm, U.: Diameter of axons and thickness of myelin sheaths of the pyramidal tract fibers in the adult human medullary pyramid. Anat Anz 157, 97–111 (1984) 16. Zel, J., et al.: Neurological and histological outcomes after subarachnoid injection of a liposomal bupivacaine suspension in pigs: a pilot study. Br. J. Anaesth. 122, 379–387 (2019)

Geometric Morphometrics Analysis of the Effect of Liposomal Bupivacaine

13

17. Joshi, G.P., Patou, G., Kharitonov, V.: The safety of liposome bupivacaine following various routes of administration in animals. J. Pain Res. 8, 781e9 (2015) 18. Bavli, Y., Rabie, M., Fellig, Y., Nevo, Y., Barenholz, Y.: Liposomal bupivacaine (Bupigel) demonstrates minimal local nerve toxicity in a rabbit functional model. Pharmaceutics 13, 185 (2021) 19. Girish, P.J., Patou, G., Kharitonov, V.: The safety of liposome bupivacaine following various routes of administration in animals. J. Pain Res. 8, 781–789 (2015) 20. Hasanbegovi´c, I., Musi´c, M., Derviševi´c, L., Saraˇc Hadžihalilovi´c, A., Redži´c, A., Džananovi´c, D.ž.: Histomorphometric differences between perineural and intraneural application of 0.75% ropivacaine in Wistar rats. Acta Medica Saliniana 50, 31–39 (2020) 21. Utkualp, N., Ercan, I.: Anthropometric measurements usage in medical sciences. Hindawi Publ Corp BioMed Res Int 2015, 7 (2015) 22. Ocakoglu, G., Ercan, I.: Traditional and modern morphometrics. Turkiye Klinikleri J. Biostat. 5, 37–41 (2013) 23. Shaffer, D.W.: Learning mathematics through design: the anatomy of Escher’s world. J. Math. Behav. 16, 95–112 (1997)

Skin Substitutes: An Overview of Current State of the Art Nina Kocivnik(B) Faculty of Pharmacy, University of Ljubljana, 1000 Ljubljana, Slovenia [email protected]

Abstract. The skin is the largest organ in the body and performs many physiological functions. Of these, the barrier between the organism and the environment is the most important. Any perturbation of the skin can lead to several pathological events, so the main goal of medicine is to repair the damage. In clinical practice, this is most commonly performed either with conventional coverages, which include petroleum gauze and silicone sheets, or biological coverages, which include allografts, xenografts, and human amnion. However, their use is limited in the case of extensive and deep wounds. Skin substitutes from tissue-engineered biomaterials are being developed to overcome these shortcomings. Research studies aim to improve their applicability mainly in terms of meeting relevant clinical requirements (tailored to the wound size and type of skin damage), accessibility, and preparation technique. An ideal biomaterial skin substitute has to be safe for the patient, has proven clinical efficacy, and is convenient for handling and use. At present, none of the available products fulfill all these requirements. In this review, we aim to present the currently available commercial skin substitutes and indications for their use in clinical practice as well as describe their advantages and disadvantages. Also, the latest progress in the development of skin models in the area of achieving vascularization, the incorporation of skin adnexa, and the creation of full-thickness skin are overviewed. Keywords: Tissue engineering · Skin substitutes · Artificial skin products · Wounds

1 Introduction The skin is the largest and structurally most complex organ of the body, accounting for approximately 15% of the total body weight of an adult. It consists of three functional layers: the epidermis, dermis, and hypodermis. The epidermis is the top layer of the skin and is composed of keratinocytes that form a barrier to the external environment and prevent water loss. Under the epidermis is the dermis, which consists mainly of fibroblasts. This layer is responsible for elasticity and resistance to deformation. The hypodermis, the subcutaneous tissue, is the deepest layer and is composed mainly of adipocytes. This layer is responsible for fat and energy storage as well as thermal and mechanical protection of the body. The skin performs numerous functions, such as protection from pathogens, thermoregulation, excretion of waste products, prevention of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 14–21, 2024. https://doi.org/10.1007/978-3-031-49068-2_2

Skin Substitutes: An Overview of Current State of the Art

15

water loss, etc. [1]. Skin wounds can occur for a variety of reasons, including trauma, infection, genetic disorders, and burns, which are the most common causes of skin loss. Burns can cause extensive and deep wounds with no possibility of skin regeneration [2, 3]. Because of the risk of infection, the wound should be covered as soon as possible. Conventional dressings include petroleum gauze and biological dressings used for temporary coverage. They have the disadvantage of needing to be changed regularly, which can have a negative effect on re-epithelialization and wound healing, and they cannot be used to cover extensive deep burns. Biologic dressings are also available, including allografts, xenografts, and human amnion. The most appropriate temporary dressing is a cadaveric allograft, which can cover a wound temporarily and for an extended period of time. When allografts are not available or affordable, xenografts can be used. The only commercially available xenografts are porcine xenografts. Amnion is a thin tissue derived from the inner layer of the amniotic membrane and is one of the most effective substitutes for partial wound coverage. For immunologic reasons, biological dressings are also used only as temporary wound coverings [4, 5].

2 Methods This paper provides an overview of currently available commercial skin substitutes and indications for their use in clinical practice. Their advantages and disadvantages are described, as well as the latest progress in the skin model development. Literature search was conducted for this review. The data about skin substitutes, tissue engineering and artificial skin products for wound healing were collected from various sources. These included electronic databases PubMed, Medline and Science Direct. The search was performed using a combination of the following terms: wounds, tissue engineering, skin substitutes and artificial skin products. We included the articles that were not older than 15 years, encompassed wound healing mechanisms and surgical treatment of skin defects with application of skin substitutes to superficial wounds. The articles were selected in correlation with the study objective and their scientific relevance.

3 Skin Substitutes The main problem of conventional dressings is that they are unable to repair extensive deep wounds. Therefore, tissue engineering is becoming increasingly important, with the aim of creating skin substitutes from biomaterials that cover full-thickness wounds and fully mimic the functions of the original tissue [6]. Currently, various biomaterials are used in clinical medicine that differ in terms of duration of coverage (temporary, semi-permanent, permanent), anatomical structure (epidermal, dermal, dermo-epidermal), type of biomaterial (biological, synthetic), composition in terms of cellular component (cellular, acellular), and method of manufacture (in vivo, in vitro) [3]. All skin substitutes must meet certain requirements. The three most important are patient safety, clinical efficacy, and ease of handling and use. In addition, biomaterials must not be toxic or immunogenic, cause excessive inflammation, or pose a risk for communicable diseases. They must be biodegradable, repairable, and capable of reconstructing a tissue with similar properties to the skin they replace [3, 7]. They must also

16

N. Kocivnik

relieve pain, prevent fluid and heat loss, and protect the wound from infection. It is also advantageous if the substitute has a long shelf life [2, 8]. Of the substitutes currently commercially available, none meet all of these requirements, so the ideal skin substitute does not yet exist. Therefore, several studies are currently underway to develop improved skin substitutes [9]. In this article, we review the most commonly used commercial skin substitutes that are currently available for clinical and therapeutic use. We also describe recent advances and innovations in the production of skin substitutes, such as vascularization, incorporation of skin adnexa, and production of triple-layered substitutes.

4 Commercially Available Skin Substitutes 4.1 Epidermal Skin Substitutes Epidermal skin substitutes are keratinocytes taken from a donor and propagated in vitro. The main steps in their production are the correct isolation of keratinocytes from the donor and their subsequent cultivation. Epidermal substitutes quickly cover wounds, but due to the lack of dermal layer, they are not elastic and do not provide mechanical stability to the regenerated skin. The dermal layer is also required for perfusion, so these substitutes cannot be used for permanent wound coverage [5]. There are two approaches to applying keratinocytes to a wound. In the cultured epithelial autograft (CEA) method, individual keratinocyte colonies grow into stratified epithelial layers that are then transferred to the wound. It is a standardized method for covering extensive burn wounds [3, 10]. Epicel is the best known commercially available CEA on the market. It is derived from the patient’s own keratinocytes, which form CEA sheets 15 days after skin biopsy. Epidex is composed of keratinocytes derived from the outer root sheath of hair follicles. It is used to treat chronic ulcers. Myskin is another type of CEA that comes as a spray with a shelf life of 2–3 weeks. It is most commonly used to treat diabetic ulcers and superficial burns. The main advantage of CEAs is that a small skin biopsy can provide enough cells to cover the entire body surface in a few weeks. Another approach to keratinocyte application is spraying directly onto the wound. Cell spray is an aerosolized form of CEAs in which subconfluent keratinocytes are sprayed onto the wound. This shortens cell growth and reduces scarring. The problem is that this approach can only be applied to wounds of small thickness [2, 3, 10–12]. 4.2 Dermal Skin Substitutes Dermal skin substitutes are a group of different biomaterials for wound coverage and closure that promote tissue regeneration and optimize wound healing conditions. Most dermal substitutes are acellular and therefore cause minimal inflammatory or immunological response. They consist of a bilayer of collagen and glycosaminoglycan overlaid by a temporary epidermal substitute [13]. The composition of the various substitutes varies slightly, but their mode of application is similar. Dermal substitutes are used to cover deep wounds or burns where both the epidermal and dermal layers need to be replaced. The dermal layer is crucial because its successful vascularization allows further successful integration of the epidermal layer and epithelialization of the wound by

Skin Substitutes: An Overview of Current State of the Art

17

keratinocytes. In addition, treatment of deeper wounds with epidermal substitutes alone results in poorer healing because dermal substitutes prevent wound contraction and provide adequate mechanical stability to the wound. The major limitation with dermal skin substitutes is that they require gradual revascularization after in vivo implantation [3, 13]. Currently, there are several commercially available skin substitutes that differ in terms of their origin, intended use, advantages and disadvantages, and structural properties. Integra is one of the most commonly used synthetic skin substitutes and is most commonly used to cover full-thickness burn wounds. It is made of bovine collagen covered by a silicone layer and contains no epidermal component. It can be used for burns, skin wounds and chronic ulcers that involve the full thickness of the skin. Compared to autografts, allografts and xenografts, Integra is superior in terms of wound healing time, but has not shown good results in terms of wound infection and graft transfer. Alloderm consists of a human acellular dermal matrix with a preserved basement membrane used to cover acute burn wounds. It has been used in single-stage surgical procedures in combination with very thin skin grafts, but longer follow-up and more clinical trials are needed to determine the success of its use. Dermagraft is a neonatal fibroblast substitute whose cellular components can produce human growth factors and cytokines that can stimulate angiogenesis, tissue proliferation, and re-epithelialization during healing. Dermagraft has not raised any safety concerns to date. It has been used in burn wounds, chronic wounds, and diabetic ulcers [2, 3, 5, 10, 12, 14, 15]. Biobrane consists of a nylon mesh with porcine collagen type 1 covered with a thin silicone layer. The nylon mesh acts as a dermal layer and the silicone membrane acts as an epidermal layer. They have been successfully used in partial-thickness burns in children, and their use in patients with toxic epidermal necrosis and chronic wounds has been reported. TransCyte is a collagen-coated nylon membrane that also has neonatal allogeneic fibroblasts embedded in its scaffold. Biobrane and TransCyte provide production of growth factors and cytokines necessary for wound healing, relieve pain, and shorten wound healing time. Both are used for temporary coverage of partial-thickness wounds [2, 3, 14]. Matriderm is a 1-mm-thick collagen coated with an elastin hydrolysate that consists of a ligamentous dermal layer. The collagen serves as a support structure for cell and blood vessel growth. Applications include full-thickness burns and deep dermal and chronic wounds [2, 3, 5, 10]. OASIS Wound Matrix is a skin substitute consisting of an acellular extracellular matrix derived from the submucosa of the porcine jejunum. It is used to cover partial or full-thickness wounds, various ulcers, drains, and surgical wounds [2, 3]. Hyalomatrix and Hyalograft 3D are substitute materials prepared from hyaluronic acid derivatives. Hyalomatrix is used to cover burns and chronic wounds. Hyalograft 3D contributes to better wound healing when used to treat diabetic ulcers [3, 5]. 4.3 Dermo-epidermal Skin Substitutes Dermo-epidermal skin substitutes contain both dermal and epidermal layers and are therefore functionally very similar to normal human skin, but are currently used only as temporary wound coverings. They contain keratinocytes and fibroblasts, but cannot yet perform all the functions of skin because they lack revitalization cells, immune cells, sweat glands and hair follicles. These skin substitutes provide various growth factors, cytokines and extracellular matrix for the host cells. They also trigger and regulate wound

18

N. Kocivnik

healing and effectively relieve pain [3, 10, 16]. Apligraft is a two-layer skin substitute material. The lower skin layer contains allogeneic neonatal fibroblasts growing on a matrix of bovine type I collagen, while the upper epidermal layer contains allogeneic neonatal keratinocytes. The substitute material is used for the treatment of partial and full thickness burns, chronic wounds and diabetic ulcers. It has also been shown to promote wound closure. Its disadvantage is its short shelf life, which is only 5 days at room temperature. OrCel is a bilayer cell matrix of allogeneic fibroblasts and keratinocytes grown in a bovine collagen type I sponge in two separate layers. The substitute produces cytokines and growth factors that have a positive effect on wound healing [2, 3, 10]. PolyActive consists of endogenous keratinocytes and fibroblasts in a matrix of polyethylene terephthalate and polybutylene terephthalate, which prevent the substitute from shrinking. The substitute is commonly used for bone reconstruction, but its use for skin regeneration has not been well studied. The Tissue Tech Autograft System combines two biomaterials, Hyalograft 3D dermal substitute and Laser skin epidermal substitute. It consists of autologous fibroblasts and keratinocytes grown on hyaluronic acid membranes. It has been successfully used for the treatment of diabetic ulcers [3].

5 Advances and Innovations 5.1 Vascularized Skin Substitutes Vascularization is a prerequisite for the successful clinical use of a skin substitute. While some of the skin substitute materials currently in use allow for angiogenesis, it is usually inadequate. If the substitute material is not well vascularized, the wound will not be properly supplied with oxygen and nutrients, which may lead to infection or necrosis [5, 15]. There are two approaches to vascularization of skin substitutes: the angiogenesis approach and the prevascularization approach. In the angiogenesis approach, blood vessel growth is stimulated, but this approach is of limited use when rapid vascularization of large grafts is required. In the prevascularization approach, microvascular networks are created on the graft before it is implanted in the patient. This implies rapid blood supply through the preformation of blood vessels [17]. Klar et al. reported a prevascularized graft that contained a dermal and an epidermal layer. They used endothelial cells derived from the vascular fraction of adipose tissue containing the capillary plexus. 3D hydrogel systems containing fibrin or type I collagen were used to develop the vascular networks. After transplantation, the preformed vascular networks were anastomosed to the recipient’s vasculature within 4 days [18]. Using 3D bioprinting, Baltazar et al. developed a multilayer vascularized skin substitute. For fabrication, they used two bioinks that contained different cells required for dermis and epidermis formation. In vitro, keratinocytes formed a multilayer barrier, while endothelial cells and placental pericytes self-assembled into interconnected microvascular networks. These grafts were able to inoculate with mouse microvessels and perfuse within four weeks of implantation [19]. 5.2 Inclusion of Skin Adnexa Currently used skin substitutes contain cells of the dermis and epidermis but not skin adnexa, so they cannot restore the structure and function of the whole skin. Therefore,

Skin Substitutes: An Overview of Current State of the Art

19

a major challenge of tissue engineering is the production of substitute materials with all skin components such as hair follicles, sweat glands and sebaceous glands. The combination of bioengineered matrices and cell aggregates such as organoids represents great potential for creating skin substitutes with appendages. However, much scientific research is still needed in this area. Regeneration of sweat glands is a major problem in deep burns. Sheng et al. solved this problem by inducing bone marrow-derived mesenchymal stem cells to develop a sweat gland cell phenotype in vitro. The prepared cells were then transferred to the wound and observed. The results showed that conversion of stem cells to functional sweat gland cells occurred in all observed areas, restoring sweating [20]. Sebaceous gland regeneration is also a major problem after deep skin wounds. In their study, Ying et al. investigated whether epidermal growth factor promoted wound healing and hair follicle regeneration by using mesenchymal stem cells. The results showed that wounds were minimized when a combination of mesenchymal stem cells and epidermal growth factor was used, and sebaceous gland regeneration also occurred [21]. Abaci et al. have produced a skin substitute that allows increased differentiation of hair follicles. Using 3D bioprinting, they physiologically arranged the cells in the hair follicles, which then contributed to the production of skin substitutes with different hair follicle densities and the formation of the dermis from phenotypically similar dermal papilla cells in vivo [22]. 5.3 Full Thickness Skin Substitutes The skin substitutes currently used clinically are either single layer (epidermis or dermis only) or double layer (epidermis and dermis), so that the regenerated skin cannot perform all the functions of normal skin. Recently, increasing emphasis has been placed on creating three-layer skin substitutes that include the epidermis, dermis, and hypodermis [23]. Zimoch et al. created a three-layer skin substitute in which the hypodermis was generated by differentiation of adipose-derived mesenchymal stem cells in a type I collagen hydrogel and fused with a prevascularized dermis composed of human dermal microvascular endothelial cells and fibroblasts. Keratinocytes were then added to form an epidermal layer [24]. In their study, Haldar et al. reported the fabrication of a three-layered scaffold of biomaterials capable of providing an environment for structural and functional regeneration of all three skin layers. It was fabricated by optimizing different techniques such as casting, electrospinning, and lyophilization. The results showed improved and accelerated wound healing in vivo [25]. Monfort et al. incorporated mesenchymal stem cells from bone marrow and adipose tissue into a human plasma hydrogel to create a three-layer skin substitute. The cells were able to differentiate into mature adipocytes, and their presence in the subcutaneous layer also contributed to epidermal differentiation. In this way, a fully differentiated trilayered human skin could be produced that could be used for in vitro drug absorption testing and in the field of regenerative therapies [26].

20

N. Kocivnik

6 Conclusion Skin substitutes play an important role in the treatment of deep dermal and full-thickness wounds. The current commercially available substitutes have advantages and disadvantages that should be considered when selecting a treatment. What they all have in common is that they have revolutionized the treatment of skin disease or injury. However, there are still unresolved issues such as wound shrinkage, impaired vascularization, scarring, and the high cost of these substitutes. To date, studies have been conducted to address certain shortcomings of current skin substitutes, such as poor vascularization, lack of sweat glands, sebaceous glands, and hair follicles, and lack of all three skin layers. However, there is not yet an ideal skin substitute; further research is needed to address these issues.

References 1. McLafferty, E., Hendry, C., Farley, A.: The integumentary system: anatomy, physiology and function of skin. Nurs. Stand. 27(3), 35–42 (2012) 2. Dai, C., Shih, S., Khachemoune, A.: Skin substitutes for acute and chronic wound healing: an updated review. J. Dermatol. Treat. 31(6), 639–648 (2020) 3. Shevchenko, R.V., James, S.L., James, S.E.: A review of tissue-engineered skin bioconstructs available for skin reconstruction. J. R. Soc. Interface 7(43), 229–258 (2010) 4. Roshangar, L., Soleimani Rad, J., Kheirjou, R., Reza Ranjkesh, M., Ferdowsi Khosroshahi, A.: Skin burns: review of molecular mechanisms and therapeutic approaches. Wounds 31(12), 308–315 (2019) 5. Haddad, A.G., Giatsidis, G., Orgill, D.P., Halvorson, E.G.: Skin substitutes and bioscaffolds: temporary and permanent coverage. Clin. Plast. Surg. 44(3), 627–634 (2017) 6. Khoshnood, N., Zamanian, A.: Decellularized extracellular matrix bioinks and their application in skin tissue engineering. Bioprinting 20(2) (2020) 7. Cantòn, I., et al.: Development of a 3D human in vitro skin co-culture model for detecting irritants in real-time. Biotechnol. Bioprocess Eng. 106(5), 794–803 (2010) 8. Groeber, F., Holeiter, M., Hampel, M., Hinderer, S., Schenke-Layland, K.: Skin tissue engineering—in vivo and in vitro applications. Adv. Drug Deliv. Rev. 63(4–5), 353–366 (2011) 9. MacNeil, S.: Biomaterials for tissue engineering of skin. Mater. Today 11(5), 26–35 (2008) 10. Savoji, H., Godau, B., Hassani, M.S., Akbari, M.: Skin tissue substitutes and biomaterial risk assessment and testing. Front. Bioeng. Biotechnol. 6, 86 (2018) 11. Chocarro-Wrona, C., López-Ruiz, E., Perán, M., Gálvez-Martín, P., Marchal, J.A.: Therapeutic strategies for skin regeneration based on biomedical substitutes. J. Eur. Acad. Dermatol. Venereol. 33(3), 484–496 (2019) 12. Nyame, T.T., Chiang, H.A., Leavitt, T., Ozambela, M., Orgill, D.P.: Tissue-engineered skin substitutes. Plast. Reconstr. Surg. 136(6), 1379–1388 (2015) 13. Rehim, S.A., Singhal, M., Chung, K.C.: Dermal skin substitutes for upper limb reconstruction: current status, indications, and contraindications. Hand Clin. 30(2), 239–252 (2014) 14. Bar-Meir, E., Mendes, D., Winkler, E.: Skin substitutes. Israel Med. Assoc. J. 8, 188–191 (2006) 15. Sheikholeslam, M., Wright, M.E.E., Jeschke, M.G., Amini-Nik, S.: Biomaterials for skin substitutes. Adv. Healthcare Mater. 7(5), 10 (2018) 16. Grayson, W.L., Martens, T.P., Eng, G.M., Radisic, M., Vunjak-Novakovic, G.: Biomimetic approach to tissue engineering. Semin. Cell Dev. Biol. 20(6), 665–673 (2009)

Skin Substitutes: An Overview of Current State of the Art

21

17. Oualla-Bachiri, W., Fernández-González, A., Quiñones-Vico, M.I., Arias-Santiago, S.: From grafts to human bioengineered vascularized skin substitutes. Int. J. Mol. Sci. 21(21), 81–97 (2020) 18. Klar, A.S., et al.: Tissue-engineered dermo-epidermal skin grafts prevascularized with adipose-derived cells. Biomaterials 35(19), 5065–5078 (2014) 19. Baltazar, T., et al.: Three dimensional bioprinting of a vascularized and perfusable skin graft using human keratinocytes, fibroblasts, pericytes, and endothelial cells. Tissue Eng. 26, 227– 238 (2020) 20. Sheng, Z., et al.: Regeneration of functional sweat gland-like structures by transplanted differentiated bone marrow mesenchymal stem cells. Wound Rep. Regen. 17(3), 427–435 (2009) 21. Xia, Y., You, X.E., Chen, H., Yan, Y.J., He, Y.C., Ding, S.Z.: Epidermal growth factor promotes mesenchymal stem cell-mediated wound healing and hair follicle regeneration. Int. J. Clin. Exp. Pathol. 10(7), 7390–7400 (2017) 22. Abaci, H.E., et al.: Tissue engineering of human hair follicles using a biomimetic developmental approach. Nat. Commun. 9(1), 5301 (2018) 23. Bellas, E., Seiberg, M., Garlick, J., Kaplan, D.L.: In vitro 3D full-thickness skin-equivalent tissue model using silk and collagen biomaterials. Macromol. Biosci. 12(12), 1627–1636 (2012) 24. Zimoch, J., et al.: Bio-engineering a prevascularized human tri-layered skin substitute containing a hypodermis. Acta Biomater. 134, 215–227 (2021) 25. Haldar, S., Sharma, A., Gupta, S., Chauhan, S., Roy, P., Lahiri, D.: Bioengineered smart trilayer skin tissue substitute for efficient deep wound healing. Mater. Sci. Eng. C-Mater. Biol. Appl. 105, 110–140 (2019) 26. Monfort, A., Soriano-Navarro, M., García-Verdugo, J.M., Izeta, A.: Production of human tissue-engineered skin trilayer on a plasma-based hypodermis. J. Tissue Eng. Regen. Med. 7(6), 479–490 (2013)

Evaluation of Platelet-Enriched Plasma Antimicrobial Effect: In Vitro Study Tea Be´cirevi´c1(B) , Izet Eminovi´c1 , Nadira Ibrišimovi´c Mehmedinovi´c2 , Edin Omeragi´c3 , Edin Falan3 , Ermin Papraˇcanin3 , and Mirza Ibrišimovi´c3,4 1 Faculty of Natural Science and Mathematics, University Sarajevo, Sarajevo, Bosnia and

Herzegovina [email protected] 2 Department of Chemistry, Faculty of Science, University of Tuzla, Tuzla, Bosnia and Herzegovina 3 Familia Institution for Accommodation, Care and Rehabilitation, Sarajevo, Bosnia and Herzegovina 4 Medical School Faculty-Sarajevo School of Science and Technology, Sarajevo, Bosnia and Herzegovina

Abstract. Platelet-enriched plasma is autologous blood product, that contains platelets concentration two to three times above normal blood level. As such, PRP is a rich source of different bioactive molecules including grow factors, enzymes, cytokines, and chemokines. Beside its regenerative properties, a limited number of studies has proven that PRP can induce antimicrobial effect against single growing pathogens and biofilms. Aim of this study was to analyze PRP antimicrobial effect against three most common biofilm forming bacteria, including S. aureus, E. coli and P. aeruginosa. The antimicrobial property of PRP was evaluated after 24 h of incubation with selected bacteria in BHI media using spectrophotometer with a light source of 600 nm. To check whether PRP can inhibit bacterial biofilm formation, after 24 h incubation, tube screening test (TM) was applied. Bacteria treated with PRP and platelet poor plasma (PPP) were compared with untreated control, composed of bacteria growing spontaneously in BHI media. PRP produced strong growth inhibition in all tested bacteria when compared to bacteria treated with PPP and control group. Based on the obtained results it can be concluded that PRP can induce antimicrobial effect on S. aureus, P. aeruginosa and E. coli. PRP also reduced biofilm formation for P. aeruginosa and E. coli. However, there was no effect on S. aureus biofilm formation. Keywords: Platelet-enriched plasma (PRP) · Antimicrobial activity · Biofilm

1 Introduction Platelet-enriched plasma (PRP) is a plasma fraction with a concentration of platelets three to five times higher than the basic concentration of autologous human plasma. The normal number of platelets in the blood ranges from 150,000 to 350,000 platelets per 1 µl of blood, while PRP is most often defined as a suspension of 1,000,000 platelets per 1 µl of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 22–28, 2024. https://doi.org/10.1007/978-3-031-49068-2_3

Evaluation of Platelet-Enriched Plasma Antimicrobial Effect

23

blood [1]. Since platelets are the main components of PRP, PRP represents a rich source of various bioactive molecules, including enzymes, growth factors, cytokines, chemokines, and molecules involved in cell adhesion. These bioactive molecules play crucial roles in tissue regeneration and healing process. For a long period of time platelets were considered to only have regenerative properties, however recent findings have revealed antimicrobial properties of platelets. Platelets can induce antimicrobial activity using different mechanism, for example platelets can directly bind and internalize pathogens, they can stimulate the production of reactive oxygen species such as superoxide and hydrogen peroxide, and most recent findings indicate that platelets can also produce small antimicrobial proteins known as platelet microbicidal proteins/peptides (PMPs) [2]. Another advantageous characteristic of PRP is that represents an autologous product, derived from the patient own blood. As such PRP minimizes any risk of allergic and immune reactions or transmission of infectious diseases. Today, one of the biggest challenges in fighting bacterial infections represent biofilm forming bacterial communities. Biofilm infections are difficult to treat due to their highly structured and organized multispecies communities, in which bacteria can exchanges genes conferring to antibiotic resistance. Furthermore, biofilm extracellular matrix made of extracellular polymers acts as a shield protecting bacterial communities from antibacterial agents [3]. More recently antibacterial properties of PRP have been elucidated. Several studies have found that PRP has bacteriostatic and bactericidal effects on most common bacterial pathogens including methicillin-resistant and methicillin-sensitive S. aureus strains (MRSA and MSSA), E. coli, P. aeruginosa, K. pneumonia and S. epidermis [4–7]. However, there is a very little number of studies, studying PRP antimicrobial activity against biofilm infections. Lately, PRP activity to com- bat S. aureus formed biofilm aggregates in equine synovial fluid causing infectious osteoarthritis was studied [8]. According to the results PRP preparations containing higher number of platelets without leukocytes had increased antimicrobial activity and showed positive synergism with antibiotic amikacin. This study also showed that anti- bacterial effect of PRP is contributed mainly to platelets and not to leukocytes. Aim of this study was to investigate PRP antimicrobial effect against three most common biofilm forming bacteria, including S. aureus, E. coli and P. aeruginosa and com- pare its antimicrobial activity with platelet poor plasma (PPP) and control.

2 Materials and Methods 2.1 PRP Preparation Institutional ethical approval was obtained before conducting this study. To prepare PRP, the same male volunteer served as blood donors (age 38). Whole blood was collected in 30ml tubes with Na-citrate as anticoagulant. PRP was prepared using two-step centrifugation. After first centrifugation the blood separates in three layers. First and second layer, known as plasma and buffy coat, are separated in new tube with thin layer of red blood cells at the bottom. After second centrifugation, the upper layer is considered platelet poor plasma (PPP) and the remaining part constitutes platelet rich plasma (PRP).

24

T. Be´cirevi´c et al.

2.2 Bacterial Strains Clinical strains used in this study were Escherichia coli, Staphylococcus aureus and Pseudomonas aeruginosa. Tested strains were collected from patients with persistent urinary and soft tissue infections. Isolated bacteria strains were stored in glycerin stocks at −20 °C. Once they were ready to use, after thawing, 50 µl of each bacteria was inoculated in polystyrene tubes containing 3 ml Brain Heart Infusion (BHI) media and incubated overnight at 37 °C. To test the ability of bacteria to form biofilm formation, tube screening test (TM) was used [9]. 2.3 Evaluation of Antimicrobial Activity This study included two different experimental groups and one negative control group containing bacteria in BHI media without addition of any human blood product. First group tested PRP antimicrobial activity, and second group was used to test activity of PPP. To test PRP antimicrobial activity a total of 100 µl of PRP were added to tubes containing 20 µl of each bacteria and 3 980 µl BHI media. To test the activity of PPP, a total of 100 µl of PPP was added to 20 µl of each bacteria and 3 980 µl BHI media. To assess bacterial proliferation OD600 was measured by a spectrophotometer with a light source of 600 nm before incubation (time 0) and after 24 h of incubation at 37 °C (time 24). Experiments were performed in triplicates. To check whether PRP can inhibit bacterial biofilm formation, after 24 h incubation tube screening test (TM) was applied.

3 Results The effect of PRP and PPP on bacterial growth inhibition were evaluated in three different bacterial strains: P. aeruginosa, S. aureus and E. coli. Bacteria treated with PRP and PPP were compared with untreated control, composed of bacteria growing spontaneously in BHI media. According to the obtained results, PRP produced strong growth inhibition in all tested bacteria when compared to bacteria treated with PPP and control group (Figs. 1, 2 and 3). Bacterial growth was assessed using absorbances (OD600), obtained results are presented in Tables 1, 2 and 3. Table 1. P. aeruginosa OD and tube test results for biofilm formation after 0h and 24h. Bacteria was treated with 100µl of PRP or PPP and compared with control group. Experiment

OD600 (0h)

OD600 (24h)

Bacterial growth

Biofilm 24h

Control

0.04

2.50

2.46

+

PRP + P. aeruginosa

1.21

2.50

1.29

–

PPP + P. aeruginosa

0.40

2.32

1.92

–

Evaluation of Platelet-Enriched Plasma Antimicrobial Effect

25

Opcal density (OD600)

3.00 2.50 2.00 1.50 1.00 0.50 0.00 Control

PRP + P. aeruginosa

PPP + P. aeruginosa

Samples aer 24h at 37°C Fig. 1. Growth rates for P. aeruginosa after 24h PRP or PPP incubation.

Table 2. S. aureus OD and tube test results for biofilm formation after 0h and 24h. Bacteria was treated with 100µl of PRP or PPP and compared with control group. Experiment

OD600 (0h)

OD600 (24h)

Bacterial growth

Biofilm 24h

Control

0.19

2.36

2.17

+

PRP + S. aureus

1.27

2.38

1.11

+

PPP + S. aureus

0.55

2.50

1.95

+

Fig. 2. Growth rates for S. aureus after 24h PRP or PPP incubation.

26

T. Be´cirevi´c et al.

Table 3. E. coli OD and tube test results for biofilm formation after 0h and 24h. Bacteria was treated with 100µl PRP or PPP and compared with control group. Experiment

OD600 (0h)

OD600 (24h)

Bacterial growth

Biofilm 24h

Control

0.23

2.20

1.97

++

PRP + E. coli

1.31

2.50

1.19

+

PPP + E. coli

0.30

2.48

2.18

++

Fig. 3. Growth rates for E. coli after 24h PRP or PPP incubation.

4 Discussion In the last decade PRP regenerative and therapeutical properties have been widely studied and PRP have been used in different medical fields including orthopedics, sports medicine, dentistry, plastic surgery, gynecology, neurosurgery, ophthalmology, urology, wound care and aesthetic medicine [10]. However, antimicrobial properties of PRP preparations are beginning to emerge and very few studies have studied PRP antimicrobial mechanism. Today, we don’t know all mechanisms by which platelets in- duce antimicrobial properties and it is in a near past that platelets have been discovered to induce antimicrobial effect. However, one thing is clear, PRP due to its rich compositions made of different bioactive molecules, most of them secreted by platelets represents a promising agent in fighting bacterial infections. This aspect of PRP is particularly important to elucidate since we are witnessing tremendous increase in antimicrobial resistance causing one of the major urgent threats to public health. In addition, antibiotic resistance is making a burden for healthcare system and increases a healthcare cost due to its prolonged hospital admission and treatment failures. In Europe it is estimated that antibiotic resistance correlates with more than nine billion euros per year [11]. S. aureus and P. aeruginosa are most common bacteria found in persistent infections and patients suffering from chronic wound [12]. Furthermore, S. aureus is most common bacteria isolated in infectious arthritis and periprosthetic joint infection (PJI) and is

Evaluation of Platelet-Enriched Plasma Antimicrobial Effect

27

also associated with the highest treatment failure rates [13]. In this study we have proven that PRP can decrease the growth rates of tree selected bacteria S. aureus, P. aeruginosa and E. coli. We have demonstrated the beneficial effect of PRP against selected bacteria, however in regards of treatment, adequate amount, and timing of PRP admissions need to be determined. It has been shown that peak PRP antimicrobial activity against MRSA and carbapenem-resistant P. aeruginosa after in vitro incubation are at the fifth and it continued until tenth hour [4]. It is also important to emphasize that PRP antimicrobial effect also depends on PRP preparation and quality. For example, higher antimicrobial PRP effect is seen in PRP preparations prepared with two-step centrifugation when compared with those prepared with one-step centrifugation [6]. However, for better understanding of PRP antimicrobial properties we need to look for molecules that affect microbial metabolism and understand their method of action. Up to our knowledge there is just one study exploring PRP bioactive molecules that induce antimicrobial effect and no studies correlating its effect with bacteria metabolites [14].

5 Conclusion Based on the obtained results it can be concluded that PRP can induce antimicrobial effect on S. aureus, P. aeruginosa and E. coli. PRP also reduced biofilm formation for P. aeruginosa and E. coli. However, there was no effect on S. aureus biofilm formation. Further research identifying PRP antimicrobial mechanism of action is needed to provide better understanding of PRP antimicrobial potential in fighting antibiotic resistance bacteria. Acknowledgment. We thank for financial support to Ministry of Science, Higher Education and Youth of Canton Sarajevo, Bosnia and Herzegovina (Funding ID: 27–02-35–35137-41/22).

References 1. Wasterlain, A.S., Braun, H.J., Dragoo, J.L.: Contents and formulations of platelet rich plasma. In: Maffulli, N. (eds.) Platelet Rich Plasma in Musculoskeletal Practice. Springer, London (2016) 2. Yeaman, M.R.: Platelets: at the nexus of antimicrobial defence. Nat. Rev. Microbiol. 12(6), 426–437 (2014) 3. Stoodley, P., Sauer, K., Davies, D.G., Costerton, J.W.: Biofilms as complex differentiated communities. Annu. Rev. Microbiol. 56, 187–209 (2002) 4. Çetinkaya, R.A., et al.: Platelet-rich plasma as an additional therapeutic option for infected wounds with multi-drug resistant bacteria: in vitro antibacterial activity study. Eur. J. Trauma Emerg. Surg. 45(3), 555–565 (2019) 5. Cie´slik-Bielecka, A., Bold, T., Ziółkowski, G., Pierchała, M., Królikowska, A., Reichert, P.: Antibacterial activity of leukocyte- and platelet-rich plasma: an in vitro study. Biomed. Res. Int. 27(2018), 947–1723 (2018) 6. Maghsoudi, O., Ranjbar, R., Mirjalili, S.H., Fasihi-Ramandi, M.: Inhibitory activities of platelet-rich and platelet-poor plasma on the growth of pathogenic bacteria. Iran. J. Pathol. 12(1), 79–87 (2017)

28

T. Be´cirevi´c et al.

7. Smith, O.J., Wicaksana, A., Davidson, D., Spratt, D., Mosahebi, A.: An evaluation of the bacteriostatic effect of platelet-rich plasma. Int. Wound J. 18(4), 448–456 (2021) 8. Gilbertie, J.M., et al.: Platelet-rich plasma lysate displays antibiofilm properties and restores anti-microbial activity against synovial fluid biofilms in vitro. J. Orthop. Res. 38(6), 1365– 1374 (2020) 9. Christensen, G.D., Simpson, W.A., Bisno, A.L., Beachey, E.H.: Adherence of slime-producing strains of Staphylococcus epidermidis to smooth surfaces. Infect. Immun. 37(1), 318–326 (1982) 10. Gupta, S., Paliczak, A., Delgado, D.: Evidence-based indications of platelet-rich plasma therapy. Expert Rev. Hematol. 14(1), 97–108 (2021) 11. Dadgostar, P.: Antimicrobial resistance: implications and costs. Infect Drug Resist. 20(12), 3903–3910 (2019) 12. Wolcott, R.D., et al.: Analysis of the chronic wound microbiota of 2,963 patients by 16S rDNA pyrosequencing. Wound Repair Regen. 24(1), 163–174 (2016) 13. Tong, S.Y., Davis, J.S., Eichenberger, E., Holland, T.L., Fowler, V.G., Jr.: Staphylococcus aureus infections: epidemiology, pathophysiology, clinical manifestations, and management. Clin. Microbiol. Rev. 28(3), 603–661 (2015) 14. Mariani, E., et al.: Platelet-rich plasma affects bacterial growth in vitro. Cytotherapy 16(9), 1294–1304 (2014)

The Influence of Popular Antibiotics in Culture Medium on the Cell Motility and Superoxide Production of Mesenchymal Stem Cells Inna Zumberg(B)

, Larisa Chmelikova , and Vratislav Cmiel

Department of Biomedical Engineering, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12, Brno 61600, Czech Republic [email protected]

Abstract. Antibiotics have become an integral part of routine cell culturing. Penicillin-streptomycin mixture and gentamicin are widely used to prevent bacterial contamination of cells. A lot of publications have elucidated the effect of these antibiotics in culture medium on differentiation and metabolism of mesenchymal stem cells (MSCs). However, much less is known about their influence on cell migration and mitochondrial superoxide production. In this study, MSCs from human adipose tissue were cultured in media containing mostly used concentrations of mentioned antibiotics. In total, two concentrations of penicillinstreptomycin (100 U·ml−1 –100 μg·ml−1 , 200 U·ml−1 –200 μg·ml−1 ) and gentamicin (10 μg·ml−1 , 50 μg·ml−1 ) were tested. The acquisition of microscopic images was performed using Leica TCS SP8 X confocal laser scanning microscope. The MitoSOX Red indicator was used to observe mitochondrial superoxide. Image processing and statistical analysis was performed using MATLAB software. We found that the addition of these antibiotics to the medium negatively affects the cell migration velocity in a dose-dependent manner, with penicillin-streptomycin inhibiting cell migration less than gentamicin. Also, penicillin-streptomycin and gentamicin induce mitochondrial superoxide production in cells. Keywords: MSCs · Penicillin-streptomycin · Gentamicin · Superoxide · Cell migration

1 Introduction Cell culture can be defined as a process of isolating animal or plant cells continued by their maintaining (or culturing) in controlled artificial conditions, often provided by an incubator. Cell cultures provide a platform for researchers to make in vitro experiments on identical cells to reach highly reproducible results. Compared to cells in a complex organism in which integrity and sterility are still maintained, isolated cells are exposed to non-sterile environments. As a consequence, this leads to increased risk of cell culture contamination. Once it happens, the best choice for contaminated cells is to get rid of them. Another way is to treat the cells with high doses of antibiotics, but this will cause the loss of reagents, cultureware, time, and efforts. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 29–36, 2024. https://doi.org/10.1007/978-3-031-49068-2_4

30

I. Zumberg et al.

To prevent all these problems, various antibiotics are usually added to the culture medium during routine cell culturing. The most widely used antibiotics in order to avoid bacterial contamination are penicillin-streptomycin [1] and gentamicin [2]. Penicillin belongs to a group of β-lactam antibiotics that inhibit bacterial cell wall biosynthesis. Streptomycin and gentamicin are aminoglycoside antibiotics that induce bacterial death by blocking the ability of 30S ribosomal units to synthetize proteins. The general recommended working concentration ranges from 0.5 to 50 μg·ml−1 for gentamicin. For penicillin-streptomycin solution, the concentration of 100 U·ml−1 of penicillin and 100 μg·ml−1 of streptomycin is commonly used. An abundant number of scientific reports related to cell culturing declare the application of the mentioned concentrations of antibiotics [3–6]. There are also plenty of studies, where higher than recommended concentrations of penicillin-streptomycin were used [7–9]. The toxic effect of higher gentamicin content in medium on different cells has already been demonstrated [10, 11], so greater than recommended working concentrations of gentamicin are almost never used. Human MSCs (hMSCs) are multipotent stem cells that have an ability to differentiate into a number of cell types. Different studies have shown that MSCs are able to accelerate wound healing, suppress tumor growth and modulate immune response [12]. These properties make them promising for use in regenerative medicine, tissue engineering and cancer treatment. Mesenchymal stem cell migration is a key aspect, enabling them to execute their functions. Therefore, the aim of this study is to evaluate the effect of penicillin-streptomycin and gentamicin in culture medium on mesenchymal stem cell migration velocity and superoxide production by mitochondria.

2 Materials and Methods 2.1 Cell Culture hMSCs from adipose tissue (C-12977, PromoCell) were thawed at passage 4 and cultured in 12-well plate (92112, TPP) containing low-serum cell growth medium for mesenchymal stem cells (C-28009, PromoCell) at 5% CO2 and 37 °C. Also, each well except the control was supplemented with appropriate antibiotic (penicillin-streptomycin (P4333, Sigma-Aldrich) or gentamicin (G1272, Sigma-Aldrich)). Table 1. Tested groups. Group

Used antibiotic

Control –

Concentration 0 μg·ml−1

PS

Penicillin-streptomycin 100 U·ml−1 of penicillin and 100 μg·ml−1 of streptomycin

2PS

Penicillin-streptomycin 200 U·ml−1 of penicillin and 200 μg·ml−1 of streptomycin

G

Gentamicin

10 μg·ml−1 of gentamicin

5G

Gentamicin

50 μg·ml−1 of gentamicin

The Influence of Popular Antibiotics in Culture Medium

31

In each experiment, 5 wells were used to create 5 tested groups: control, PS, 2PS, G and 5G (see Table 1). For cell passaging, cells were treated with accutase (SCR005, Merck Millipore) after reaching 90% confluence. 2.2 Cell Migration Study For cell migration study, cells were seeded in 8-well chambered cover glass (C8-1.5HN, Cellvis) at a density of 5 × 103 cells per well. After one day of incubation, cell migration was monitored by Leica TCS SP8 X confocal laser scanning microscope (in bright field mode) using the stage top incubator (The Stage Top Chamber, OKOLAB). Three selected ROI from each well were monitored during 18 h in 15-min intervals. Thereby, a sequence of 73 images from each ROI was obtained (image size 1024 × 1024 px, pixel size 1.136 μm × 1.136 μm, 10 × magnification). The image sequences were then processed using the algorithm previously described in [13]. Briefly, the algorithm allows to track the migration of single cells in MATLAB software, calculate the migration parameters for each cell and then visualize the result in a form of rose plot. The cell migration parameters include accumulated distance (the length of cell migration path), velocity (calculated as an accumulated distance divided by total observation time) and Euclidean distance (describing the straight-line distance between start and end position of cell) [14]. Rose plot shows single-cell trajectories, where the beginning of each trajectory is normalized to the coordinate system center. In total, the migration of 45 cells from 3 independent experiments within each tested group was monitored. The obtained rose plots are shown in Fig. 1. 2.3 Mitochondrial Superoxide Production MitoSOX Red Mitochondrial Superoxide Indicator (M36008, Invitrogen) was used for visualization of superoxide produced by mitochondria. 5 mM stock solution was prepared by dissolving the content of a vial (50 μg) in 13 μl of dimethyl sulfoxide (DMSO, MERCI). For labelling cells, 5 μM working solution was prepared by diluting 1 μl of stock solution in 1 ml of Hanks’ Balanced Salt solution with calcium and magnesium (HBSS, 55037C, SAFC, Sigma Aldrich). For monitoring the mitochondrial superoxide production, cells were seeded in 35 mm glass bottom dishes with 14 mm well size (D35-14-1.5-N, Cellvis) at a density of 15 × 103 cells per well. After two days of incubation, the media were aspirated. Then, each dish was washed with warm HBSS and filled with 1 ml of MitoSOX Red working solution. Cells were incubated for 15 min at 37 °C and 5% CO2 . Finally, the dishes were washed three times and filled with warm buffer. The images (image size 2048 × 2048 px, pixel size 0.568 μm × 0.568 μm) were acquired by Leica TCS SP8 X confocal laser scanning microscope using 10 × magnification. The excitation wavelength was set to 510 nm, while the emission was measured in range of 550 and 610 nm. The images obtained from each group are represented in Fig. 3. The acquired images were then processed using a custom algorithm in MATLAB software. The algorithm allows to generate a mask of segmented cells for each image and then to calculate mean intensity of fluorescence in cell areas only.

32

I. Zumberg et al.

3 Results The first aim of this study was to test whether the use of antibiotics during routine cell culturing has any (positive or negative) effect on cell migration. In Fig. 1, we can observe that the rose plot trajectories in control group are longer than in other groups. This observation is also confirmed by Fig. 2, showing that the median value of trajectory lengths decreases with increasing antibiotic concentration. Furthermore, statistical analysis was provided. A one-way ANOVA was used for evaluation. The normality of data and the homogeneity of variances were confirmed by the Lilliefors test and Bartlett’s test, respectively. First, the cell migration velocity was analyzed. We found that the cell migration velocity is significantly inhibited in all tested groups compared to the cells in control group (p < 0.05). Also, there is a significant difference between cell velocities in G group and 5G group (p = 3·10–6 ). Meanwhile, the difference in cell velocity between PS group and 2PS is not statistically significant (p = 0.95).

Fig. 1. Rose plots for (A) control group (B) PS group (C) 2PS group (D) G group (E) 5G group

Another migration parameter to analyze was the directness of cell movement that is calculated by dividing the Euclidean distance by accumulated distance (see Fig. 2). Here, the non-parametric Kruskal-Wallis test was used. As a result, the directness of cell migration was not influenced by antibiotics in all groups compared to control group (p > 0.05). The second aim of this study was to compare the mitochondrial superoxide production in hMSCs cultured with and without antibiotics. The mean intensity of fluorescence measured in each group is 36.84 ± 0.95 (control group), 39.30 ± 1.32 (PS group), 42.31 ± 0.80 (2PS group), 38.04 ± 0.92 (G group), and 45.72 ± 2.28 (5G group). As shown

The Influence of Popular Antibiotics in Culture Medium

33

Fig. 2. Parameters of hMSCs migration quantified in each tested group.

Fig. 3. Mitochondrial superoxide production by hMSCs in (A) control group (B) PS group (C) 2PS group (D) G group (E) 5G group. Scale bar 250 μm.

in Fig. 3, the intensity of fluorescence slightly increases with increasing antibiotic concentration, which indicates the increasing production of mitochondria superoxide. The statistical analysis was provided using one-way ANOVA (as described before). It was discovered that only the higher concentrations of penicillin-streptomycin and gentamicin in culture medium (i.e., 2PS and 5G groups) significantly increase the superoxide production compared to control group (p < 0.05). Conversely, the lower concentrations have not been shown to cause statistically significant increase in superoxide production (p > 0.05 for PS and G groups).

34

I. Zumberg et al.

4 Discussion Adding antibiotics to cell culture media helps to prevent bacterial contamination of the cells and to avoid many of the problems associated with it. Penicillin-streptomycin and gentamicin, probably the most popular antibiotics used in cell cultures, allow to maintain the aseptic conditions that are necessary for cell culture handling. All of this makes researchers’ work easier and saves a lot of time. Much research has been devoted to the study of side effects of both antibiotics in culture medium on cell metabolism and morphology. It has been shown that high doses of penicillin-streptomycin and gentamicin have an antiproliferative effect on mammalian cells, decrease the gene expression and induce apoptosis [15]. Gentamicin at a concentration of 50 μg·ml−1 changes cell morphology, inhibits mitochondria membrane potential and increases lactate and mitochondria superoxide production in human mammary epithelial cells [11]. The standard concentration of penicillin (100 U·ml−1 ) and streptomycin (100 μg·ml−1 ) affects calcium ion channel function in human induced pluripotent stem cell-derived cardiomyocytes [16]. There are also a few studies that describe the side effects of mentioned antibiotics on hMSCs. Penicillin-streptomycin and gentamicin decrease the cell growth rate, inhibit cell proliferation and affect cell differentiation and mitochondrial activity [10, 17, 18]. In this study, we investigate the effect of culture medium supplemented with these antibiotics on the hMSCs motility and mitochondrial superoxide production. We found that the recommended concentration of penicillin (100 U·ml−1 ) and streptomycin (100 μg·ml−1 ) negatively affects cell migration velocity as well as double concentration. In the case of gentamicin, both tested concentrations are now considered as recommended. Here we have demonstrated that gentamicin inhibits cell migration velocity at both concentrations, with the lowest migration velocity observed in the group, where the 50 μg·ml−1 concentration was used. Also, the directness of cell migration was examined in all groups. The results indicate that penicillin-streptomycin and gentamicin added to the culture medium do not have any impact on the directness of hMSCs migration. Superoxide as a free radical from a group of reactive oxygen species (ROS) produced mainly in mitochondria plays an important role in different physiological processes including cell differentiation and immune response. Increased production of superoxide is related to the increased oxidative stress that causes cell damage and aging [19]. The evaluation of mitochondria superoxide production in this study demonstrates that the standard recommended concentration of penicillin (100 U·ml−1 ) and streptomycin (100 μg·ml−1 ) as well as the lower recommended concentration of gentamicin (10 μg·ml−1 ) do not induce higher superoxide production compared to the cells cultured without antibiotics. In contrast, higher concentrations of mentioned antibiotics cause a significant increase in superoxide production. These findings lead to the conclusion that the absence of antibiotics in culture medium is worth considering at least during the execution of the experiments. Otherwise, it will be impossible to analyze changes in cell migration caused by a specific substance if it is tested in cell culture medium already containing antibiotic that affect cell migration [20, 21]. The same applies to superoxide level measurement in the cells [22] and other cellular aspects evaluation [23].

The Influence of Popular Antibiotics in Culture Medium

35

5 Conclusion Penicillin-streptomycin are widely used as culture media supplements that help to avoid bacterial contamination and simplify the maintenance of cells outside the organism. In recent years, a significant number of studies have been published that examine the influence of such antibiotics in medium on differentiation, proliferation and metabolism of human mesenchymal stem cells. However, the effect of antibiotics on cell migration remained unknown. In this study, it was confirmed that penicillin-streptomycin and gentamicin added to the culture medium at recommended concentrations negatively affect cell migration velocity and cause the oxidative stress in hMSCs. Based on the results of this study, it is highly recommended to weigh and consider the use of antibiotics in culture medium when planning and performing experiments on cell culture to obtain more accurate and reproducible results.

References 1. Penicillin-Streptomycin: https://www.sigmaaldrich.com/CZ/en/product/sigma/p4333?con text=product. Accessed 17 Feb 2023 2. Gentamicin solution: https://www.sigmaaldrich.com/CZ/en/product/sigma/g1272. Accessed 17 Feb 2023 3. Schulte-Michels, J., et al.: Anti-inflammatory effects of ivy leaves dry extract: influence on transcriptional activity of NFκB. Inflammopharmacology 27, 339–347 (2019). https://doi. org/10.1007/s10787-018-0494-9 4. Ceylan, U., et al.: Clozapine regulates microglia and is effective in chronic experimental autoimmune encephalomyelitis. Front. Immunol. 12, 656941 (2021). https://doi.org/10.3389/ fimmu.2021.656941 5. Petrenko, Y., et al.: A comparative analysis of multipotent mesenchymal stromal cells derived from different sources, with a focus on neuroregenerative potential. Sci. Rep. 10, 4290 (2020). https://doi.org/10.1038/s41598-020-61167-z 6. Cholet, J., et al.: Anti-inflammatory and antioxidant activity of an extract of luzula sylvatica in a co-culture model of fibroblasts and macrophages. Curr. Res. Compl. Alternat. Med. 6(1), 152. https://doi.org/10.29011/2577-2201.100052 7. Zhang, Y., et al.: Osteoblast behaviors on titania nanotube and mesopore layers. Regen. Biomater. 4(2), 81–87 (2017). https://doi.org/10.1093/rb/rbw042 8. Shin, J., et al.: Effect of doxycycline on epithelial-mesenchymal transition via the p38/Smad pathway in respiratory epithelial cells. Am. J. Rhinol. Allergy 31(2), 71–77 (2017). https:// doi.org/10.2500/ajra.2017.31.4410 9. Tweedell, R.E., et al.: The selection of a hepatocyte cell line susceptible to Plasmodium falciparum sporozoite invasion that is associated with expression of glypican-3. Front. Microbiol. 10 (2019). https://doi.org/10.3389/fmicb.2019.00127 10. Chang, Y., et al.: Toxic effects of gentamicin on marrow-derived human mesenchymal stem cells. Clin. Orthop. Relat. Res. 452, 242–249 (2006). https://doi.org/10.1097/01.blo.000022 9324.75911.c7 11. Elliott, R.L., Jiang, X.: The adverse effect of gentamicin on cell metabolism in three cultured mammary cell lines: “Are cell culture data skewed?”. PLOS ONE 14(4) (2019) 12. Lim, S.K., Khoo, B.Y.: An overview of mesenchymal stem cells and their potential therapeutic benefits in cancer therapy (Review). Oncol. Lett. 22(5), 785 (2021). https://doi.org/10.3892/ ol.2021.13046

36

I. Zumberg et al.

13. Zumberg, I.: Monitoring the parameters of cells migrating in pseudo-3D extracellular matrix. In: Proceedings I of the 27th Conference STUDENT EEICT 2021, pp. 306–311. Brno, Czech Republic: Vysoké uˇcení technické v Brnˇe, Fakulta elektrotechniky a komunikaˇcních technologií (2021) 14. Zengel, P., et al.: μ-Slide chemotaxis: a new chamber for long-term chemotaxis studies. BMC Cell Biol. 12, 21 (2011). https://doi.org/10.1186/1471-2121-12-21 15. Hassan, S., Farizan, A.: The relevance of antibiotic supplements in mammalian cell cultures: towards a paradigm shift. Gulhane Med. J. 62(4), 224–230 (2020). https://doi.org/10.4274/ gulhane.galenos.2020.871 16. Hyun, S-W., et al.: The effects of gentamicin and penicillin/streptomycin on the electrophysiology of human induced pluripotent stem cell-derived cardiomyocytes in manual patch clamp and multi-electrode array system. J. Pharmacol. Toxicol. Methods 91, 1–6 (2018). https://doi. org/10.1016/j.vascn.2017.12.002 17. Llobet, L., et al.: Side effects of culture media antibiotics on cell differentiation. Tissue Eng. Part C 21(11) (2015). https://doi.org/10.1089/ten.TEC.2015.0062 18. Skubis, A., et al.: Impact of antibiotics on the proliferation and differentiation of human adipose-derived mesenchymal stem cells. Int. J. Mol. Sci. 18(12), 2522 (2017). https://doi. org/10.3390/ijms18122522 19. Indo, H.P., et al.: A mitochondrial superoxide theory for oxidative stress diseases and aging. J. Clin. Biochem. Nutr. 56(1), 1–7 (2015). https://doi.org/10.3164/jcbn.14-42 20. Wahl, E., et al.: Acute stimulation of mesenchymal stem cells with cigarette smoke extract affects their migration, differentiation and paracrine potential. Sci. Rep. 6, 22957 (2016). https://doi.org/10.1038/srep22957 21. Naaldijk, Y., et al.: Migrational changes of mesenchymal stem cells in response to cytokines, growth factors, hypoxia, and aging. Exp. Cell Res. 338(1), 97–104 (2015). https://doi.org/10. 1016/j.yexcr.2015.08.019 22. Wang, X., et al.: Hypoxic preconditioning combined with curcumin promotes cell survival and mitochondrial quality of bone marrow mesenchymal stem cells, and accelerates cutaneous wound healing via PGC-1α/SIRT3/HIF-1α signaling. Free Radic. Biol. Med. 159, 164–176 (2020). https://doi.org/10.1016/j.freeradbiomed.2020.07.023 23. Suh, N., Lee, E.: Antioxidant effects of selenocysteine on replicative senescence in human adipose-derived mesenchymal stem cells. BMB Rep. 50(11), 572–577 (2017). https://doi.org/ 10.5483/bmbrep.2017.50.11.174

Treatment Planning for Electrochemotherapy of Spinal Metastases Helena Cindriˇc(B)

, Damijan Miklavˇciˇc , and Bor Kos

Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Slovenia [email protected]

Abstract. Bone metastases can be effectively treated by electrochemotherapy (ECT). Clinical studies report a reduction in pain, preservation of bone stability, and absence of neurologic symptoms after treatment. The transpedicular approach is a promising approach for the treatment of spinal metastases with ECT. The success and safety of treatment can be improved with computer-assisted treatment planning. Patients’ medical images are used to create an anatomically correct numerical models in which optimal electrode placement and treatment parameters are determined to ensure complete tumor coverage and minimal risk of spinal cord injury. We developed an algorithm to optimize electrode positions and voltages, which significantly reduces the time and expertise required to develop a treatment plan for the transpedicular approach. Keywords: Bone metastases · Transpedicular approach · Electroporation · Treat-ment planning · Minimally invasive treatment

1 Electrochemotherapy of Bone Metastases Spinal metastases are a common complication in cancer patients. Symptoms include acute pain, fractures, limited mobility, and neurological dysfunction, which significantly affect patients’ quality of life [1]. The goal of treating spinal metastases is primarily palliative; however, the treatment is particularly complicated because the mechanical stability of the spine needs to be preserved, and the spinal cord and nerves are often involved. Electroporation is a phenomenon in which short high-voltage electric pulses are tissue to tissues that alter the integrity of cell membranes and consequently increase their permeability. Electroporation can be either reversible, where cell membranes completely recover afterwards, or irreversible, where the membranes cannot recover, and the cells die. Electrochemotherapy (ECT) is a combination of reversible electroporation of the tumor mass and administration of specific chemotherapeutic agents (bleomycin or cisplatin). ECT significantly increases the cytotoxicity of chemotherapeutic agents, thereby increasing the efficacy of treatment and reducing the negative side effects of chemotherapy [2]. Several preclinical and clinical studies have shown that bone metastases can be effectively treated by ECT at various skeletal sites [1, 3] and in the spine [4, 5]. All © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 37–42, 2024. https://doi.org/10.1007/978-3-031-49068-2_5

38

H. Cindriˇc et al.

studies report a remarkable reduction in pain after ECT, no bone instability, and no neural symptoms associated with treatment. Electroporation also offers several advantages over other treatment options, as it preserves the structure and density of bone tissue, enables bone regeneration, and has low neural toxicity [6, 7]. Therefore, ECT is a promising, minimally invasive option for the treatment of spinal metastases.

2 Transpedicular Approach for Treating Spinal Metastases In 2018, we presented a new minimally invasive approach to treat spinal metastases with electrochemotherapy—the so called transpedicular approach [8]. In this approach the needle electrodes for ECT are inserted into the vertebral body through the pedicles, similar to the placement of the pedicular screws for spinal fixation surgery (Fig. 1). The electrodes are shorter than what is currently used for ECT of bones and are gradually retracted during treatment, allowing better targeting of the tumor.

Fig. 1. Transpedicular approach: electrodes for ECT of a spinal metastasis (shown in blue) are inserted into the vertebral body through the pedicles. Figure is reproduced from [8].

We used numerical modeling to investigate the feasibility and safety of the proposed approach for the treatment of spinal metastases. To develop a numerical model of spinal metastases we numerically reconstructed the first clinical case of ECT of a spinal metastasis, presented by Gasbarrini et al. [4]. We used experimental data from a preclinical study on a sheep vertebra [7] and measurements from the Gasbarrini case to fine-tune and validate the model. Two additional cases of patients with spinal metastases with varying degrees of spinal canal and pedicle involvement were collected, and patient-specific numerical models were constructed for each case. A treatment plan with the transpedicular approach was developed for all three cases. Using numerical calculations, we evaluated the treatment outcome and the potential risk of spinal cord injury. The results of our numerical study (Fig. 2) demonstrated that the

Treatment Planning for Electrochemotherapy of Spinal Metastases

39

transpedicular approach is a feasible and safe option for minimally invasive treatment of spinal metastases. In 2019, the first clinical case report was published [5] in which two patients with spinal metastases were treated with percutaneous electrochemotherapy via a transpedicular approach. The treatment achieved local tumor control as well as pain relief and disability improvement.

3 Numerical Modelling for Treatment Planning It is generally accepted, that electroporation occurs in tissue when the local electric field strength exceeds a certain (tissue-specific) threshold. Therefore, a prerequisite for successful ECT is complete coverage of the tumor volume with an electric field above the threshold. Determining the distribution of electric field in tissue is not a trivial task; it depends on the geometry of the electrodes used for pulse delivery, the pulse parameters, and the electrical properties of the tissue. The target area usually contains different types of tissues, which have very different electrical properties. Numerical modelling is an effective way of predicting the electric field distribution for selected pulse parameters, electrodes, and tissues [9]. Numerical models of electroporation are based on solving the Laplace equation for electric potential V (Eq. 1): ∇ · (σ ∇V ) = 0

(1)

σ → σ (E)

(2)

Tissue electrical conductivity σ is not constant during electroporation. In short, after the tissue is reversibly electroporated, the base electrical conductivity increases as a function of the local electric field E (Eq. 2). If the tissue becomes irreversibly electroporated, the conductivity reaches its maximum value and plateaus. In most models a sigmoid function is used to describe the increase in conductivity [9, 10].

4 Construction of Patient-Specific Treatment Plans In planning electroporation-based treatments, the main objective is to determine the optimal electrode positions and applied voltage amplitude that ensure complete coverage of the CTV (tumor mass with a safety margin of 5–10 mm) and avoid damage to nearby sensitive anatomical structures such as the spinal cord and nerves. To create a patient-specific treatment plan, the patient’s medical images are first segmented into tissues of interest, e.g., the tumor, healthy bone, spinal cord, and nerves. An anatomically accurate numerical model is then created from the segmented tissue masks (Fig. 1) and imported into finite element analysis software, such as COMSOL Multiphysics (Comsol Inc, Sweden). The model is finalized by assigning the (nonlinear) electrical properties to each tissue in the model. The next step is to determine the electrode positions and pulse parameters, while taking into account any anatomical and technical constraints. For the transpedicular approach the number of electrodes and their position relative to the tumor needs to be determined, as well as any electrode retractions. The last step is to determine the voltage

40

H. Cindriˇc et al.

amplitude to be delivered to specific electrode pairs. The final distribution of the electric field in the tissue is then calculated using the finite element software. The finalized treatment plan provides the physician with a graphical representation of the electrode placement, the optimal pulse parameters to be delivered to specific electrode pairs, the expected electric current consumption, and a visualization of the expected electric field distribution in the tissue and coverage of the CTV (Fig. 2).

Fig. 2. A visual representation of electric field distributions and coverage, shown as an overlay on patients’ CT images. RE–reversible electroporation threshold; IRE–irreversible electroporation threshold; UT–untreated tissue. Figure is reproduced from [8].

5 Optimization of Electrode Positions Treatment planning is still mainly performed manually. Typically, multiple iterations are needed, and the process requires a significant knowledge on the distribution of the electric field in inhomogeneous tissue and expertise in modelling. Most attempts to optimize this process have been based on genetic algorithms, which are very time consuming and require high computational power. Recently, we presented an algorithm for optimizing electrode positions (and voltage amplitudes) for ECT of spinal metastases using the transpedicular approach [11]. This study was the first attempt to use spatial information about the distribution of the

Treatment Planning for Electrochemotherapy of Spinal Metastases

41

electric field in the tissue to optimize electrode positioning and pulses without using computationally intensive genetic algorithms. In developing the algorithm, we adhered to the concept of a manual approach, but the iterative process is fully automated, requiring minimal operator input. The operator must identify two points in the patients’ CT scan for each electrode (Fig. 3A), which determine the starting positions of two electrodes (Fig. 3B). The algorithm then calculates the distribution of the electric field in the CTV, identifies any inadequately treated regions, and uses this information to iteratively move the electrodes from their starting positions towards the target, to cover the entire CTV (Fig. 3C). The algorithm considers several constraints related to electrode positions and the technical limitations of the pulse generator. The algorithm is executed in MATLAB (MathWorks, USA) but the calculations are performed in COMSOL Multiphysics and are connected to MATLAB via LiveLink.

Fig. 3. A) Point selection shown on the axial CT slice. B) Vertebral model showing the tumor and starting electrode positions, obtained from the selected points. C) Electrode positions are adjusted by the algorithm. Figure is adapted from [11].

The performance of the algorithm was tested on realistic vertebral models (created from patients’ CT images) to which spherical tumors of different sizes were added, yielding a total of 108 test models. The algorithm performed successfully for different spinal segments, tumor sizes, and tumor positions within the vertebral body. The average time to find a solution was 71 s (range: 17–253 s), and the average number of iterations was 4.9 (range: 1–15). The algorithm presents a significant improvement over finding a solution with genetic algorithms that require at least 100 iterations. The planning process requires a much lower level of expertise on the part of the operator, which is also an improvement over the manual approach. However, before a treatment planning workflow can be established using the presented algorithm, validation must be performed on real clinical cases of spinal tumors. The source code of the algorithm and the entire test data set have been made publicly available in an online repository (https://doi.org/10.6084/m9.figshare.212 70111.v1).

42

H. Cindriˇc et al.

6 Conclusions Electrochemotherapy is a promising treatment for painful bone metastases that offers several advantages over existing treatments. Clinical studies report a remarkable reduction in pain after treatment. Numerical studies and initial clinical reports indicate that the transpedicular approach is a feasible and safe method for treating spinal metastases. The use of computer assisted treatment planning and fast and robust optimization could make electrochemotherapy treatment safer and more accessible to treating physicians and, as an extension, to patients.

References 1. Campanacci, L., Bianchi, G., Cevolani, L., et al.: Operating procedures for electrochemotherapy in bone metastases: results from a multicenter prospective study on 102 patients. Eur. J. Surg. Oncol. 47, 2609–2617 (2021). https://doi.org/10.1016/j.ejso.2021.05.004 2. Campana, L.G., Edhemovic, I., Soden, D., et al.: Electrochemotherapy–emerging applications technical advances, new indications, combined approaches, and multi-institutional collaboration. Eur. J. Surg. Oncol. 45, 92–102 (2019). https://doi.org/10.1016/j.ejso.2018. 11.023 3. Bianchi, G., Campanacci, L., Ronchetti, M., Donati, D.: Electrochemotherapy in the treatment of bone metastases: a Phase II trial. World J. Surg. 40, 3088–3094 (2016). https://doi.org/10. 1007/s00268-016-3627-6 4. Gasbarrini, A., Campos, W.K., Campanacci, L., Boriani, S.: Electrochemotherapy to metastatic spinal melanoma: a novel treatment of spinal metastasis? Spine 40, E1340-1346 (2015). https://doi.org/10.1097/BRS.0000000000001125 5. Cornelis, F.H., Ben Ammar, M., Nouri-Neuville, M., et al.: Percutaneous image-guided electrochemotherapy of spine metastases: initial experience. Cardiovasc. Intervent. Radiol. 42, 1806–1809 (2019). https://doi.org/10.1007/s00270-019-02316-4 6. Fini, M., Salamanna, F., Parrilli, A., et al.: Electrochemotherapy is effective in the treatment of rat bone metastases. Clin. Exp. Metastasis 30, 1033–1045 (2013). https://doi.org/10.1007/ s10585-013-9601-x 7. Tschon, M., Salamanna, F., Ronchetti, M., et al.: Feasibility of electroporation in bone and in the surrounding clinically relevant structures: a preclinical investigation. Technol. Cancer Res. Treat. 15, 737–748 (2016). https://doi.org/10.1177/1533034615604454 8. Cindriˇc, H., Kos, B., Tedesco, G., et al.: Electrochemotherapy of spinal metastases using transpedicular approach—a numerical feasibility study. Technol. Cancer Res. Treat. 17, 1533034618770253 (2018). https://doi.org/10.1177/1533034618770253 9. Corovic, S., Lackovic, I., Sustaric, P., et al.: Modeling of electric field distribution in tissues during electroporation. Biomed. Eng. Online 12, 16 (2013). https://doi.org/10.1186/1475925X-12-16 10. Pavšelj, N., Bregar, Z., Cukjati, D., et al.: The course of tissue permeabilization studied on a mathematical model of a subcutaneous tumor in small animals. IEEE Trans. Biomed. Eng. 52, 1373–1381 (2005). https://doi.org/10.1109/TBME.2005.851524 11. Cindriˇc, H., Miklavˇciˇc, D., Cornelis, F.H., Kos, B.: Optimization of transpedicular electrode insertion for electroporation-based treatments of vertebral tumors. Cancers 14, 5412 (2022). https://doi.org/10.3390/cancers14215412

Bio-micro/nano Technologies

Development of Polymer-Based Nanoparticles for the Reduction of Melittin Toxicity Berrin Chatzi Memet1,2 , Eren Demirpolat3 , Turgay Yildirim4 and Omer Aydin1,2,5,6(B)

,

1 NanoThera Lab, ERFARMA-Drug Application and Research Center, Erciyes University,

38280 Kayseri, Turkey [email protected], [email protected] 2 Department of Biomedical Engineering, Faculty of Engineering, Erciyes University, Kayseri 38039, Turkey 3 Department of Pharmacognosy, Faculty of Pharmacy, Erciyes University, Kayseri 38280, Turkey 4 Helmholtz Institute for Pharmaceutical Research, Saarland University, 66440 Saarland, Germany 5 ERNAM-Nanotechnology Research and Application Center, Erciyes University, Kayseri 38039, Turkey 6 ERKAM-Clinical Engineering Research and Implementation Center, Erciyes University, Kayseri 38030, Turkey

Abstract. Melittin is a naturally occurring cytotoxic peptide, derived from a component in the poison of the European honeybee Apis mellifera. Bee venom has been used for the treatment of pain, rheumatoid arthritis, and chronic inflammatory diseases, and its knowns with anti-cancer, anti-microbial, anti-viral peptide, and other pharmacological properties. Melittin is a peptide that has potential in the treatment of various diseases. However, melittin induces pore formation in the cell membrane due to its positive charge and causes cell lysis. To overcome this disadvantage and take advantage of the potential effect of melittin, novel negatively charged PMMA/SPMA was synthesized by RAFT polymerization, and nanoparticles were formed by the nanoprecipitation method. After melittin was complexed with the nanoparticles, the size of the nanoparticles were slightly increased from 223 to 550 nm, whereas the charge of the nanoparticles were increased from −40 to −2.97 mV. Critic micelle concentration (CMC) was 17 μg/mL and the binding efficiency was determined as 83.45% by measuring the fluorescence density in the supernatants. The cytotoxicity results indicate that the toxicity of melittin (IC50: 2.23 μM) has been successfully abolished with MEL NPs in Raw 264.7 cells. These nanoparticles have the potential to facilitate the successful and effective utilization of melittin in various treatments, creating potential for its use in diverse therapies. Keywords: Melittin · Polymeric nanoparticle · Micelle · Anti-cancer peptide

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 45–50, 2024. https://doi.org/10.1007/978-3-031-49068-2_6

46

B. Chatzi Memet et al.

1 Introduction Nanotechnology is not just a theoretical concept for the future world; it is inherently multidisciplinary and encompasses various fields such as engineering, physics, chemistry, biology, and many others. Nanotechnology, as a technology that enables scientific and engineering activities at the nanoscale, has become the key technology of the 21st century [1, 2]. The advantage of nanotechnology lies in the unique and fundamentally different properties exhibited by materials at the nanoscale compared to larger-size bulk materials composed of the same substance. These properties include physical, chemical, electrical, mechanical, and biological characteristics. Reducing material dimensions to the nanometer level creates phase interfaces that are crucial for enhancing material properties. Nanoparticles present unique opportunities for addressing challenges in various fields, including the medical and pharmaceutical industries, food packaging, electronics, and the energy industry [3]. By harnessing the properties of nanoparticles, we can explore new avenues for overcoming obstacles and achieving advancements in these diverse sectors. Bee venom is a biotoxin or apitoxin synthesized and secreted by a gland located in the bee’s abdominal cavity. Bee venom consists of a complex mixture of various biologically active peptides such as melittin, including enzymes like phospholipase A2, bioactive amines, and components with various non-peptide pharmaceutical properties. Melittin is the main naturally occurring peptide derived from the toxic component in the venom of the European honeybee Apis mellifera. It is a small-sized, water-soluble [4, 5], linear, cationic, anti-inflammatory [6] and amphipathic peptide consisting of 26 amino acids [7–11]. The amino terminal region (residues 1–20) of this peptide is predominantly hydrophobic, the carboxyl-terminal region (residues 21–26) is hydrophilic due to the presence of a stretch of positively charged amino acids [12], generally predominantly hydrophobic [8–11] (Fig. 1).

Fig. 1. Amino acid sequence of melittin.

Melittin is an antibacterial peptide known to have pharmacological effects [13]. It also has other pharmacological properties such as anti-tumor [14], anti-inflammatory [6], anti-HIV [15], anti-cancer [16], anti-viral [17], analgesic [18], and anti-mutagenic

Development of Polymer-Based Nanoparticles for the Reduction

47

[19]. In addition, melittin has a +6 charge, which it binds to the negatively charged membrane surface and then induces pore formation in the cell membrane, ultimately leading to cell lysis [7, 8, 20, 21]. For the same reason, the non-specific cytolytic activity of melittin can cause serious off-target effects such as hemolysis when administered intravenously [10, 21]. The aim of this study is to develop a nanotechnology-based drug delivery system to eliminate these non-specific side effects of melittin such as hemolysis and cell lysis and benefit from its all-pharmacological properties. In this study, novel poly(methyl methacrylate)-b-sulfopropyl methacrylate polymer was synthesized and, nanoparticles were obtained from this polymer and complexed with melittin (MEL NP) to eliminate the toxic effect.

2 Experimental 2.1 Synhtesis of Poly(Methyl Methacrylate)-b-Sulfopropyl Methacrylate (PMMA-b-SPMA) Copolymer of poly(methyl methacrylate) and sulfopropyl methacrylate was synthesized by reversible addition–fragmentation chain transfer (RAFT) polymerization. Briefly, poly(methyl methacrylate) (PMMA) (macro RAFT agent) (128 mmol), 3-Sulfopropyl methacrylate potassium salt (SPMA) (64 mmol), AIBN (2.2’ azobisisobutyronitrile) (128 mmol) and 1,3,5-trioxane were dissolved in of N,N-Dimethylformamide (DMF) and Dimethyl sulfoxide (DMSO). Immersed in pre-heated oil bath for 6h. 2.2 Micelle Formation and Complexation with Melittin Polymeric nanoparticles were obtained using the nanoprecipitation method. Melittin complex with polymeric nanoparticles (MEL NP) was realized by electrostatic bonding. To remove unbound melittin, centrifugation at 15000 rpm at 0 °C was applied. 2.3 Characteriztion Critic Micelle Concentration (CMC) was determined using nile red [22] encapsulation into polymeric nanoparticles. Nuclear magnetic resonance (1 H NMR) (Brucker Avance 400 MHz) was used to determine the conversion ratio of monomer to polymer. Size and surface charge analysis of complexed and empty nanoparticles were performed using Nano ZS (Malvern Zetasizer). The binding efficiency (BE) of the formed complexes was calculated using the BCA Assay (bikinchoninic acid assay). The following equation was used to calculate the binding efficiency. BE% =

Total Melittin Weight − Free Melittin Weight x100 Total Melittin Weight

(1)

48

B. Chatzi Memet et al.

2.4 Cytotoxicity Assay RAW 264.7 murine macrophage cells were seeded in 96-well plates at a seeding density of 1x104 cells/well and allowed to adhere for 24 h before incubating with empty NP, and MEL NP for 24h.

3 Results and Discussion 3.1 Synthesis of Amphiphilic PMMA-b-SPMA Polymer To ensure chain end retention, polymerization was stopped after 6 h at 70% monomer conversion. The resulting M n values (calculated according to DP of polymer) from 1 H NMR spectrum of PMMA-b-SPMA polymer was 8277.64 g/mol. 3.2 Micelle Formation and Characterization The critical micelle concentration (CMC) of PMMA-b-SPMA polymer was determined via monitoring the change in the intensity of NR fluorescence upon increasing polymer concentration. When low concentrations of PMMA-b-SPMA polymer were present, the baseline NR fluorescence increased slightly as the polymer concentration increased above 17 μg/mL. Table 1. DLS and ζ-potential results of prepared nanoparticles. Zeta-average [d.nm]

PDI

ζ-potential [mV]

Empty NP

223 ± 4

0.07

− 40 ± 0.89

MEL NP

550 ± 64

0.3

− 2.97 ± 0.64

The decrement in ζ- potential and increment in size of the complexed nanoparticles is probably due to the success binding of melittin (Table 1). The binding efficiency was calculated by measuring the fluorescence density in the supernatants of the complexes (Fig. 2). According to the fluorescence intensity the binding efficiency was calculated 83.45% ± 3.48 using Eq. 1. 3.3 Cytotoxicity of MEL NP Firstly we determined the IC50 value of melittin on RAW 264.7 cell line. Then we compared the effect of empty nanoparticles and MEL NP. The IC50 value of melittin at 24 h was calculated 2.23 μM. When the cytotoxicity results are examined, as shown in Fig. 3 synthesized nanoparticles completely eliminate the toxicity of melittin. No toxic effects were observed even in nanoparticles containing melittin concentrations above IC50 of melittin. These results indicate the biocompatibility of the polymeric carrier.

Development of Polymer-Based Nanoparticles for the Reduction

49

Fig. 2. Dinamic light scattering and ζ-potential results. (A) Size of empty nanoparticle. (B) Surface charge of empty nanoparticle. (C) Size of complexed MEL NP. (D) Surface charge of complexed MEL NP.

Fig. 3. Cytotoxicity assay of empty and complexed nanoparticles. There is no toxic effect of melittin above its IC50 concentration when complexed with nanoparticles.

4 Conclution In consequence, biocompatible polymeric nanoparticles were obtained that eliminated the toxicity of melittin. These melittin complexes have potential in many biomedical applications and treatment of varying diseases. Acknowledgement. This study has been supported by Erciyes University Scientific Research Project Coordination Unit under grant number TYL-22–11454. Berrin Chatzi Memet is supported by the TUBITAK (121C221).

50

B. Chatzi Memet et al.

References 1. Ozimek, L., Pospiech, E., Aliment, S.N.J.A.S.P.T.: Nanotechnologies in food and meat processing. 9(4), 401–412 (2010) 2. Maksimovi´c, M., Omanovi´c-Mikliˇcanin, E., Badnjevi´c, A.: How technology can help? In: Maksimovi´c, M., Omanovi´c-Mikliˇcanin, E., Badnjevi´c, A. (eds.) Nanofood and Internet of Nano Things: For the Next Generation of Agriculture and Food Sciences, pp. 9–86. Springer International Publishing, Cham (2019) 3. Omanovi´c-Mikliˇcanin, E., et al.: Nanocomposites: a brief review. Heal. Technol. 10(1), 51–59 (2020) 4. Al-Rabia, M.W., et al.: Repurposing of sitagliptin-melittin optimized nanoformula against SARS-CoV-2; antiviral screening and molecular docking studies. 13(3), 307 (2021) 5. Gajski, G., Garaj-Vrhovac, V.: Melittin: a lytic peptide with anticancer properties. Environ. Toxicol. Pharmacol. 36(2), 697–705 (2013) 6. Shin, S.H., et al.: The effects of melittin and apamin on airborne fungi-induced chemical mediator and extracellular matrix production from nasal polyp fibroblasts. Toxins (Basel) 9(11) (2017) 7. Rady, I., et al.: Melittin, a major peptide component of bee venom, and its conjugates in cancer therapy. Cancer Lett. 402, 16–31 (2017) 8. Gonzalez-Horta, A., et al.: Biodegradable nanoparticles loaded with tetrameric melittin: preparation and membrane disruption evaluation. Gen. Physiol. Biophys. 36(4), 373–381 (2017) 9. Memariani, H., et al.: Melittin: a venom-derived peptide with promising anti-viral properties. Eur. J. Clin. Microbiol. Infect. Dis. 39(1), 5–17 (2020) 10. Memariani, H., et al.: Melittin: from honeybees to superbugs. Appl. Microbiol. Biotechnol. 103(8), 3265–3276 (2019) 11. Raghuraman, H., Chattopadhyay, A.: Melittin: a membrane-active peptide with diverse functions. Biosci. Rep. 27(4–5), 189–223 (2007) 12. Klocek, G., Seelig, J.: Melittin interaction with sulfated cell surface sugars. Biochemistry 47(9), 2841–2849 (2008) 13. Habermann, E.: Bee and wasp venoms. Science 177(4046), 314–322 (1972) 14. Brogden, K.A.: Antimicrobial peptides: pore formers or metabolic inhibitors in bacteria? Nat. Rev. Microbiol. 3(3), 238–250 (2005) 15. Uzair, B., et al.: Potential uses of venom proteins in treatment of HIV. Protein Pept. Lett. 25(7), 619–625 (2018) 16. Aufschnaiter, A., et al.: Apitoxin and its components against cancer. Neurodegen. Rheumat. Arthrit. Limit. Possibil. 12(2), 66 (2020) 17. Pascoal, A., et al.: An overview of the bioactive compounds, therapeutic properties and toxic effects of apitoxin. Food Chem. Toxicol. 134, 110864 (2019) 18. Lin, L., Zhu, B.P., Cai, L.: Therapeutic effect of melittin on a rat model of chronic prostatitis induced by complete Freund’s adjuvant. Biomed. Pharmacother. 90, 921–927 (2017) 19. Al-Ani, I., et al.: Pharmacological synergism of bee venom and melittin with antibiotics and plant secondary metabolites against multi-drug resistant microbial pathogens. Phytomedicine 22(2), 245–255 (2015) 20. Lee, S.-J., et al.: Interaction of melittin peptides with perfluorocarbon nanoemulsion particles. J. Phys. Chem. B 115(51), 15271–15279 (2011) 21. Cheng, B., Xu, P.: Redox-sensitive nanocomplex for targeted delivery of melittin. Toxins (Basel) 12(9) (2020) 22. Aydin, O., et al.: Formulation of acid-sensitive Micelles for delivery of cabazitaxel into prostate cancer cells. Mol. Pharm. 13(4), 1413–1429 (2016)

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both Biosynthesizing Silver Nanoparticles and Anti-microbial Activity Berna Oyku Ozbey1,2

and Gulizar Caliskan3(B)

1 Department of Stem Cell and Tissue Engineering, Institute of Health Sciences, Istinye

University, 34010 Istanbul, Turkey 2 Pera Labs, 3675 Market St Suite 200, Philadelphia 38039, PA, USA 3 Department of Genetics and Bioengineering, Faculty of Engineering, Izmir University of

Economics, Izmir 35330, Turkey [email protected]

Abstract. Green synthesis has recently become one of the most popular methods, as it is both low-budget and environmentally friendly. One of the important considerations in green synthesis is to perform an optimization study because it is necessary to understand how different application conditions (pH, incubation time, metal concentration, etc.) can affect the formation of nanoparticles with different morphology and efficiency, underlining the need for optimization of the process. In this study, firstly the endemic Inula Viscosa (L.) plant, popularly known as cancer grass, was extracted using distillation method. Then, silver nanoparticle (AgNPs) biosynthesis was carried out using the extract of Inula Viscosa (L.) plant. Their physicochemical characterization was conducted using Fourier-transformed infrared spectroscopy (FTIR), UV-visible spectrophotometry (UV-Vis), Scanning Electron Microscopy (SEM), and Dynamic Light Scattering (DLS). The time, pH, and AgNO3 concentration, which affect the characteristic and morphological properties of AgNPs, were optimized with the Box Behnken Design (BBD) method, with statistical and experimental design determined by means of a Design Expert statistical software program. The disk diffusion method was also implemented and optimized to increase antimicrobial activity. The study determined the optimal levels of AgNPs, which were green synthesized by Inula Viscosa (L.), provided proof of its antimicrobial properties, and demonstrated their potential to be used as a low-budget aid to new generation clinical treatment methods. Keywords: Silver nanoparticles · Biosynthesis · Inula Viscosa (L.) · Antimicrobial activity

1 Introduction Nanotechnology focuses on structures with dimensions below 100nm, and can be defined as an interdisciplinary technology. Recently, in nanotechnology, there has been a trend towards green synthesis method, which is low cost, environmentally friendly and relatively easy to synthesize. In chemical synthesis, the product is synthesized in a shorter © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 51–61, 2024. https://doi.org/10.1007/978-3-031-49068-2_7

52

B. O. Ozbey et al.

time and in desired sizes, but this is a toxic and expensive method. Also, the yield is higher in green synthesis, which uses various biological sources such as plant extracts, bacteria, microalgae and fungi as reducing agents [1]. Inula viscosa (L), which was used in the study, is an endemic species in countries such as Turkey, Algeria, and Tunisia, is also popularly called ‘cancer grass’ and is a plant widely used in traditional medicine. Among the phytochemicals in Inula Viscosa (L.), tomentosin in particular, has anticancer, antimicrobial, and anti-inflammatory properties. However, there are few medical studies on this plant in the literature [2]. These few preliminary studies show that it is important to investigate the potential of the plant in nanoparticle synthesis. Thus, we focused on the effect of the extract of this plant on silver nanoparticle synthesis. One of the major reasons for the selection of silver nanoparticles is that these are known to play an effective role in destroying the permeability of the bacterial membrane [3]. Chemical methods used to synthesize silver nanoparticles increase the harmfulness of the product by using toxic solvents, and reduce its potential. Due to these limitations, it was considered more appropriate to select the biosynthesis method, which is environmentally friendly and has less toxic effects [4]. So far, there has been no attempt to achieve silver nanoparticle biosynthesis from the extracts of Inula viscosa (L.) or its species in the literature. This study sets out to determine the nanoparticle synthesis potential and usage areas of Inula viscosa (L.), which have not yet been studied, and also to optimize efficiency, which has never been previously attempted. At the same time, optimization of the process is important for biological synthesis, considering that different application conditions (pH, incubation time, metal concentration, etc.) may cause nanoparticle formation with different morphology and efficiency [5]. Box Behnken Design (BBD) method, a response surface methodology, was used for optimization steps with Design-Expert program and optimization was found for the effect of incubation time (30–240 min), pH (4–8) and AgNO3 (metal) concentration (1–8 mM) on the average size diameter in nanoparticle synthesis. After nanoparticle production, characterization was performed [6] and the characterization of nanoparticles was conducted using FourierTransformed Infrared Spectroscopy (FTIR), Scanning Electron Microscopy (SEM), UV-Vis spectrophotometry and Dynamic Light Scattering (DLS), and disc diffusion method.

2 Material and Method 2.1 The Extraction of the Plant Extract of Inula viscosa (L.) plant was obtained using the infusion technique of extraction process, which included distilled water as a solvent. In the first step, the raw plant was passed through the grinder to obtain its ground form, and in the second, the infusion system was established. In the infusion system, distilled water is transferred to a 500 mL beaker, and the magnetic stirrer is run at 350 rpm and set at 60 °C. Then the ground herb is added to the beaker, and allowed to infuse for 20–30 min. The temperature is maintained at 60 °C or below in order to obtain the most optimal level of the plant extract without losing its phytochemical and antimicrobial properties. After this process, a vacuum filtration device and Whatman No:1 filter paper were used to eliminate

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both

53

unwanted particles. The plant extract obtained after filtration was stored at + 4 °C for use in nanoparticle synthesis. 2.2 The Experiment Management with the Box Behnken Design Method A three-level fractional factorial design developed by Box Behnken Design (BBD) was applied to determine the nature of the surface response in an experimental region. This design has many advantages, such as having three levels that can be coded as 1 (low), 0 (medium), and + 1 (high), creating an independent quadratic design and facilitating the organization and interpretation of results. Each block consists of the maximum and minimum values, the factorial design values, and the factors’ central values. The BBD method has the advantage that it saves time as it has fewer factor levels and has no excessively high or low number/levels of experiments. The BBD method reduces the number of experimental sets and is widely used in biotechnological research [7]. The parameters to be used in nanoparticle optimization in this project and their minimum and maximum values are as follows. – Incubation time (30–240 min) – pH (4- 8). – AgNO3 concentration (1–8 mM) [8, 9]. 2.3 The Synthesis Silver Nanoparticles (AgNPs) by Using Inula Viscosa (L.) Extract The BBD method was used to both identify and optimize parameters that affect various unique properties of nanoparticles after plant extraction. For optimum nanoparticle synthesis using the extract of Inula Viscosa (L.) plant, 17 experimental design runs were determined by Design Expert program. This involved the use of metal concentrations of 1, 4.5 and 8 mM AgNO3. The pH values of 3 different metal concentrations were adjusted as 4, 6 and 8 using a pH meter. Then, for the synthesis of silver nanoparticles, the metal solution and plant extract were prepared in a 1:1 ratio in a working volume of 40 mL. The stock solution containing the plant extract was added dropwise to the medium containing AgNO3 at a rate of 1 drop/s, thus initiating the nanoparticle production process. Incubation was completed at 25 °C at a mixing speed of 200 rpm using a shaking incubator at intervals appropriate to the experimental design. The aim of stirring was to increase the interaction of the plant extract with the metal solution, to produce nanoparticles in smaller sizes, and to prevent aggregation and homogeneous distribution. After the incubation period for all experimental trials, procedures were applied to calculate the nanoparticle concentration and the samples were dried and stored using the lyophilization process before characterization. 2.4 The Characterization of Nanoparticles In this study, UV-Vis spectrophotometry, FTIR, SEM and DLS characterization methods were used to determine the properties of the synthesized nanoparticles. UV-Vis spectrophotometry is a technique used to measure the light absorbed and emitted by a sample. UV-Vis spectrophotometry is very important for the characterization

54

B. O. Ozbey et al.

and examination of nanoparticles due to its optical properties, which are sensitive to nanoparticles. UV visible region devices operate between 200 and 900 nm [10]. In order to identify structural, compositional and functional groups associated with AgNPs formations, Infra-red spectra analysis of AgNPs was performed using Per-kin Elmer Spectrum FTIR Spectrometer at room temperature, within the range of 4004000 cm−1 [9, 10]. Scanning Electron Microscopy (SEM) works by scanning the particle surface with high-energy electrons to characterize the morphology of nanoparticles [11, 12]. AgNPs coated with Au-Pd under 7.50 kV high vacuum were recorded by Scanning Electron Microscopy (Thermo Scientific Apreo S). Malvern Zeta Sizer was used to find the harmonic mean diameters and polydispersity index (PI). 2.5 The Antimicrobial Activity Test Agar disk diffusion method was used to examine the antimicrobial activity of plantbased nanoparticles [13]. The test was performed on gram-negative (Escherichia coli (EA)) and gram-positive (Staphylococcus aureus (SA)) bacteria, and the antimicrobial effect of nanoparticles was compared with the control group in terms of the diameter of white plaque formed on the agar surface after incubation under appropriate conditions. All steps were followed inside the Class II Laminar Flow Biosafety Cabinet to ensure aseptic conditions. The results were evaluated in the Design Expert program.

3 Result and Discussion The nanoparticles synthesized by green synthesis were produced sustainably from the plant source, and during this process, cell metabolites were also used. In this study, optimized results were obtained after characterization. This study can also provide a model for further studies on biological synthesis for this selected Inula Viscosa (L.) plant species. In addition, there are very few studies on Inula viscosa (L.) plant used in the study at either national or international levels. The reasons for choosing to study a flowering plant species native to Turkey, and its- other bioactive compounds are the plant’s anti-cancer, anti-microbial, total flavonoid content, total phenolic values, and the resulting high antioxidant activity. At the same time, antimicrobial tests have proven the antimicrobial effects of nanoparticles biosynthesized with the extract of this plant. Addressing the lack of research on this herb is one of the main goals of the study. The anticancer and antimicrobial effects of the plant species Inula Viscosa (L.) have already been reported [14], but the current study is notable as the first to carry out the green synthesis of Inula viscosa (L.) plant based nanoparticles with silver. 3.1 The Results from Characterization Studies The factors chosen for optimization are metal concentration, incubation time and pH values that affect the biosynthesis of nanoparticles.

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both

55

UV-Vis spectrophotometry is a technique used to measure the light absorbed and emitted by a sample, using the measurement of the intensity of a beam after passing through it. The absorption here is mostly due to the excitation of bond electrons in the molecules. It is known that silver nanoparticles exhibit a yellowish or brownish color in aqueous solutions due to their dispersion and surface plasmon resonance absorption, which has characteristic optical absorption in the visible region. The absorbance value giving the highest peak is Run 11 at approximately 270 nm. For test run 1, the wavelength value is 258.97 nm, the absorbance value is 9.7217, and the full width at half maximum (FWHM) value is 1.149 nm. For test trial 4, the wavelength value is 293.78 nm, the absorbance value is 10, and the FWHM value is 6.7897 nm. For test trial 11, the wavelength value is 295.26 nm, the absorbance value is 10, and the FWHM value is 4.7154 nm. In general, absorbance and wavelength values were very similar.. In addition, the peak points are also within the range given for silver nanoparticles in the literature, i.e., in the region of 300–450 nm [15]. The surface plasmon bands of all NPs were observed in the range of 290–400 nm, which is as expected for this metal (Fig. 1).

Fig. 1. Surface plasmon resonance values of nanoparticles

The FTIR spectra of silver nanoparticles synthesized by using Inula viscosa (L.) plant extract is shown in Fig. 2. FTIR results for AgNPs with average 56 nm, 200 nm size depicted in Fig. 2 coded as r11 and r3, respectively. Distinct peaks were observed at 3263.74, 1519.02, 1596.39, 1370.42, 1262.30, 1155.52, 1066.64, 813.05 and 523.74 cm−1 for AgNPs whereas for r3 coded AgNPs, the peaks were observed at 1598.08, 1076.3 cm−1 only. The peaks at around 3400 cm−1 are the bond vibration of water molecules [5]. Vibrations of C = O and CC = groups ranged from 1650 and 1500 cm−1 [5]. The surface morphology and size distribution of the AgNPs was examined by scanning electron microscopy (SEM). SEM images show that spherical shaped particles in range from 40 to 250 nm are prevalent (Fig. 3). The minimum nanoparticle sizes are as follows, respectively; Trial 1 is 59.42- 59.23– 62.98 nm, Trial 4 is 48.26–45.94–52.11 nm, and Trial 11 is 42.66–62.58–62.98 nm. The nanoparticle sizes synthesized by green synthesis under normal conditions are expected to be larger than 100–1000, and sometimes as much as 200 nm. Nanoparticles synthesized with chemical substances are generally very likely to be smaller than 50 nm because

56

B. O. Ozbey et al.

Fig. 2. FTIR-ATR analysis results

Fig. 3. SEM images of AgNPs

their skeletons have been formed beforehand [16]. On the other hand, according to the biosynthesis and extraction process using water only in this project, the nanoparticle sizes are generally below 100. This provides the desired size range, according to the literature. The produced nanoparticle had “Good” quality, shown by the zeta-potential value, low aggregation formation, harmonic mean diameter of nanoparticle size obtained as a result of DLS characterization. AgNPs showed a homogeneous distribution with a polydispersity index (PI) value as a 0.1938. This value represents the distribution of size within a sample. PDI values of 0.2 and below are acceptable for metal nanoparticles [17]. The DLS results in our study appear to have been successful in terms of monodispersal and stability. In addition, the nanoparticle concentration after lyophilization was calculated to be approximately 1.5 mgNP/mL. According to the BBD, the F value of the Model is 33.23%. A value of p < 0.0001 indicates that the model is significant (Table 1). The probability of such a large “Model F-Value” due to noise is only 0.01%. Overall values of Probe > F less than 0.05 indicate that the model terms are significant. In this case B (incubation time), C (metal concentration), AB (pH-incubation time), AC (pH-metal concentration), BC (Incubation timemetal concentration), B2 (incubation time-incubation time), C2 (metal concentrationmetal concentration) can be considered important model terms, as long as there is another

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both

57

parameter with values greater than 0.1, indicating that the model terms are not significant, i.e. the model reduction option model. However, there is no trivial model term in this study. A “Lack of Fit F-value” of 0.13 means no Missing Fit significant to pure error. Such a large “Missing F-value” due to noise has a probability of occurrence of 94.02%. An insignificant lack of fit is interpreted positively, because it is a desirable for the model to fit or be compatible [18, 19]. Table 1. Results of the ANOVA Source

SS

Model

13808.73

9

1534.30

33.23

< 0.0001

165.62

1

165.62

3.59

0.1001

2457.01

1

2457.01

53.21

0.0002

505.62

1

505.62

10.95

0.0130

AB

1162.81

1

1162.81

25.18

0.0015

AC

2830.24

1

2830.24

61.30

0.0001

BC

380.25

1

380.25

8.24

0.0240

A2

28.35

1

28.35

0.61

0.4589

B2

5903.70

1

5903.70

127.86

< 0.0001

C2

496.13

1

496.13

10.74

0.0135

Residual

323.22

7

46.17

A-pH B-incubation time C-metal concentration

DF

MS

F-value

Lack of fit

27.81

3

9.27

Pure error

295.41

4

73.85

Core total

14131.95

16

p-value

0.13

SS, sum of squares; DF, degrees of freedom; MS, mean square

The “Pred R-Square” value of 0.9359 is in reasonable agreement with the “Adj RSquare” of 0.9477. “Adeq Precision” measures the signal-to-noise ratio. It is expected to have a ratio greater than 4. The obtained ratio of 24.277 is an indication of a sufficient signal, and therefore, the optimization success [20]. These results are shown in Table 1. Appropriate trial patterns were examined at the points where the harmonic mean particle diameter (nm) was optimized to a minimum. As can be seen from Fig. 4, when both the pH and the incubation time increased, the harmonic mean particle diameter decreased. The three-dimensional (3 D) visual of the model also indicates that 8 mM silver concentration is suitable to diminish the nanoparticle diameters in the solution. Based on the overall characterization results, it can be said that metal concentration is the most important determinant of success in nanoparticle synthesis, followed by small size, low p-value, volume, and other important values.

58

B. O. Ozbey et al.

Fig. 4. A 3 D interaction plot of harmonic mean particle diameter of AgNPs

3.2 The Results from Antimicrobial Activity Studies The first part of the study involves the optimization of the harmonic mean particle diameter of plant-based nanoparticles. The second part presents an optimization of the antimicrobial activity of these nanoparticles. Design expert results of antimicrobial activities against the E.coli and S.aureus are shown in Figs. 5 and 6, respectively. In this set of experiments, the highest desirability to achieve the maximum antimicrobial activity for EC was optimized and obtained as 11.06 mm for EC, where the metal concentration was 8 mM, the incubation time was 30 min, pH at 6.46 and the harmonic mean diameter was 99 nm (Fig. 5). After the statistical analysis of the model was conducted with analysis of variance (ANOVA), the F value of the model was found as 20.57, which indicates that the model is significant (p < 0.05). According to the ANOVA test, the metal concentration and pH and the interactions of these two variables were found to be significant (p < 0.05). The “Pred R-Squared” of 0.7812 is in reasonable agreement with the “Adj R-Squared” of 0.8607. In this set of experiments, the highest desirability to achieve the maximum antimicrobial activity for SA was optimized and obtained as 12.5 mm for SA, where the metal concentration was 8 mM, the incubation time was 30 min, ph at 6.5 and the harmonic mean diameter was 98 nm (Fig. 6). The statistical analysis of the model was made with an analysis of variance (ANOVA). The model F value of 14.98 implies that the model is significant (p < 0.05). According to the ANOVA test, the metal concentration and pH and their interactions were found to be significant (p < 0.05). The coefficient of determination (R2 ) was used to check the fitness of the model. An R2 value closer to one implies a better correlation between experimental and predicted responses, such as more fit model explanation. In this model, the correlation coefficient (R2 ) value of 0.8736 is in reasonable agreement with the adjusted determination coefficient (R2 Adj) value of 0.8153 in terms of the high significance of the model.

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both

59

Fig. 5. Design expert results of antimicrobial activities against the E.coli

Fig. 6. Design expert results of antimicrobial activities against the S.aureus

Therefore, the antimicrobial activity showed a reasonable effect on these plant- based nanoparticles. Also, the green synthesized nanoparticles produced showed, interestingly, 99% antistatic properties. Consequently, it is clear that advanced characterization methods are suitable for investigating the optimization potential of nanoparticle production specific to Inula Viscosa (L.) plant, and this approach should be developed in further studies. 3.3 Conclusions The biological synthesis of nanoparticles from plants has been extensively studied, yet no study has focused specifically on the potential of biological components of Inula viscosa (L.) as anticancer, antimicrobial reagent in the synthesis of silver nanoparticles. The present work is a model study using optimization of both synthesis and

60

B. O. Ozbey et al.

antimicrobial activity of silver nanoparticles. Moreover, in these efforts, it is essential to choose endemic plants, and understand the effects of different parameters such as incubation time, metal concentrations, pH level and their interactions on both biosynthesis of nanoparticles and antimicrobial effects, and also, to conduct eco-friendly green synthesis. AgNPs of Inula viscosa (L.) indicated significant antibacterial effect against E. coli and S. aureus bacteria in this study, which will guide future studies on the pharmaceutical and biomedical applications of biogenic AgNPs. Nanoparticles were identified with certain characterization features in our study; however, stabilization and further characterization studies are still needed. Future studies should also consider challenges related to biosynthesized AgNPs such as safety profile, genotoxicity, pharmacokinetics, and antibacterial resistance. Eventually, coating techniques with biocompatible chitosan etc. may be employed to increase stabilization. Another possible innovative direction is the evaluation of these nanoparticles in drug delivery systems and anticancer research. In short, these special nanoparticles are material components with great potential for development. This present study shows the potential for applications of synthesized AgNPs as a novel antibacterial and cytotoxic agent for biomedical applications such as drug delivery, biosensor, and hyperthermia.

References 1. Korkmaz, N., Kaçan, F.N., Kaya, M.Y.: Çevre dostu gümü¸s nanoparçacık sentezi. Bozok J. Eng. Arch. 1(1), 14–20 (2022) 2. Khan, A.L., Hussain, J., Hamayun, M., Gilani, S.A., Ahmad, S., Rehman, G., et al.: Secondary metabolites from Inula britannica L. and their biological activities. Molecules 15(3):1562– 1577 (2010) 3. Li, W.R., Xie, X.B., Shi, Q.S., Zeng, H.Y., Yang, Y.S., Chen, Y.B.: Antibacterial activity and mechanism of silver nanoparticles on Escherichia coli. Appl. Microbiol. Biotechnol. 85(4), 1115–1122 (2010) 4. Abbasi, E., et al.: Silver nanoparticles: synthesis methods, bio-applications and properties. Crit. Rev. Microbiol. 42(2), 173–180 (2016) 5. Caliskan, G., Mutaf, T., Agba, H.C., Elibol, M.: Green synthesis and characterization of titanium nanoparticles using microalga, Phaeodactylum tricornutum. Geomicrobiol. J. 39(1), 83–96 (2022) 6. Onba¸slı, D.: Kitosan-g¨umü s¸ nanopartik¨u l¨u ve kitosan-g¨u mü s¸ -grafen oksit nanokompozitinin sentezi, karakterizasyonu ve antimikrobiyal aktivitelerinin belirlenmesi. Erciyes Ü niversitesi Vet. Fak¨u ltesi Derg. 13(3), 208–215 (2022) 7. Kraisit, P., Hirun, N., Mahadlek, J., Limmatvapirat, S.: Fluconazole-loaded solid lipid nanoparticles (SLNs) as a potential carrier for buccal drug delivery of oral candidiasis treatment using the Box-Behnken design. J. Drug Deliv. Sci. Technol. 63 (2021) 8. Bhau, B.S., Ghosh, S., Puri, S., Borah, B., Sarmah, D.K., Khan, R.: Green synthesis of gold nanoparticles from the leaf extract of Nepenthes khasiana and antimicrobial assay. Adv. Mater. Lett. 6(1), 55–58 (2015) 9. Ballottin, D., et al.: Elucidating protein involvement in the stabilization of the biogenic silver nanoparticles. Nanoscale Res. Lett. 11(1) (2016) 10. Gholami-Shabani, M., et al.: Antimicrobial activity and physical characterization of silver nanoparticles green synthesized using nitrate reductase from Fusarium oxysporum. Appl. Biochem. Biotechnol. 172(8), 4084–4098 (2014)

Endemic Inula Viscosa (L.) Extracts and Their Potential for Both

61

11. Daglıoglu, Y.: Ekotoksisite deneylerinde nanopartik¨u l karakterizasyonunun önemi ve y¨o ntemleri. Marmara Fen Bilim. Derg (2018) 12. Paper, N., Submission, P.: Synthesis of zinc oxide nanoparticles via sol—gel route and their characterization. 5(2009):2010–2014 (2016) 13. Balouiri, M., Sadiki, M., Ibnsouda, S.K.: Methods for in vitro evaluating antimicrobial activity: a review. J. Pharmaceut. Anal. 6(2), 71–79 (2016) 14. Aydin, T., Saglamtas, R., Dogan, B., Kostekci, E., Durmus, R., Cakir, A.: A new specific method forisolation of tomentosin with a high yield from Inula viscosa(L.) and determination of its bioactivities. Phytochem. Anal. 1–7 (2022) 15. Nased, S.M., Khozemy, E.E., Kamoun, E.A., Gendi, H.E.: Gamma radiation-induced crosslinked composite membranes based on polyvinyl alcohol/chitosan/AgNO3/vitamin E for biomedical applications. Int. J. Biol. Macromol. 137, 878–885 (2019) 16. Chandrika, K., Chaudhary, A., Mareedu, T., Sirisha, U., Vangalapati, M.: Adsorptive removal of acridine orange dye by green tea/copper-activated carbon nanoparticles (Gt/Cu- AC np). Mater. Today Proc. 44(1), 2283–2289 (2021) 17. Danaei, M., et al.: Impact of particle size and polydispersity index on the clinical applications of lipidic nanocarrier systems. Pharmaceutics 10(2), 57 (2018) 18. Al Mousa, A.A., Hassane, A.M.A., Gomaa, A.E.-R.F., Aljuriss, J.A., Dahmash, N.D., AboDahab, N.F.: Response-surface statistical optimization of submerged fermentation for pectinase and cellulase production by Mucor circinelloides and M. hiemalis. Fermentation 8, 205 (2022) 19. Chaker, H., Ameur, N., Saidi-Bendahou, K., Djennas, M., Fourmentin, S.: Modeling and Box-Behnken design optimization of photocatalytic parameters for efficient removal of dye by lanthanum-doped mesoporous TiO2. J. Environ. Chem. Eng. 9(1) (2021) 20. Abosabaa, S.A., ElMeshad, A.N., Arafa, M.G.: Chitosan nanocarrier entrapping hydrophilic drugs as advanced polymeric system for dual pharmaceutical and cosmeceutical application: a comprehensive analysis using box-behnken design. Polymers 13(5), 677 (2021)

Metrology in Medical Measurements

Lessons Learned from External Audits in Medical Device Testing Laboratories: Best Practices and Recommendations for Quality Management Baki Karaböce(B) TÜB˙ITAK UME, Gebze Kocaeli 41470, Türkiye [email protected]

Abstract. Calibration, measurement and test laboratories should provide reliable, accurate and timely service in accordance with the technological conditions of the day. The ability of laboratories to provide such a service is explained in detail in the ISO/IEC 17025:2017 “General Requirements for the Competence of Experiment and Calibration Laboratories” standard. There are some unclear situations in the application of the ISO/IEC 17025 standard in laboratories serving in the medical field, in other words, performing calibration, measurement and testing of devices in hospitals. Most of the medical devices are multi-parameter measuring devices. Therefore, the calibrators, analyzers or simulators used in the measurements and testing of these devices are generally multifunctional devices. In this article, the most common problems in the technical accreditation audits of medical laboratories will be examined and solution suggestions will be presented. In this paper, lessons learned from external audits in medical device testing laboratories are shared. Best practices and recommendations for quality management will be presented. Keywords: TS EN ISO/IEC 17025 Standard · Accreditation · Medical devices · Metrology · Audit

1 Introduction ISO/IEC 17025 is applied for laboratories aiming to provide quality service to customers in post-market surveillance [1, 2]. Calibration, measurement and test laboratories are expected to provide reliable, accurate and timely service. A Quality management system (QMS) should be established in order to implement the ISO/IEC 17025 standard. Nonconformance with the practices and activities described in the QMS documentation and/or standards is defined as “non-compliance”. For the continuity and improvement of the quality management system, it is very important to eliminate the problems arising from the audits carried out in accordance with ISO/IEC 17025 by the accreditation institution in every developed or developing country, as well as the findings obtained from the internal audits. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 65–73, 2024. https://doi.org/10.1007/978-3-031-49068-2_8

66

B. Karaböce

Every country has its own accreditation agency. For example, Deutsche Akkreditierungsstelle (DAkkS) in Germany, United States Accreditation Services (USAS) in the USA, Türk Akreditasyon Kurumu-Turkish Accreditation Agency (TÜRKAK) in Turkey and Institute for Accreditation of Bosnia and Herzegovina (BATA) in Bosnia and Herzegovina were assigned [3–6] for accreditation,. After the quality management sys- tem is established, major and minor Non-conformances detected in the first audit differ from those in subsequent audits. In the ongoing audits at certain periods, major non-conformances decrease and minor non-conformances increase, but these can be corrected with corrective actions. The requirements for the accreditation of the test/calibration laboratory are defined in the ISO/IEC 17025 standard, and the routinely applied accreditation audits are carried out according to the conditions specified in this standard [7, 8]. The terms and conditions that must be met according to this standard are described in the section under the headings of Clause 6 “management requirements” and Clause 7 “technical requirements”. Documents listed in Table 1. Should be ready for an audit [9]. In laboratories for testing medical devices that apply ISO 17025 international standard, defining scope of accreditation is particularly important. Medical devices measure various parameters, thus etalons/analyzer used for the testing purposes are also multiparameter devices that make them more critical. Beyond the basic ECG (electrocardiogram), patient monitors are used to quickly review a range of vital signs, both in real time, to better understand a patient’s current condition, improvement or worsening. A typical multiparameter patient monitor simultaneously looks at 12-lead cardiac ECG, oxygen saturation (SpO2), hemoglobin, temperature, invasive or non-invasive blood pressure, respiration, and pacemaker activity. Medical laboratories which conduct performance tests of medical devices apply the same ISO/IEC 17025 and/or ISO/IEC 17020 standard [10–13]. Medical laboratories must prepare quality documents that include device list, traceability, personnel list, procedures, uncertainties, forms, calibration certificates, proficiency tests and comparison measurements participation list, location and activities etc. In some countries, the ministry of health may apply extra regulations other than accreditation.

2 Important Findings in External Audits for Medical Devices In order to get an idea about the quality of the service provided by medical device testing laboratories, its performance in audits can be examined. For most laboratories, non-conformances in traceability, personnel training, and frequent use, transportation and set up of calibrator devices and validity of results are among the most important findings in an accreditation audit in medical measurements and tests [14, 15]. Strong points and positive sides of testing laboratories for medical devices should be mentioned. Since many measurements are realized with many different devices, the device usage information by users increases considerably. Problems (if any) encountered in the field can be resolved quickly. They generally have no customer complaints. In common, handling of test or calibration items, reporting of results, selection, verification and validation of methods are clear for laboratories. Non-conformances will be detailed below.

Lessons Learned from External Audits in Medical Device Testing Laboratories: Best …

67

Table 1. Preparations for ISO/IEC 17025 accreditation No

Factors in uncertainty calculation

Legal status

Documents showing that the body is a legal entity or a part of a legal entity

Professional liability insurance

Professional indemnity insurance that covers the damages of third parties

Authorized signature lists

Authorized signature lists

Scope of service

Information on the areas in which the laboratory provide service

Organization chart and job descriptions

Organization chart and job descriptions

Main document list

Up to date list of all the documentation in the laboratory with re vision status)

Quality handbook

Quality manual of the laboratory or any document for this purpose

Personnel

Qualified and trained personnel

Procedures

Procedures for the implementation of the management system, prepared in accordance with ISO/IEC 17025

Quality instructions and lists

Quality-related instructions prepared in accordance with ISO/IEC 17025

Testing instructions/calibration procedures

All testing/calibration/sampling instructions. It must be specified if in case of using the in-house method

Calibration certificates

Up to date calibration certificates of the devices, the list of calibration frequency, calibration technical specifications, records of evaluation of calibration suitability etc

2.1 Measurement Traceability Traceability is the most important quality show for a medical laboratory. According to item 6.5 Metrological Traceability in ISO/IEC 17025, an accredited laboratory should be able to show traceability of the result declared in the test report it produces to international standard (SI) basic units. For example, in order to verify the sound pressure level, vibration and force value in audiometric devices, an artificial ear and a mastoid which must have calibration certificates are used. They are traceable to primary standards throughout the microphone, accelerometer and force sensor in the relevant measurement range as seen in an example traceability scheme in Fig. 1. Audit findings can be seen in a given certificate. If the certificate is not found, traceability means broken. In this case, all measurement and test reports made with that reference device are withdrawn.

68

B. Karaböce

Fig. 1. Traceability in audiometric measurements

An active and busy medical calibration laboratory usually has 3-4 sets of calibrator device systems. The multiplicity, variety and frequent use of devices can make tracking somewhat difficult, and traceability may be at risk as observed in audits. 2.2 Measurement Uncertainty Under ISO/IEC 17025 requirements, in item 6.5 Metrological Traceability, the best uncertainty of measurement for each accredited method must be calculated. In the test reports given, the measurement uncertainty must be given, provided that it is not lower than this uncertainty. The uncertainty of measurement can be estimated or calculated using different methods. The medical measurement and testing laboratory can decide for itself which method to calculate uncertainty. Whether given in % error or in units, the minimum uncertainty calculation for each measurement should be made in detail. On site, in other words, in the hospital, the most important factors affecting the uncertainty will be device resolution, standard deviation and changes in ambient conditions. Considering that the personnel in the field will be technicians or technicians in some cases, for example, a ready-made excel calculation control sheet can be prepared first for uncertainty calculations. The most accepted and most widely used approach to uncertainty calculation is to follow the steps described in the JCGM 100:2008 Guidance on Expressing Uncertainty in Measurement (GUM) [16]. In a measurement, all factors that can affect the measurement result are taken into account and combined as shown in Table 2. The model function with measurement range is generally not described as observed in many audits. For example an equation for audiometer calibration as seen in Eq. 1 can be a model function for uncertainty calculations. It should be noted that the calculation of total uncertainty is different. Presentation of the uncertainty calculation method is beyond the scope of this article. Depending of different type of measurements, uncertainty sources may be listed as seen in Table 3. E = SPL + RETSPL + δ res + δ mon − SPLan − δ lin − δ resp − δ ear − δ dr + δ rep In Equation 1, SPL: Sound pressure level set on the audiometer RETSPL: Reference equivalent sound pressure level threshold value δres: Error form device resolution

(1)

Lessons Learned from External Audits in Medical Device Testing Laboratories: Best …

69

Table 2. Table captions should be placed above the tables. No

Factors in uncertainty calculation

1

Description of the measured value in terms of measurement process

2

Listing the input quantities

3

Determining the uncertainty for each input quantity

4

Evaluation of any covariances/correlations in input quantities

5

Calculation of the measured value

6

Appropriate combination of the uncertainty components

7

Multiplication of the combined uncertainty by a coverage factor

8

Reporting the results in the proper format

δmon,: Correction due to the effect of mounting the audiometer headset SPLan,: Sound pressure level measured from the analyzer δlin: Correction due to linearity error of the analyzer δresp,: Correction due to the frequency response of the analyzer δear: Artificial ear response correction δdr: Correction for drift in calibration assembly performance since last calibration δrep: Repeatability of measurements Table 3. Possible sources of uncertainty in measurement No

Error sources in uncertainty calculation

1

Incomplete definition of the quantity being measured

2

Imperfect realization of the definition of the quantity being measured

3

Lack of knowledge of environmental effects on measurements

4

Person based error in analog reading

5

Resolution of device under test

6

Inaccurate values of measurement standards

7

Inappropriate approximations and assumptions

8

Rounding errors

9

Variations in values (repeatability) under similar conditions

Generally many medical devices are busy (in use). So the time for handling of devices is short. Lack of time may create errors in measurements. Moreover it may not be possible to take enough data for calculation of repeatability in measurements.

70

B. Karaböce

2.3 ILC and PTs According to item 7.7 Ensuring the Validity of Results in ISO/IEC 17025, medical measurement/test laboratories must participate in proficiency test (PT) at least once every 4 years with the coordination of an independent pilot laboratory, for example a metrology institute, and obtain successful results. Alternatively, it should participate in the interlaboratory comparison (ILC), preferably with the coordination of an independent pilot laboratory, for example a metrology institute, and obtain successful results. According to the ISO/IEC 17043:2010 standard, proficiency testing is the evaluation of participant performance against predetermined criteria through interlaboratory comparisons. In other words, proficiency testing is a method that a laboratory uses to see its proficiency and verify the laboratory’s measurement process by comparing test results with those of a reference laboratory and other participating laboratories. According to the ISO/IEC 17043 standard, the ILC test is the evaluation by comparing the measurement and/or test results of two or more laboratories on the same or similar substances in accordance with predetermined conditions. z (mostly used) scores outside the ≤2.0 limit and En scores outside the ≤1 limit considered as unsuccessful results. If the participating laboratory achieves unsuitable/unsuccessful performance, it is expected that the participating laboratory will promptly initiate corrective action and/or demonstrate that the situation is under control. The formula used for calculating the Z score in Eq. 2. And En score in Eq. 3 are shown below.   (2) z = xlab − Xreference /σPT      (Xlab − Xreference )    En =     U 2 lab + U 2 reference 

(3)

In Eqs. 2 and 3. Xlab = participant’s result, xreference = mean or assigned value, σPT = standard deviation of PT, Ulab = participants’s expanded uncertainty and Ureference = reference’s expanded uncertainty. The absence of an independent reference laboratory in PT or ILC may cast doubt on the results of the participating laboratory. 2.4 Multiparameter Feature of Medical Devices According to item 6.4 Equipment in ISO/IEC 17025, medical laboratory shall establish the metrological traceability of their reference devices. In medical devices has ECG, Respiration, NIBP (non-invasive blood pressure), IBP (invasive blood pressure), Temperature, SpO2, EtCO2 (End-tidal carbon dioxide) that needs time, temperature, and pressure traceability according to standards. Sometimes a medical laboratory that gives service may have traceability in pressure only but it performs the measurements of all parameters.

Lessons Learned from External Audits in Medical Device Testing Laboratories: Best …

71

2.5 Training of Personel Item 6.2 Personnel in ISO/IEC 17025 declares, the laboratory shall have procedure(s) and retain records for training of personnel and determining the competence requirements. Personel training is another problem in medical tests laboratories as observed in audits. Since the staff circulation is high, new staff are constantly coming and the training of the incoming staff cannot be completed on time. Generally, employees prefer an established job. However, medical test laboratories usually go to hospitals in the country for measurements and spend at least 1 week to 1 month at the place they go. This constantly hectic work life can get boring over time. Non-conformances are detected in training certificates, personnel competency evaluation and personnel comparisons. 2.6 Handling of Test or Calibration Items Some devices, such as an ultrasonic power calibrator or an incubator calibrator, for example, require installation prior to use. Some ultrasonic power calibrators are prone to error by design. For example, in the device in Fig. 2, the connection of the reflective target that detects the ultrasonic power to the balance sensor element (load cell) is very complex, and it is prone to distortion and erroneous measurement results. While the direct transfer of the radiation force generated by the ultrasonic probe to the balance sensor is the most accurate way, the indirect transfer (with 5 more extensions as you can see in Fig 2c) of the force as seen in the picture will lead to measurement errors. It is difficult to use the device in this way for a long time and by inexperienced users without errors. There may be errors and time losses in the placement and positioning of around 7 sensors in the incubator calibrator shown in Fig. 3. As defined in the TS EN IEC 606012-19: 2021 standard, it is necessary to manually determine and fix the scales one by one in order to place the sensors in the correct positions. There is a high probability of making mistakes here. During the audits, such devices with faulty, broken and missing parts were frequently encountered.

Fig. 2. Ultrasound power meter (a) whole set up (b) force/power transmission system (c) drawback of force/power transmission system (1 to 5 are extensions)

In audits, ultrasound power calibrator and incubator calibrators are generally problematic and create non-conformances.

72

B. Karaböce

Fig. 3. Incubator calibration system with sensors (a) with sensors (b) with positioned sensors.

3 Conclusions The general non-conformances and their elimination, which emerged in the accreditation audits carried out within the scope of EN ISO/IEC 17025 Standard, are instructive and guiding for the measurement/test laboratory. The findings show us the importance of the document structure to be created in the establishment of the quality system and that it should overlap with the operation of the system [17]. Compliance and validation of the methods used with the practices in the laboratories, meticulous and proper work on this subject, the appropriate qualifications of the devices used in the laboratories, calibration, maintenance, etc [18]. It shows that the processes and the fulfillment of all quality requirements are important in ensuring the continuity of the system. A properly functioning quality system and the fact that the non-conformances identified in the internal audit and initial audits at the beginning can be reduced over time is an expression of the system working properly. Corrective actions determined after the audits carried out by the accreditation institution, elimination of non-conformances, quality control studies, sensitivity to be shown in the reporting stages will contribute to institutional improvement activities. It should be kept in mind that all identified non-conformance will contribute positively to system improvement. The auditing process should be considered as a training, because the auditors who come to the audit are usually experts in their field. Audit results and outputs will guide the activities to be carried out to improve the system and reduce risks by identifying weak points within the scope of the quality system. It is observed that significant non-conformance have decreased after regular audits. It is a corporate prestige expression that means competence and recognized in accreditation according to the ISO/IEC 17025 standard. Traceability, personnel training, and frequent use, transportation and set up of calibrator devices and validity of results constitute the biggest inconveniences for most medical measurement/testing laboratories during an accreditation audit. A certification report from an accredited supplier to ensure traceability can easily get the job done. There are many methods and examples in the literature to calculate measurement uncertainty. With a programmatic management of measurement and testing campaigns, they can more easily overcome device and setup challenges. With the use of these tools at hand, medical laboratories will be in a better position to successfully manage the audit process.

Lessons Learned from External Audits in Medical Device Testing Laboratories: Best …

73

The audit carried out within the scope of ISO/IEC 17025 accreditation can be considered as a consultancy service in a way. Because inspectors often work in the same field and are experts in medical device measurements. They evaluate the technical competence and knowledge of the laboratory.

References 1. Badnjevi´c, A., Pokvi´c, L.G., Be´cirovi´c, L.S., Deumi´c, A.: Post- market surveillance of medical devices: a review. Technol. Health Care (2022) 2. ISO homepage: ISO/IEC 17025:2017 General requirements for the competence of testing and calibration laboratories 3. Dakks homepage: https://www.dakks.de/en/home-en.html 4. USAS homepage: http://usaaccreditation.com/laboratory-accreditation.html 5. BATA homepage: http://www.bata.gov.ba/ 6. TÜRKAK homepage: https://turkak.org.tr/ 7. Badnjevi´c, A., Pokvi´c, L.G.: Legal metrology framework for medical devices. In: Clinical Engineering Handbook (2nd Edn), pp. 774–779. 8. Badnjevi´c, A., Pokvi´c, L.G, Boskovic, D., Dzemic, Z.: Medical devices in legal metrology. In: Conference: 4th Mediterranean Conference on Embedded Computing MECO 2015. https:// doi.org/10.13140/RG.2.1.4741.2324 9. Complying with ISO 17025: https://www.unido.org/sites/default/files/2010-08/Complying_ with_ISO_17025_A_practical_guidebook_0.pdf 10. Pichler, E.: Top three challenges in external accreditation audits, 2016–05–12, https://www. romerlabs.com/fr/centre-d-expertise/base-documentaire/articles/news/top-three-challengesin-external-accreditation-audits/ 11. Bozkurt, E.N., Bayram, G., Gültop, F., Topcu, U., Gevrek, N.: Evaluation of audit findings performed in laboratories according to TS EN ISO, IEC 17025 Standard, January 2017 Türk hijyen ve deneysel biyoloji dergisi. Turkish Bull. Hygiene Exp. Biol. 74(1), 83–94. https:// doi.org/10.5505/TurkHijyen.2016.04557] 12. Ayub, Y., Anwar Z., Shah, Z., Sharif, M.M.: Non-conformities against ISO/IEC 17025:2017 in Pakistani labs: a study based on Auditing Body Reports April 2021. Indus. Eng. Manag. 10(4), 293. Project: ISO 17025:2017 Lab Accreditation (Chemical Sector) 13. Okezue, M.A., Adeyeye, M.C., Byrn, S.J., Abiola, V.O., Clase, K.L.: Impact of ISO/IEC 17025 laboratory accreditation in sub-Saharan Africa: a case study. BMC Health Serv. Res. 20, 1065 (2020). https://doi.org/10.1186/s12913-020-05934-8 14. Zima, T.: Accreditation of medical laboratories–system, process, benefits for labs. J. Med. Biochem. 36(3), 231–237 (2017) 15. Jang, M., Yoon, Y., Song, J., Kim, J., Min, W., Lee, J., et al.: Effect of accreditation on accuracy of diagnostic tests in medical laboratories. Ann Lab Med. 37(3), 213–222 (2017) 16. ISO homepage: https://www.iso.org/standard/50461.html 17. Linko, S.: Internal audits in private medical laboratory practice—a Finnish experience in 1996–2000. Practitioner’s Report. Accred Qual Assur 7, 55–59 (2002) 18. Grochau, I.H., Schwengber ten Caten, C.: A process approach to ISO/IEC 17025 in the implementation of a quality management system in testing laboratories, Practitioner’s Report. Accred Qual Assur 17, 519–27 (2012)

Implementation of Legal Metrology Framework for Medical Devices to Healthcare Sector in the Republic of Uzbekistan Vohobjon Nishonov1(B) , Lemana Spahi´c2 , Amar Deumi´c3 , Ammar Traki´c2 , Najmiddin Muminov4 , Sheroz Ismatullev1 , and Lejla Gurbeta Pokvi´c2 1 Uzbek National Institute of Metrology, Tashkent, Uzbekistan

[email protected]

2 Verlab Research Institute for Biomedical Engineering, Medical Devices and Artificial

Intelligence, Sarajevo, Bosnia and Herzegovina 3 Verlab Ltd., Sarajevo, Bosnia and Herzegovina 4 Tashkent State Agrarian University, Tashkent, Uzbekistan

Abstract. Background: The safety and reliability of medical devices are becoming more emphasized as time passes. Crisis situations such as COVID-19 have shown the negative effects that can be brought upon by a lack of medical device surveillance mechanisms. Objectives: The aim of this study is to present the results of the implementation of the legal metrology framework conducted by joint efforts of the State Center of Expertise and Standardization of Medicines, Medical Devices, and Medical Equipment and the Uzbek National Institute of Metrology. Methods: The study was based on the data collected from annual performance inspections in all healthcare institutions in the period from 2016 to 2021 for 37 types of medical devices. Data envelopment analysis was used to derive conclusions and the results were compared with results from Bosnia and Herzegovina. Results: Results indicate that the implementation of legal metrology the framework leads to a significant increase in the accuracy of medical devices hence leading to increased reliability and patient safety in diagnostic and therapeutic processes. Conclusions: Fault prediction of medical devices enables more effective maintenance strategies thus enhancing cost-effectiveness and decreasing the downtime of equipment. This in turn leads to a more efficient healthcare system capable of facing challenges of the society. Keywords: Legal metrology · Medical devices · Accuracy · Healthcare system

1 Introduction Crisis situations like COVID-19 always point out bottlenecks in the sector in which they occur. Emergency authorization of medical devices during the COVID-19 pandemic has shown a great need for dedicating a significant degree of attention to the entire healthcare © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 74–90, 2024. https://doi.org/10.1007/978-3-031-49068-2_9

Implementation of Legal Metrology Framework for Medical Devices …

75

sector and the operations within with an emphasis on medical devices [1]. Besides skilled healthcare professionals, medical devices are a key component ensuring functionality of the healthcare system [2]. Today, there are over 2 million different kinds of medical devices on the world market, categorized into more than 7000 generic devices groups [3]. More than 90% of diagnoses and treatments made by medical professionals are based on the results of analysis using medical devices, and the remaining 10% are based on other diagnostic methods such as direct physical examination and anamnesis. Since the measurements made by medical devices are used for diagnosis, prevention, monitoring of diseases, and life support, they pose a potential risk to the patient’s life [4]. Medical devices, as such, have to be overseen and regulated at all phases of their lifespan, from ideation phase to utilization in healthcare institutions [5]. Rapid progress in the healthcare systems in recent years has been brought upon by the emergence of biomedical engineering and significant advancements in digital technologies. This rapid growth in both interest and volume of modern medical devices developed and produced each day has made policy makers more aware of adverse events that can occur if the devices are not overseen in a proper way [5]. Hence, medical device oversight while the devices are operating on the market has become more stringent in the past years and enforced by the relevant directives. The European Union and the US Food and Drug Administration have developed databases containing information regarding adverse events caused by medical device malfunction or failure. The Manufacturer User Facility Database (MAUDE) contains reports regarding medical device-associated adverse events and device recalls reported by manufacturers, importers and device user facilities and voluntary reporters such as health care professionals, patients and consumers [6]. European database on medical devices (EUDAMED) is one of the key mechanisms for implementation of new rules on medical devices [7]. It provides the public with evidence-based representation of the lifecycle of medical devices used on the market of the European Union. It integrates a variety of electronic systems that collate and process information regarding medical devices and manufacturers. Hence, it enhances the transparency and coordination of activities related to medical devices between different EU member states. Legislation on medical devices in the European Union (EU) is managed by the European Commission in close cooperation with Health Authorities of the member states. Medical devices manufacturers must have CE Marking, comply with international standards and EU legislation in order to place the certain device on the EU market. The EU legislation has been in a transition period from the Medical Devices Directives (MDD 90/385/EEC and 93/42/EEC) [8, 9] to the Medical Device Regulations (MDR 2017/745 and 2017/746) [10, 11]. The new MDRs entered into force in 2017 and replaced the previous MDDs [10, 11]. In the United States of America (USA), the Food and Drug Administration (FDA) is the responsible institution for regulating medical devices [12]. In Australia, the responsible body for regulating medicine and medical devices is The Therapeutic Goods Administration (TGA) which is part of the Australian Government Department of Health and Ageing [13]. China Food and Drug Administration (CFDA) oversees medical devices and ensures their safety and effectiveness, and protects human health and life in China [14]. This responsibility is taken by Pharmaceuticals and Medical Devices Agency (PMDA) in Japan [15]. In terms of scope, all of the aforementioned documents cover the same,

76

V. Nishonov et al.

they all require premarket evaluation of medical devices, and they mention post-market surveillance. However, the methodology of post-market surveillance is quite variable amongst the aforementioned guidelines and there is a large discrepancy on the market [5]. Countries worldwide have initiated efforts and constructed guidelines for medical device surveillance. Best practices of MD surveillance are those that are evidence-based and Bosnia and Herzegovina [16–22], Portugal [23], Serbia [24] and Saudi Arabia [25] are already implementing these practices. A paper by Badnjevic et al. [5] revises all PMS strategies laid out by national and international guidelines and proposes the best solution that could be used as a reference standard. Evidence-based medical device performance assessment is founded on metrology principles with an aim of ensuring traceability of measurements and treatments made by medical devices. Metrology is the science of measurement process ensuring that measurement meets specified degrees of both accuracy and precision. Metrology institutes from all over the world have different infrastructure and experience in the field of measurements in healthcare. The most influential metrology institutes in the world, that regulate subject areas such as bioscience and health, medical devices, diagnostics, standard reference materials, biomedical optics sections and medical devices metrology and standards are NIST (National Institute of Standards and Technology) [26] from USA, LNE (Laboratoire national de métrologie et d’essais) [27] from France and PTB (The Physikalisch-Technische Bundesanstalt) [28] from Germany. In Spain, Portugal, Saudi Arabia, Republic of Serbia, Bosnia and Herzegovina [5], and the Chezch Republic medical measurement devices are subject to legal metrology and metrological inspection carried out periodically. These inspections are carried out either by metrology organizations or accredited inspection laboratories and the results of inspection are entered in a single registry. This approach allows for continuous monitoring of medical devices, assessment of their performance, and identification of potential risks and their prevention [29]. In Uzbekistan, the demand for healthcare increased dramatically in the last years because of the rising prevalence of preventable, non-communicable diseases and the suboptimal use of healthcare resources [31]. Along with it, the field of legal metrology for medical devices advanced as well. For example, e-metrology online platform has been created and it has been used since the beginning of 2021. The two main organizations taking responsibility for medical device performance inspection according to metrological characteristics of measurands are the Uzbek National Institute of Metrology (UzNIM) [32] and the State Center of Expertise and Standardization of Medicines, Medical Devices, and Medical Equipment. (SCESMMDME) of The Ministry of Health. The Institute of Metrology [33] has the following functions: (a) implementation of a unified state policy in the field of metrology, ensuring the uniformity and reliability of measurements in assigned regions and by types of measurements; (b) storage and maintenance at an appropriate level of high-precision initial and exemplary measuring instruments; (c) development of the national metrological service;

Implementation of Legal Metrology Framework for Medical Devices …

77

(d) improvement of existing methods and means of measurement of the highest accuracy, standardization of methods and means of measurement, control and testing; (e) organization of work on state tests in order to approve the type of measuring instruments; (f) organization of work on metrological certification, verification, and calibration of measuring instruments [32]. The main objectives of SCESMMDME include the following activities: (a) organization and implementation of quality control by the state; (b) registration and permitting the use of local and foreign drugs, medical measurement devices, and medical equipment; (c) coordination of activities and management of institutions and organizations engaged in the examination, standardization, certification of pharmaceutical and medical products; (d) metrological control of medical measurement devices; (e) conducting examinations, laboratory and clinical trials, as well as approval of regulatory documents [33]. In cooperation with these two organizations, a list of medical devices with measuring function has been developed and registered by the Ministry of Justice of the Republic of Uzbekistan on August 22, 2017, under No. 2916 [34]. This decree was developed based on the learning experience of Moldova [35], Russia [36], Belarus [37], Ukraine [38], Kirgizia [39], and Kazakhstan [40] and comparison between their laws and decrees related to list of measuring and testing devices that are subject to metrological control. According to the decree, a total of 32 measuring medical devices and 5 testing medical devices are subject to metrological control. For the 37 medical devices included in the legal metrology framework, there are 11 different measurement and testing functions. For the purpose of this study, the devices that share the same measurement or testing function are grouped and presented as a single device, and those are: Electroencephalograph (EEG), Electrocardiograph (ECG, including holter, cardio monitor), Pulse oximeter, Sphygmomanometer, Electromyograph, Exoencephalograph, Rheographs (including Rheoanalyzers), Ultrasound diagnostic device, hematological analyzer, biochemical analyzer, and physiotherapy equipment (including low frequency therapy devices, ultra high frequency therapy devices, ultrasound therapy devices). After the introduction of these devices into the legal metrology of Uzbekistan, there was a firm foundation for carrying out metrological control once a year by two responsible organizations based on decree No2916 [34]. As there is a significant discrepancy, and lack of overall harmonization of procedures for post-market surveillance of medical devices, it is of high importance to develop traceable methods that show significant results in decreasing the rate of faulty medical devices. Such method was developed by the UzNIM and SCESMMDME and the results of this method, along with the impact on the state of the medical device performance in healthcare institutions in Uzbekistan is presented in this study.

78

V. Nishonov et al.

2 Methods and Materials According to the Uzbek legal metrology framework, medical device performance inspections are conducted annually. In this study, data was collected on the territory of Uzbekistan during the period of January 1, 2016 to December 31, 2021. The methodology according to which the devices were inspected, along with the permissible deviations is shown in Table 1. Table 1. Description of monitored parameters, maximum allowed error limits and reference guidelines. Medical device under test

Parameters that are monitored

Maximum allowed error limit

Guidelines

Electroencephalograph (EEG)

Relative error of calibration error (amplitude calibrator and time stamp calibrator) signal

±2%

O‘zDSt 8.089:2020 State system for ensuring the uniformity of measurements of the Republic of Uzbekistan electroencephalographs, electroencephaloscopes and electroencephalo-analyzer Verification procedure

Measurement range and 7–15% relative error of amplitude and time parameters of EEG signals

Pulse oximeter

Level of internal noise brought to the entrance

2%

Relative error of signal spectral composition

1,5–4 µV Frequency-≤10%, Amplitude- ≤15%

Voltage measurement error

±10% (±15%)

Time measurement error

±10%

Internal noise voltage applied to the input

≤25 µV

Amplitude-frequency description (AChT-AChX) unevenness

From—10% to 5%

Heart rate range and measurement error

Based on user manual

Pressure measurement error

Based on user manual

O‘z DSt 8.091:2020 State system for ensuring the uniformity of measurements of the Republic of Uzbekistan. Patient monitors. Methods and means of verification

(continued)

Implementation of Legal Metrology Framework for Medical Devices …

79

Table 1. (continued) Medical device under test

Parameters that are monitored

Maximum allowed error limit

Guidelines

SpO2 oxygen saturation Based on user measurement error manual

Sphygmomanometer

Temperature measurement error

Based on user manual

Hermeticity of the pneumo-system of pressure gauges

From O‘zDSt 8.090:2020 State 2 mmHg/impulse to system for ensuring the 3 mmHg/impulse uniformity of More than 20 M measurements of the Republic of Uzbekistan. Arterial pressure meters – non-invasive mechanical, semi-automatic and automatic. Methods and ± 3 mmHg means of verification

Electrical resistance of the insulation Electrical strength of the insulation Pressure measurement error

Electrocardiograph Measurement range and Based on standard (ECG, including holter, relative error of forms of signal and cardio monitor) amplitude and time ± 3% parameters of signals

Voltage measurement error

± 10% (± 15%)

Time interval measurement error

± 10%

Calibration voltage error

± 5%

Internal noise voltage applied to the input

Less than 25 µV

Shifting of signals between channels

Less than 1 mm

O‘zDSt 8.086:2019 State system for ensuring the uniformity of measurements of the Republic of Uzbekistan. Electrocardiographers, electrocardioscopes and electrocardio analyzers. Methods and means of verification

Range of input voltages Based on user manual Amplitude-frequency description (AChT-AChX) unevenness

Based on user manual

Time constant

Based on user manual

(continued)

80

V. Nishonov et al. Table 1. (continued)

Medical device under test

Parameters that are monitored

Maximum allowed error limit

Guidelines

Heart rate range and measurement error

± 5%

ST segment level measurement error

± 25 µV

Electromyograph

Bioelectric activity of muscle and nerve structures

Based on user manual

QU 15.92:2018 Methods and means of verification of electromyograph

Exoencephalograph

Functional parameters Based on user based on the geometric manual dimensions of the examined internal anatomical structures of the patient’s body organs and tissues, the size of the fetus, the cardiovascular system, the speed of blood flow in the vessels, the parameters of the structure of the brain and the principle of ultrasound examination

QU 15.92:2018 Methods and means of exoencephalograph

Rheographs (including Rheoanalyzers)

Muscle impedance, electrical resistance of the blood system and tissues

Based on user manual

QU 15–350:2016 Methods and means of rheographs (including rheoanalyzers)

Ultrasound diagnostic device

Resting zone

±10 mm (for all sensors), ± 3 mm (for linier sensor)

O‘z DSt 8.085:2019 State system for ensuring the uniformity of measurements of the Republic of Uzbekistan. Medical ultrasonic diagnostic echo pulse scanning systems with doppler function. Methods and means of verification

Measurements in vertical direction, distance measurement

± 1 mm

Measurements in horizontal direction, distance measurement

± 2 mm

(continued)

Implementation of Legal Metrology Framework for Medical Devices …

81

Table 1. (continued) Medical device under test

Parameters that are monitored

Maximum allowed error limit

Axial and lateral solutions

At least 4 targets from each group of phantom targets should be clearly imaged without distortion

Guidelines

Width of the focal zone At least 8 targets should be clearly imaged without distortion Sensitivity (maximum depth of ultrasound penetration)

Based on user manual

Gray scale and dynamic – range to be reflected Hematological analyzer Relative error of Based on user Hematological analyzer manual (Blood test, hematological analysis (analysis) parameters)

GOCT 8.627–2013 Interstate Standart. Hematological analyzer. Means of verification

Biochemical analyzer

O‘z DSt 8.045:2015 State system for ensuring the uniformity of Measurements of the Republic of Uzbekistan. Bio-chemical analyzers methods and means of verification

Physiotherapy equipment (including low frequency therapy devices, ultra high frequency therapy devices, ultrasound therapy devices)

Relative error of biochemical analyzers

Based on user manual

Mean square error of optical density measurement results

Based on user manual

Output parameters of low-frequency therapeutic devices

Based on user manual

QU 15.93:2018 Methods and means of verification of physiotherapy equipment

Table 1 shows that methods which are used for the verification of medical devices in Uzbekistan fits at some points with the methods which are used for the verification of medical devices in Bosnia and Herzegovina [41–49]. The data from 2016 to 2021 was collected manually based on the reports of UzNIM and SCESMMDME while the data for 2021 was retrieved from the e-metrology platform.

82

V. Nishonov et al.

In order to analyze the results of implementation of the legal metrology framework, a total number of medical devices was identified and all reports corresponding to those devices were reviewed. Table 2 presents the data regarding inspected and non-inspected medical devices in healthcare institutions. The inspected medical devices were further classified into two categories, accurate (A) where measured output of the device is within maximum permissible error for all measurement points and faulty (F) where measured output of the device is outside the permissible error limit for the inspected device. Data envelopment analysis (DEA) is a helpful nonparametric method in operations research for performance evaluation by measuring the efficiency scores of the decisionmaking units (DMUs). The fundamentals of DEA are based on a nonparametric approach that solves the problem of determining the efficiency of various DMUs according to how inputs are converted into outputs [42]. Two inputs and one output were chosen: X1—percentage of medical devices that were carried out metrological control in 2016. X2—percentage of medical devices that were carried out metrological control in 2021. Y1—percentage of increase of accurate medical devices, where applicable.

3 Results and Discussion When comparing the ratios of inspected and non-inspected medical devices in the first year of implementation of annual performance inspections and 2021, the results indicate a substantial improvement and a 71% increase in the percentage of inspected devices accompanied with the same reduction in non-inspected devices (Table 2.). Table 2. Percentage of inspected and non-inspected medical devices Year

2016 (%)

2021 (%)

Inspected medical devices

15

86

Non-inspected medical devices

85

14

The fraction of non-inspected medical devices decreased significantly when comparing 2016 and 2021. In 2016, the fraction of non-inspected medical devices decreased from 75 to 14% while the fraction of inspected devices increased accordingly from 15 to 86%. The estimated figures for 2016 and 2021 indicate that this trend will continue, and all medical devices will be covered by metrological control in the upcoming years. An important aspect to consider, both regardless and with respect to the number of inspected devices is the ratio of accurate and faulty devices. As indicated in Fig. 1, in 2016, almost 21.3% of inspected medical devices were faulty, while 78.7% of them were accurate. In 2021, 7.4% of inspected medical devices were faulty, while 92.6% of them

Implementation of Legal Metrology Framework for Medical Devices …

83

Fig. 1. Increase in the accuracy of medical devices between 2016 and 2021.

were accurate. This shows a substantial increase in the safety and performance reliability of medical devices due to the implementation of the legal metrology framework. Table 3 illustrates key findings of metrological control of medical devices after the 6 year implementation of the legal metrology framework. The most significant improvement in the number of devices subjected to metrological control was observed in electromyography devices, where an 84% point increase is observed. On the other hand, the fraction of pulse oximeter devices subjected to metrological control has experienced the lowest increase of 41%. The reason for the low percent increase observed with pulse oximeters is the fact that they have been incorporated in legal metrology in 2020 while other devices were incorporated much earlier. A factor that has to be taken into consideration is the import rate, which significantly affected the rate of metrological inspection of electromyography devices, where a significant increase in their metrological inspection rates was observed in accordance with their increased presence on the market. The fraction of accurate electromyograph, exoencephalograph, and rheograph devices inspected in 2021. Is 100%, as indicated in Table 3. In comparison with 2016, metrological inspection of these devices indicates a 21%, 18% and 19% increase in the percentage of accurate devices respectively. This emphasizes the positive effect that regular metrological control has. When observing all of the medical devices encompassed under the legal metrology framework in Uzbekistan, the percentage of accurate medical devices has increased from 69 to 93% over the course of 6 years, while the overall percentage of inspected devices increased from 15 to 86% over the same time period. To emphasize the results, the discussion focuses only on the medical devices exhibiting the highest and lowest increase in the percentage of inspected devices in the period of 6 years. The highest increase is observed in physiotherapy equipment while the lowest increase is observed in sphygmomanometers. Figure 2 shows that in 2016 faulty physiotherapy equipment had the highest rate of faulty devices and in 2021, that decreased significantly. The percentage of faulty devices was lowered by 24%.

84

V. Nishonov et al.

Table 3. Improvement in the accuracy of each inspected medical device in 2021 com- pared to 2016 Purposed medical device

The percentage of inspected medical devices in 2016, X1 (%)

The percentage of inspected medical devices in 2021, X2 (%)

Percent point, (%)

2016 X (%)

✓ (%)

2021 X (%)

✓ (%)

Percent point, Y1 (%)

EEG

23

89

76

23

77

2

98

21

ECG

17

92

75

25

75

6

94

19

Pulse oximeter

0

41

41

0

0

39

61

0

Sphygmomanometer

21

85

64

24

76

8

92

16

Electromyograph

7

91

84

21

79

0

100

21

Exoencephalograph

11

93

82

18

82

0

100

18

Rheographs

14

87

73

19

81

0

100

19

Ultrasound diagnostic device

19

90

71

25

75

9

91

16

Hematological analyzer

22

94

72

22

78

1

99

21

Biochemical analyzer 20

95

75

23

77

4

96

19

Physiotherapy equipment

15

86

71

35

65

11

89

24

Average:

15

86

71

21

69

7

93

17

Fig. 2. Improvement in the accuracy and performance of Physiotherapy equipment in 2021 compared to 2016.

Implementation of Legal Metrology Framework for Medical Devices …

85

The main factor for such kind of increase is the widespread use of physiotherapy equipment, and the high interest in their performance inspection in order to ensure high quality treatment for patients. Figure 3 shows an increase in accuracy and performance of sphygmomanometers that is the lowest one among the inspected medical devices. In general, this result is also satisfactory, because in 2016, the percentage of accurate sphygmomanometers was 76% and it went up to 92% in 2021. However, considering the high utilization of these devices in regular medical practice, the 16% increase is expected to improve in the following years.

Fig. 3. Improvement in the accuracy and performance of Sphygmomanometer in

During the analysis process of metrological control of Sphygmomanometers, the fact that many healthcare institutions still use mercury sphygmomanometers and low-quality electronic sphygmomanometers was taken into consideration. Hence, the accuracy drawbacks were expected. However, in order to ensure safety and accurate treatment of all patients, UzNIM and SCESMMDME should give more attention to strengthening the metrological control of the Sphygmomanometer and make adjustments to the verification procedure. In order to emphasize the significance of this methodology, the results achieved in Uzbekistan are compared with the results achieved in Bosnia and Herzegovina. The results of implementation of the legal metrology framework in these two countries are comparable due to the fact that the guidelines for performance evaluation have the same permissible error limits. When considering all medical devices, the accuracy was significantly higher in Bosnia and Herzegovina in the initial year, hence the implementation of legal metrology framework did not result in a large increase in the percentage of accurate devices (Fig. 4.) The percent increase in the accuracy of physiotherapy equipment due to implementation of the legal metrology framework in physiotherapy equipment is more significant in Uzbekistan in comparison with Bosnia and Herzegovina (Fig. 5.). The reason for this is the same as for all medical devices, the better initial state of the equipment in healthcare institutions.

86

V. Nishonov et al.

Fig. 4. Improvement in the accuracy and performance of medical devices in Bosnia and Herzegovina and Uzbekistan

Fig. 5. Improvement in the accuracy and performance of physiotherapy equipment in Bosnia and Herzegovina and Uzbekistan

The percent increase in the accuracy of sphygmomanometers due to implementation of the legal metrology framework in physiotherapy equipment is more significant in Uzbekistan in comparison with Bosnia and Herzegovina (Fig. 6). However, the implementation of the legal metrology framework has resulted in higher accuracy of sphygmomanometers in Uzbekistan in 2021 when compared with the accuracy of these medical devices in Bosnia and Herzegovina. The statistical analysis shows that the market coverage of medical devices is directly proportional to their higher accuracy since most prominently used medical devices are the ones whose performance is most vital for healthcare workers. Additionally, increased import rate of medical devices consequently leads to a conclusion that old and faulty medical devices are decommissioned to use new medical devices. The most crucial point is that every hospital, clinic and diagnostic center should understand decommissioning impacts to increase the accuracy and reliability of medical devices.

Implementation of Legal Metrology Framework for Medical Devices …

87

Fig. 6. Improvement in the accuracy and performance of blood pressure monitors (sphygmomanometers) in Bosnia and Herzegovina and Uzbekistan

The data collected during annual performance inspections regarding the visual state of the device, electrical safety and performance is digitized [42] to maximize its usefulness. As this data is collected in a standardized manner, it ensures traceability of measurements made by medical devices and the best usefulness of data is derived by using it for predictive modeling of device behavior to implement the best possible maintenance strategies [43]. Performance predictions for several medical devices were done using artificial intelligence based on the data collected during annual performance inspections in Bosnia and Herzegovina [44–49].

4 Conclusion Legal norms in the field of medical metrology in the Republic of Uzbekistan are aimed at protecting the rights and legitimate interests of citizens and are regulated by the Uzbek Agency for Technical Regulation. Annual metrological inspection of medical devices serves to ensure continuous quality of diagnoses and treatments given to patients and human health and well-being. The results of this study, which is the first of this kind for Uzbekistan, shows that effective implementation of the legal metrology framework for medical measurement devices has the potential to prevent possible errors in the use of medical devices. In comparison between the results of case studies which were carried out in Bosnia and Herzegovina, the results of this study suggest similar trends as the results of research carried out in Uzbekistan. Hence, it can be concluded that the proposed methodology is indeed effective. In addition to providing health benefits for all patients and healthcare workers, implementation of the legal metrology framework leads to an increase in cost-effectiveness of maintenance systems and makes planning of maintenance possible. Future endeavors will be devoted towards development of artificial intelligence based algorithms for performance prediction of medical devices thus resulting in predictive maintenance

88

V. Nishonov et al.

infrastructure that will significantly contribute to both cost-effectiveness and accuracy of diagnostic processes in healthcare. Acknowledgements. This paper and the research behind it would not have been possible without the exceptional support of Ministry of Innovative Development of the Republic of Uzbekistan, Uzbek National Institute of Metrology and “Verlab” verification laboratory. I would like to thank professors Najmiddin Muminov, Almir Badnjevic and Lejla Gurbeta for comments that greatly improved the scientific paper. I would like to show my gratitude Verlab team for sharing their pearls of wisdom with me during the my fellowship program.

References 1. Badnjevi´c, A., Pokvi´c, L.G., Džemi´c, Z., Beˇci´c, F.: Risks of emergency use authorizations for medical products during outbreak situations: a COVID-19 case study. Biomed. Eng. Online 19(1), 1–4 (2020) 2. World Health Organization.: Monitoring the building blocks of health systems: a handbook of indicators and their measurement strategies. World Health Organization (2010) 3. Bosnjakovic, A., Dzemic, Z.: Legal metrology: medical devices. In: Proceedings of the International Conference on Medical and Biological Engineering, vol. 62. Sarajevo, B&H, (2017 March 15). https://doi.org/10.1007/978-981-10-4166-2_88 4. Badnjevi´c, A., Cifrek, M., Magjarevi´c, R., Džemi´c, Z. (eds.): Inspection of medical devices. In: Series in Biomedical Engineering. Springer, Singapore (2018) 5. Badnjevi´c, A., et al.: Post-market surveillance of medical devices: a review. Technol. Health Care 30(6), 1315–1329 (2022) 6. MAUDE database: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/search.cfm 7. EUDAMED database: Retrieved from https://ec.europa.eu/tools/eudamed/eudamed 8. OáMalley, D.J.: Council Directive 90/385/EEC on the approximation of the laws of the Member States relating to active implantable medical devices. Retrieved from https://eurlex.eur opa.eu/legal-content/EN/ALL/?uri=celex:31990L0385 Accessed 20 June 1990 9. Council, E.: Council directive 93/42/EEC concerning medical devices. Retrieved from https:// eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:31993L0042Accessed 14 June 1993 10. REGULATION (EU) 2017/745 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on medical devices, amending Directive 2001/83/EC, Regulation (EC) No 178/2002 and Regulation (EC) No 1223/2009 and repealing Council Directives 90/385/EEC and 93/42/EEC. Retrieved from https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri= CELEX:32017R0745 Accessed 5 April 2017 11. REGULATION (EU) 2017/746 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on in vitro diagnostic medical devices and repealing Directive 98/79/EC and Commission Decision 2010/227/EU. Retrieved from https://eur-lex.europa.eu/legal-content/ EN/TXT/PDF/?uri=CELEX:32017R0746 Accessed 5 April 2017 12. World Health Organization: https://www.who.int/health-topics/medical-devices 13. Australian regulatory guidelines for medical devices (ARGMD), 2011 May, Version 1.1 14. Zhang, W., Liu, R., Chatwin, C.: Marketing authorization of medical devices in China. J. Commer. Biotechnol. (2016). https://doi.org/10.5912/jcb720 15. Konishi, A., Isobe, S., Sato, D.: New regulatory framework for medical devices in Japan: current regulatory considerations regarding clinical studies. J. Vasc. Intervent. Radiol. 29(5), 657–660 (2018 May 01). https://doi.org/10.1016/j.jvir.2017.12.022

Implementation of Legal Metrology Framework for Medical Devices …

89

16. Gurbeta, L., Alic, B., Dzemic, Z., Badnjevic, A.: Testing of infusion pumps in healthcare institutions in Bosnia and Herzegovina. In EMBEC & NBC 2017, pp. 390–393. Springer, Singapore (2017) 17. Gurbeta, L., Alic, B., Dzemic, Z., Badnjevic, A.: Testing of dialysis machines in healthcare institutions in Bosnia and Herzegovina. In: EMBEC & NBC, 2017, pp. 470–473. Springer, Singapore 18. Gurbeta, L., Badnjevic, A., Dzemic, Z., Jimenez, E.R., Jakupovic, A.: Testing of therapeutic ultrasound in healthcare institutions in Bosnia and Herzegovina. In: 2nd EAI International Conference on Future Access Enablers of Ubiquitous and Intelligent Infrastructures, pp. 24– 25. 2016 19. Badnjevic, A., Gurbeta, L., Jimenez, E.R., Iadanza, E.: Testing of mechanical ventilators and infant incubators in healthcare institutions. Technol. Health Care 25(2), 237–250 (2017) 20. Gurbeta, L., Dzemic, Z., Bego, T., Sejdic, E., Badnjevic, A.: Testing of anesthesia machines and defibrillators in healthcare institutions. J. Med. Syst. 41(9), 1–10 (2017) 21. Gurbeta, L., Badnjevi´c, A.: Inspection process of medical devices in healthcare institutions: software solution. Heal. Technol. 7(1), 109–117 (2017) 22. Gurbeta, L., Badnjevi´c, A., Žuni´c, E., Pinjo, N., Ljumi´c, F.: Software package for tracking status of inspection dates and reports of medical devices in healthcare institutions of Bosnia and Herzegovina. In: 2015 XXV International Conference on Information, Communication and Automation Technologies (ICAT) 2015 Oct 29, pp. 1–5. IEEE 23. Pires, C., Duarte, D., Cavaco, A.: Analysis of medical device alerts issued by the Portuguese medicines agency: scoping the purpose of new regulatory recommendations. Acta Med. Port. 34(3), 201–208 (2021) 24. Ivanovska, E., Toni´c-Ribarska, J., Lazova, J., Popstefanova, N., Davcheva-Jovanoska, M., Trajkovi´c-Jolevska, S.: Obezbedivanje kliniˇckih dokaza u skladu sa MDR 2017/745-novi izazovi za proizvodaˇce u industriji medicinskih sredstava. Arhiv za farmaciju. 69(1), 39–49 (2019) 25. Al-Surimi, K., Househ, M., Almohandis, E., Alshagathrh, F.: Establishing a national medical device registry in Saudi Arabia: lessons learned and future work. In: Enabling Health Informatics Applications 2015, pp. 23–26. IOS Press 26. The National Institute of standards and technology: http://www.nist.gov/medical-devices-por tal.cfm 27. France’s National Metrology Institute.: http://www.lne.eu/en/markets/medical-health.asp 28. National Metrology Institute of Germany: http://www.ptb.de/cms/en/fachabteilungen/abt8 29. Karaboci, B., et al.: Medical metrology studies at Tübitak UME. In: 17 International Congress of Metrology, Paris, France (2015) 30. Graham, R.D.: Jonesa, Craig Jackson, the joint committee for traceability in laboratory medicine (JCTLM) –its history and operation. Clin. Chim. Acta 453, 86–94 (2016) 31. Ministry of Health of the Republic of Uzbekistan https://ssv.uz/uz/open_ministry/category/ ochi-malumotlar 32. Uzbek National Institute of Metrology: https://nim.uz/en/about-us/general-information 33. State center for expertise and standardization of medicines, medical devices and medical equipment. https://uzpharm-control.uz/en/ 34. Decree of the Uzbek Agency for Standardization, Metrology and Certification and the Ministry of Health of the Republic of Uzbekistan No 2916 On approval of the list of measuring instruments and test instruments intended for medicine to be subject to metrological examination. (2017) 35. Decree of the Government of the Republic of Moldova No 1042 On approval of the official list of measuring instruments and measurements subject to legal metrological control. (2016)

90

V. Nishonov et al.

36. Decree of the Government of the Russian Federation No 250 On the list of measuring instruments, the verification of which is carried out only by state regional metrology centers accredited in the prescribed manner in the field of ensuring the uniformity of measurements. (2010) 37. Decree of the Ministry of Health of the Republic of Belarus No 255 On approval and enforcement of the Regulations on metrological support of healthcare institutions of the Republic of Belarus. (1996) 38. Decree of the Cabinet Ministry of the Republic of Ukraine No 374 “On approval of the list of categories of measuring equipment subject to periodic verification. (2015) 39. Decree of the Government of the Republic of Kyrgyz No 664 On measures to implement the Law of the Kyrgyz Republic “On ensuring the uniformity of measurements. (2012) 40. Decree of the Ministry of Health of the Republic of Kazakhstan No 765 On approval of the List of medical equipment, which is a means of measurement. (2009) 41. Badnjevi´c, A., et al.: A novel method for conformity assessment testing of therapeutic ultrasounds for post-market surveillance purposes. Technol. Health Care 31(1):339–346 (2023) 42. Wong, W.-P.: A global search method for inputs and outputs in data envelopment analysis: procedures and managerial perspectives. Symmetry 13, 1155 (2021). https://doi.org/10.3390/ sym13071155 43. Gurbeta, L., Badnjevi´c, A., Kurta, E.: eVerlab: Software tool for medical device safety and performance inspection management. In: Badnjevic, A., Škrbi´c, R., Gurbeta Pokvi´c, L. (eds) CMBEBIH 2019. CMBEBIH 2019. IFMBE Proceedings, vol 73. Springer, Cham (2020) 44. Badnjevic, A.: Evidence-based maintenance of medical devices: current shortage and pathway towards solution. Technol. Health Care 31(1), 293–305 (2023) 45. Hrvat, F., Spahic, L., Gurbeta Pokvic, L., Badnjevic, A.: Artificial neural networks for prediction of medical device performance based on conformity assessment data: Infusion and perfusor pumps case study. In: IEEE 9th Mediterranean Conference on Embedded Computing (MECO), 08–11 June 2020, Budva, Montenegro (2020) 46. Spahi´c, L., Kurta, E., Cordic, S., Becirovic, M., Gurbeta, L., Kovacevic, Z., Izetbegovic, S., Badnjevic, A.: Machine learning techniques for performance prediction of medical devices: infant incubators. In: Badnjevic, A., Škrbi´c, R., Gurbeta Pokvi´c, L. (eds) CMBEBIH 2019. CMBEBIH 2019. IFMBE Proceedings, vol. 73. Springer, Cham (2020) 47. Gurbeta, L., Džemic, Z., Badnjevic, A.: establishing traceability chain of infusion and perfusor pumps using legal metrology procedures in Bosnia and Herzegovina. In: Lhotska, L., Sukupova, L., Lackovi´c, I., Ibbott, G. (eds.) World Congress on Medical Physics and Biomedical Engineering 2018. IFMBE Proceedings, vol 68/2. Springer, Singapore (2018) 48. Gurbeta, L., Vukovic, D., Džemic, Z., Badnjevic, A.: Legal metrology procedures for increasing safety and performance characteristics with cost benefits analysis: case study dialysis machines. In: Lhotska, L., Sukupova, L., Lackovi´c, I., Ibbott, G. (eds) World Congress on Medical Physics and Biomedical Engineering 2018. IFMBE Proceedings, vol 68/2. Springer, Singapore (2018) 49. Badnjevi´c, A., Pokvi´c, L.G., Hasiˇci´c, M., Bandi´c, L., Mašeti´c, Z., Kovaˇcevi´c, Ž., Kevri´c, J., Pecchia, L.: Evidence-based clinical engineering: machine learning algorithms for prediction of defibrillator performance. Biomed. Signal Proc. Control 54, 101629 (2019). ISSN 17468094. https://doi.org/10.1016/j.bspc.2019.101629

Performance Test and Calibration in End Tidal Carbon Dioxide Measurements Mana Sezdi1(B)

and Nazif ˙Ilker Sezdi2

1 Istanbul University-Cerrahpa¸sa, Buyukcekmece Kampus, Istanbul, Turkey

[email protected] 2 Medibim Medikal Kalibrasyon Ltd. Sti, ¸ Gokdemir Plaza, Esenyurt, Istanbul, Turkey

Abstract. End-tidal CO2 (EtCO2 ) is the partial pressure of carbon dioxide in the breath at the end of expiration. The measurement of the partial pressure of carbon dioxide from the airway during respiration is called capnography, and the device that makes the measurement is called capnometry. During measurement, the capnometry is placed between the tracheal tube and the balloon or mechanical ventilator in the intubated patient. Capnography is vital in the evaluation of respiration, ventilation and circulation. Normal EtCO2 value in expiration is (30–45) mmHg. Values between (25–30) mmHg indicate hyperventilation, and EtCO2 above 45 mmHg indicates hypo-ventilation. It is important that these measurement values, which determine the way of intervention to the patient, are measured correctly. This vitally important device, like all other medical devices, should be checked for correct measurement and should be subjected to measurements within the scope of medical metrology. Although there are separate standards and procedures for each medical device to be used in test, control and calibration activities of medical devices, there is still no standardized measurement method for capnometries. In this paper, an improved method is proposed to test the accuracy of End-tidal CO2 measurements in capnometries. Testing and control of a capnometry were also carried out using the mentioned method. The applicability of the measurement method has passed the accreditation audit of the Turkish Accreditation Agency (TÜRKAK) and is within the scope of accreditation according to the TS EN ISO/IEC 17025:2017 standard. Keywords: EtCO2 · Capnometry · Performance Test · Calibration

1 Introduction 1.1 End Tidal CO2 End tidal carbon dioxide (EtCO2 ) is the partial pressure of carbon dioxide in the air exhaled during expiration. The absorption of carbon dioxide at a specific wavelength (4.26 micrometers) is measured using infrared radiation through capnometry or capnography. It was first used by anesthesiologists in the 1950s to confirm the location of the endotracheal tube after intubation. The advantages of EtCO2 measurement, such as being an easy-to-apply, showing the ventilation status simultaneously with each breath, and allowing to monitor the change over time, have increased in use [1]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 91–97, 2024. https://doi.org/10.1007/978-3-031-49068-2_10

92

M. Sezdi and N. ˙I. Sezdi

The graphic obtained from capnography is known as capnogram. People with normal lung function characteristically have the trapezoidal capnogram plot seen in Fig. 1. The EtCO2 value starts at 0 at the end of inspiration and rises with the start of expiration, reaches (35–45) mmHg at the end of expiration, and returns to 0 with the start of inspiration. The amplitude in the graph depends on the EtCO2 concentration, and the width of the graph is directly proportional to the expiration time.

Fig. 1. The graphic of end tidal CO2 [2].

1.2 End Tidal CO2 Monitoring The end tidal carbon dioxide (EtCO2 ) monitor can be qualitative or quantitative. Qualitative devices give rough information about the EtCO2 level. These devices are calorimetric and use the principle of changing the color of litmus paper according to the EtCO2 level. Although there is a slight difference in the threshold values according to the brand of the device used, the EtCO2 measured in these devices roughly turns purple if it is low, light brown if it is medium, and yellow if it is high. Today, these devices have left their place to devices that make more quantitative measurements. Quantitative devices measure the EtCO2 level and give specific results in milimeters of mercury [1]. In this measurements, infrared (IR) spectroscopy is utilized. In this way, it is measured the CO2 molecules absorption of IR light as the light passes through a gas sample. While only the value is measured in capnometric devices, the waveform showing the change in EtCO2 value over time is also monitored in capnographic devices. Capnography, the graphic display of the exhaled and inhaled carbon dioxide concentration plotted against time, is used to monitor ventilation [3] (See Fig. 2]. Types of End Tidal CO2 Monitoring. There are two types of EtCO2 monitoring. These are; mainstream CO2 monitoring and sidestream CO2 monitoring. Mainstream CO2 Monitoring. Mainstream CO2 monitoring devices utilize a small infrared sensor consisting of a sample cell or cuvette and an infrared optical bench placed within the gas flow pathway at the airway. This measurement location results in real-time CO2 values within the airway and a real-time graphical representation of the CO2 waveform plotted over time or by the exhaled volume [4].

Performance Test and Calibration in End Tidal Carbon Dioxide Measurements

93

Fig. 2. Quantitative measuring device (left), capnometry (middle), capnograph (right) [1]

Sidestream CO2 Monitoring. Sidestream CO2 monitoring devices aspirate a gas sample from a ventilator breathing circuit or other patient interface device through a length of small-bore tubing. Different interface designs for use on non-intubated patients incorporate nasal and oral sampling points to improve measurement accuracy. Several interface designs include the ability to simultaneously administer oxygen through a split channel device. Sidestream sampling devices utilize an infrared CO2 sensor in a monitor located away from the patient and can only be displayed in a time- based waveform. Time-based sidestream CO2 monitoring is the type more commonly used in the operating room setting, during non-intubated patient monitoring, and during CPR [4]. 1.3 The Performance Test of End Tidal CO2 Monitoring Capnography is vital in the evaluation of respiration, ventilation and circulation. Normal EtCO2 value in expiration is (30–45) mmHg. Values between (25–30) mmHg indicate hyperventilation, and EtCO2 above 45 mmHg indicates hypoventilation. It is important that these measurement values, which determine the way of intervention to the patient, are measured correctly. Athmospheric pressure is also another affect to measurements and should be measured and recorded during tests. This vital sign monitoring device, like any other medical devices, must be checked for accurate measurement and subjected to measurements within the scope of medical metrology. Although there are separate standards and procedures for each medical device to be used in testing, control and calibration activities of medical devices, there is still no standardized measurement method for capnometries. Recently, Et CO2 simulators have been developed for testing, control and calibration of the EtCO2 monitor. The simulators are designed to generate a gas flow connected to capnometries or patient monitors with capnography functionality. This patient’s breath simulation allows for scheduled preventive maintenance or functional testing for both mainstream and sidestream capnography monitors [5]. However, it has not yet been standardized how the EtCO2 monitor will be tested using the simulator and the procedure according to which measurements will be made. In this paper, an improved method is proposed to test the accuracy of EtCO2 measurements on capnometer devices.

94

M. Sezdi and N. ˙I. Sezdi

2 Method In the performans test of capnography device, Datrend vPad- CO2 EtCO2 simulator was used as a test device. Datrend vPad- CO2 is designed to perform CO2 gas calibration, BrPM rate and flow tests of capnography devices. vPad- CO2 is a simulator that can simulate in the range of (100–150) BrPM. It can perform simulation with an accuracy of 0.1% in the range of (2–100) BrPM, and with an accuracy of 0.1 in the range of (101–150) BrPM [5]. As it is seen in Fig. 3, CO2 gas cylinder supplied for health usage is used as a reference gas and connected to capnography device.

CO2 vPadCapnograph

CO2 simulator

Fig. 3. The measurement set-up [5].

There are 2 steps of test: Breath simulation and CO2 measurement. 2.1 Breath Simulation Breath simulation is based on to test number of breath per minute. Vpad- CO2 simulates the breathing by alternating between higher concentration of CO2 from the CO2 source and the air from the room for each breath. The duty cycle sets how much of breath is CO2 and how much of breath is air as percentage. The full range is 20 + 80% changing valued of Air- CO2 mixing values mutually. There should be also measurement test timing and it is adjusting as minute values. vPad- CO2 averages the values measured. It is adjusted with control screen of vPadCO2 BrPM simulations as given images below (see Fig. 4). 2.2 CO2 Measurement Test Health purposed CO2 gas is suppling with cylinders industrially. General and typical percentage is 5% CO2 . Specified percentage value of CO2 cylinder and atmospheric pressure should be adjusted at vPad- CO2 (see Fig. 5). The results can be display and evaluate with the View Results screen (see Fig. 6).

3 Results The CO2 % and BrPM measurement results of a capnograph tested using the Datrend vPad-CO2 EtCO2 simulator are presented in following Tables 1 and 2. In the CO2 % measurements, a 5% CO2 gas cylinder was used and the measurements were repeated

Performance Test and Calibration in End Tidal Carbon Dioxide Measurements

95

Fig. 4. The control screen of vPad- CO2 BrPM [5].

Fig. 5. The screen for the adjustment of percentage value of CO2 cylinder and atmospheric pressure [5].

5 times. The average of these 5 measurements was compared with the tolerance value determined by the manufacturer and the Pass or Fail status of the device was decided. For BrPM measurements, 5 different BrPM values set on the simulator were tried to be seen on the capnograph screen. After measurements, the measurement uncertainties were calculated.

96

M. Sezdi and N. ˙I. Sezdi

Fig. 6. The view results screen [5].

Table 1. The measured values for CO2 %. CO2 (%)

Measured (%)

Average

Tolerance

Uncertainty

Pass/Fail

5

4.96

5.01%

0.08%

0.0073%

Pass

5

4.97

5

5.04

5

5.06

5

5.02

Table 2. The measured values for BrPM. BrPM

Measured

Tolerance

Error

Uncertainty

Pass/Fail

10

10

±2

0

0.1077

Pass

30

30

±2

0

0.2084

Pass

60

60

±2

0

0.5218

Pass

120

120

±2

0

0.9173

Pass

150

149

±2

1

1.2333

Pass

4 Conclusion In this study, a preliminary study was carried out for EtCO2 measurements, which have not yet had an international determinant standard. Although EtCO2 is vital in the evaluation of respiration, ventilation and circulation, there is no standard calibration procedure

Performance Test and Calibration in End Tidal Carbon Dioxide Measurements

97

for capnograph measuring EtCO2 . Like any other medical device, capnographs need to be checked in the test, control and calibration activities. To generate a base for these test/control activities, using Datrend VPAD-CO2 EtCO2 simulator, both CO2 and BRPM measurements were performed on a capnography. Since there is no standard yet, the measurement results were evaluated according to the tolerance-acceptance criteria of the device manufacturer. The measurement results were found to be acceptable with measurement uncertainties. The measurement method was presented to the Turkish Accreditation Institution (TÜRKAK) and the approval of applicability was obtained. Our next goal is to take the measurement method into the scope of medical metrology and to expand the study.

References 1. Kudu, E.: Endtidal karbondioksit (EtCO2 ). Retrieved from https://acilci.net/endtidal-karbon dioksit/ Accessed 26 Feb 2023 2. Erlich, C., Lamer, A., Moussa, M.D.: End-tidal carbon dioxide for diagnosing anaphylaxis in patients with severe postinduction hypotension. Anesthesiology 136, 472–481 (2022) 3. Ortega, R., Connor, C., Kim, S.: Monitoring ventilation with capnography. N Engl J Med 367(19) (2012) 4. Siobal, M.S.: Monitoring exhaled carbon dioxide. Respir. Care 61(10), 1397–1416 (2016) 5. Datrend vPad-CO2 EtCO2 simulator user manual

A New and Fast Approach for Antimicrobial Resistance Detection: Combination of Artificial Intelligence and Surface-Enhanced Raman Spectra Omer Aydin1,2,3,4,4(B) , Zakarya Al-Shaebi1,2 , Munevver Akdeniz1,2 , Gizem Kursunluoglu2 , Gokmen Zarasız5,6 , Serra ˙Ilayda Yerlitas5,6 , Ahmet Sezgin5,6 , Mustafa Altay Atalay7 , and Pınar Sagiroglu7 1 Department of Biomedical Engineering, Erciyes University, 38039 Kayseri, Turkey

[email protected], [email protected]

2 Nanothera Lab, Drug Application and Research Center (ERFARMA), Erciyes University,

38039 Kayseri, Turkey 3 Clinical Engineering Research and Implementation Center, (ERKAM), Erciyes University,

38030 Kayseri, Turkey 4 Nanotechnology Research and Application Center (ERNAM), Erciyes University, 38039

Kayseri, Turkey 5 Department of Biostatistics, Erciyes University School of Medicine, 38039 Kayseri, Turkey 6 Drug Application and Research Center (ERFARMA), Erciyes University, Kayseri, Turkey 7 Department of Medical Microbiology, Faculty of Medicine, Erciyes Üniversitesi, 38039

Kayseri, Turkey

Abstract. Antimicrobial resistance (AMR) in bacteria is a global health crisis due to the rapid emergence of multidrug-resistant bacteria and the lengthy development of new antimicrobials. Surface-enhanced Raman scattering (SERS) is a powerful technique for sensitive label-free analysis of chemical and biological samples, which can be used for rapid detection and identification of bacterial strains. However, distinguishing the antibiotic-resistant and susceptible bacteria by SERS spectra is challenging due to the high molecular similarity of the bacterial strains. To overcome this challenge, we proposed to use artificial intelligence (AI) methods to assist SERS-based diagnostics of AMR bacteria. We used machine learning to optimize the sampling of SERS substrates, improving the data collection efficiency and reliability. We also used deep learning to analyze the SERS spectra of bacteria. Our AI-assisted SERS strategy enables label-free spectroscopic profiling of AMR bacteria in complex clinical settings, offering a promising solution for combating the AMR threat. Keywords: Antimicrobial resistance · Surface-enhanced raman spectroscopy · Machine learnings · Methicillin-resistant S. aureus (MRSA)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 98–103, 2024. https://doi.org/10.1007/978-3-031-49068-2_11

A New and Fast Approach for Antimicrobial Resistance Detection

99

1 Introduction We have all globally experienced the COVID-19 pandemic in the past 3 years. But how many do we know about antibiotic resistance that is knocking at our door as a silent pandemic, and how many are we aware of this impending pandemic? As defined by international institutions such as the World Health Organization (WHO); AMR is cited as one of the most important issues to health, food security, and development worldwide [1]. Although AMR occurs naturally, the wrong and unnecessary use of antibiotics on humans and animals accelerates this process. All over the world, bacterial diseases such as pneumonia, tuberculosis, gonorrhea, Staphylococcus aureus-related resistance and salmonella are becoming increasingly difficult to treat due to the diminishing effects of antibiotics. In addition, basic antibiotics are 15% less effective against common pathogens. On the other hand, despite the decrease in drug costs of antibiotics, the tests required to determine which bacteria they originate from are more expensive than these antibiotics, which makes the use of wrong antibiotics more widespread [2]. The majority of currently in use techniques are polymerase chain reaction (PCR) or antibody-based techniques like enzyme-linked immunosorbent assays (ELISA), which have some drawbacks, like high-cost procedures, time-consuming processes, and requiring trained personnel [3, 4]. Thus, rapid and reliable methods for detecting resistance and susceptibility to these infections are therefore essential for effective treatment and prevention. Surface-enhanced Raman spectroscopy (SERS) is a vibrational technique based on inelastic scattering of laser light, which comes from noble metal nanostructures [5–7]. This technique can be used in a label-free manner. Thus, it is capable of directly identifying life-threatening bacteria. SERS can be used not only for the identification of different bacteria but also for the discrimination of antibiotic-resistant and susceptible strains of bacteria [8]. However, the SERS spectra of the antibiotic-resistant and susceptible bacteria show only subtle differences due to the high molecular similarity of the bacterial strains. Machine learning techniques are indispensable to detect minor spectral differences and can discriminate bacteria at the strain level [9]. These techniques can successfully reveal subtle differences which are extremely difficult to distinguish with the naked eye. In this study, we used SERS to detect methicillin-resistant and susceptible S. aureus (MRSA and MSSA, respectively), and applied a range of machine learning algorithms, including deep learning and traditional approaches, to distinguish between the two bacterial species (Scheme 1). Among all classifiers used, the U-Net deep learning model performed best with an average accuracy of 99.04 ± 0.003%. The Random Forest traditional machine learning algorithm also performed well, with an average accuracy of 97.8 ± 0.371%.

2 Methods In this research, we analyzed a set of 22 bacterial samples, 19 of which were MRSA and the remaining 3 were MSSA. We grew these bacteria on agar plates at 37 degrees Celsius for 24 h. Following that, using sterile inoculating loops, we collected the bacterial

100

O. Aydin et al.

Scheme 1. Schematic illustration of detection of MSSA and MRSA with SERS and artificial intelligence

specimens. After washing, we mixed these specimens with a 4X concentration of silver nanoparticles (AgNPs). This prepared sample was then placed on a CaF2 slide for SERS analysis. A Raman microscopy system with a 785 nm diode laser as the excitation source was used to capture the SERS spectra. We focused on the 550–1650 cm−1 Raman shift range and gathered a total of 33,975 spectra. Among these, MRSA samples accounted for 29,475 spectra, and the remaining 4,500 belonged to MSSA. Our goal was to classify these bacteria based on their antibiotic resistance or susceptibility. For this, we utilized a variety of classification methods, including deep learning models like U-Net and VGG-16, as well as conventional machine learning algorithms such as Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbor (kNN), and naïve Bayes (NB). We employed a 10-fold cross-validation strategy to prevent overfitting. For training the deep learning models, we used the Adam optimizer. The learning rates for the U-Net and VGG-16 models were set at 0.000001 and 0.000005 respectively. All the classification algorithms were run in Python using TensorFlow and Keras, with computations performed on a Google Colab GPU.

3 Results and Discussion 3.1 AgNPs Characterization and SERS Spectra The silver nanoparticles were characterized using UV/Vis spectrophotometry for absorbance, Zetasizer for hydrodynamic size distribution and surface charge (zeta potential). The results show maximum absorption at 418 nm, a size of approximately 60 nm, and a zeta potential of -41 mV. The particles interact with the bacterium cell wall, which contains lipids, polysaccharides, phosphate, and nucleic acids. SERS spectra for MRSA and MSSA were also obtained, which showed similar spectral features with slight differences in peak intensity as seen Fig. 1. Specific peaks in the range of 550–1650 cm−1 , such as 658, 732, 1333, 1450, and 1576 cm−1 , were identified as fingerprints for bacteria cell wall components.

A New and Fast Approach for Antimicrobial Resistance Detection

101

0,6

MRSA MSSA

Raman Intensity

0,5 0,4 0,3 0,2 0,1 600

750

900

1050

1200

Raman shift (cm-1)

1350

1500

1650

Fig.1. SERS spectra of MRSA and MSSA

3.2 Deep Learning Models’ Performance The U-Net architecture was selected for its ability to extract features from both the encoder and decoder and has shown good results in antibiotic resistance and susceptibility classification [10]. After training using 10-fold cross-validation on the data, the UNet model achieved an average accuracy of 99.04 ± 0.003% and a high precision and sensitivity as it can be seen in the confusion matrix (Fig. 2a). The specificity was not as high due to the imbalance between the two classes. The AUC of the ROC curve was 0.99, indicating excellent performance. The U-Net model was also efficient in terms of training and testing time, taking only around two hours to complete.

Fig. 2. Confusion matrix of a) U-Net b) VGG-16

The VGG-16 model was used for its well-known performance and advantages. It achieved an average accuracy of 98.86 ± 0.01% and an AUC of 0.98, see Fig. 2b. The confusion matrix showed lower misclassification in the first class compared to the UNet model, resulting in higher precision and specificity. However, the U-Net model had superiority in other statistical measurements. The training and testing time for VGG-16 was longer than the U-Net model, taking approximately 4.5 h.

102

O. Aydin et al.

3.3 Traditional Machine Learning Algorithms’ Performances The performances of traditional machine learning algorithms (SVM with three kernel functions, kNN, RF, and LR) were evaluated for classifying MRSA and MSSA. RF had the best accuracy (97.8%) and AUC (0.95), while kNN was faster and had good results. Among the three kernel functions of the SVM, Gaussian performed the best, but it took a long time. LR also had good results with an average accuracy of 94.8%. The sensitivity and precision of the models were high, but the specificity was low due to the imbalance of data. The deep learning models performed well due to the massive data of SERS for antibiotic-resistant and susceptible S. aureus. The proposed method showed good results overall and could be extended to clinical practice. Table 1 demonstrates the overall results of the deep learning and machine learning algorithms used in this study. Table 1. Comparison between deep learning models and machine learning algorithms Algorithm

Average accuracy (%)

AUC

Standard deviation

Sensitivity (%)

Specificity (%)

Precision (%)

U-Net

99.04

0.99

0.003

99.59

95.51

99.30

VGG-16

98.86

0.98

0.01

99.20

96.71

99.51

RF

97.16

0.95

0.371

98.79

87.17

97.93

SVM(Gaussian)

96.10

0.88

0.315

95.88

98.15

99.79

SVM (Linear)

95.76

0.87

0.251

95.53

96.66

99.63

kNN

95.62

0.85

0.242

95.73

96.10

99.56

LR

94.73

0.81

0.165

94.58

96.34

99.64

SVM (Poly)

93.24

0.77

0.333

93.03

95.99

99.67

4 Conclusion This study used SERS and machine learning to accurately detect antibiotic resistance and susceptibility in S. aureus. The results showed that deep learning was more effective than traditional machine learning, and the method has potential for clinical use, although further studies are needed to address data imbalance and other bacterial resistances. Acknowledgment. This work was supported by the Erciyes University Scientific Research Projects Coordination Unit under grant number: FBAÜ-2023–12265.

A New and Fast Approach for Antimicrobial Resistance Detection

103

References 1. WHO.: New report calls for urgent action to avert antimicrobial resistance crisis (2019) 2. Jayaraman, R.: Antibiotic resistance: an overview of mechanisms and a paradigm shift. Curr. Sci. 96(11), 1475–1484 (2009) 3. Anjum, M.F., Zankari, E., Hasman, H.: Molecular methods for detection of antimicrobial resistance. In: Antimicrobial Resistance in Bacteria from Livestock and Companion Animals, pp. 33–50 (2018) 4. Gajic, I., et al.: Antimicrobial susceptibility testing: a comprehensive review of currently used methods. Antibiotics 11 (2022). https://doi.org/10.3390/antibiotics11040427 5. Cialla, D., et al.: Surface-enhanced Raman spectroscopy (SERS): progress and trends. Anal. Bioanal. Chem. 403(1), 27–54 (2012) 6. Uysal Ciloglu, F., et al.: Identification of methicillin-resistant Staphylococcus aureus bacteria using surface-enhanced Raman spectroscopy and machine learning techniques. Analyst 145(23), 7559–7570 (2020) 7. Ciloglu, F.U., et al.: Drug-resistant Staphylococcus aureus bacteria detection by combining surface-enhanced Raman spectroscopy (SERS) and deep learning techniques. Sci. Rep. 11(1), 18444 (2021) 8. Ciloglu, F.U., et al.: SERS-based sensor with a machine learning based effective feature extraction technique for fast detection of colistin-resistant Klebsiella pneumoniae. Anal. Chim. Acta 1221, 340094 (2022) 9. Lussier, F., et al.: Deep learning and artificial intelligence methods for Raman and surfaceenhanced Raman scattering. TrAC Trends Anal. Chem. 124, 115796 (2020) 10. Al-Shaebi, Z., et al.: Highly accurate identification of bacteria’s antibiotic resistance based on raman spectroscopy and U-Net deep learning algorithms. ACS Omega 7(33), 29443–29451 (2022)

Metrology Versus Medical Metrology Baki Karaböce(B) TÜB˙ITAK UME, Kocaeli 41470, Türkiye [email protected]

Abstract. Metrology aims to provide the reliability and accuracy of all measurements by defining the units (SI and its derivatives), which are the basis of all measurement systems, and to offer them to the use of science and technology. Metrology specializes in three different fields: scientific, legal and industrial metrology. The main purpose of Medical Metrology is to ensure and improve the traceability of measurement quantities in the field of health, to integrate them into the international metrology system by making international comparisons, and to establish a measurement association by providing traceability to low-level laboratories established in the country (with calibration, measurement and test services in the country or abroad). Medical devices range from a simple thermometer to computerized imaging systems, clinical measurement systems or multi-parameter computer-controlled devices. Regular maintenance/repair and testing, measurement, verification and calibration should be performed at regular intervals to ensure the correct operation of medical devices throughout their use. Some instruments and apparatus used for testing, measurement, verification and calibration are commercially available. However, sometimes primary standards are not always available in medical calibrations and measurement is based on physiological models or assumptions. In this article, traceability in classical/conventional metrology and medical metrology will be examined. Keywords: Metrology · Medical metrology · Traceability

1 Introduction Metrology is the science of measuring systems and units. The International Bureau of Weights and Measures (BIPM) defines metrology in general as all studies that include theoretical and experimental measurement studies in the field of science and technology and that contain uncertainty information. Activities in the field of metrology are organized by national institutes in each country and are an activity that requires international cooperation and coordination. For example, Physikalisch-Technische Bundesanstalt (PTB) in Germany, National Physical Laboratory (NPL) in England, National Metrology Institute (UME) in Türkiye and Institute of Metrology of Bosnia and Herzegovina (IMBiH) in Bosnia and Herzegovina carry out this activity. Cooperation and coordination is ensured with regional organizations such as EURAMET in Europe, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 104–111, 2024. https://doi.org/10.1007/978-3-031-49068-2_12

Metrology Versus Medical Metrology

105

COOMET in Eurasia and SIM in America. In order to examine the activities in the field of metrology, it is more appropriate to approach on the basis of three sub-fields; scientific metrology, applied metrology and legal metrology. Legal metrology aims to legally ensure the reliability of measurements and measuring devices used in the context of consumer rights protection and fair trade. Applied metrology ensures reliability in production and other processes by ensuring the suitability, calibration and quality control of measuring instruments in the industrial field. Scientific metrology, as the basis of all fields, realizes 7 basic units and derived units, develops and disseminates methods and standards for the realization of accurate measurements. International System of Units (SI), seven basic units, length (m), mass (kg), time (s), electric current (A), thermodynamic temperature (K), amount of substance (mole) and luminous intensity (cd) and based on derived units. The SI system is a system that can evolve according to developments in science and technology. From 20 May 2019, SI units are defined in terms of universal constants. Thus, it enables the use of new technologies, including quantum technologies for SI and its derived units. It is envisaged that the entire unit system can be derived in terms of values of universal constants expressed in SI units [1]. Since the product of the numeric value and the unit must equal the value of the constant, the unit becomes defined and it is by fixing the exact numeric value of each. Their numerical values and the units they define can be seen in Table 1. Table 1. SI units and universal constants Defining constant

Description

Numerical value

Unit

νCs

Cesium ultrafine frequency

9 192 631 770

Hz

c

Speed of light in vacuum

299 792 458

m s–1

h

Planck constant

6.626 070 15 × 10–34

Js

e

Fundamental charge

1.602 176 634 × 10–19

C

k

Boltzmann constant

1.380 649 × 10–23

J K–1

NA

Avogadro’s constant

6.022 140 76 × 1023

mol–1

Kcd

Luminous efficiency of a defined visible radiation

683

lm W–1

The establishment of the infrastructure of the measurement, standardization, control and quality system in a country, provides a comprehensive system for the determination of product characteristics and quality control for quality assurance. Today’s globalizing economy relies on reliable, internationally recognized measurements and tests. The prerequisite for this is a widely used, robust metrology infrastructure. Metrology studies are important in order to contribute to the improvement of people’s quality of life and global competitiveness. The science of metrology as a technical and scientific activity is a set of activities that play a key role in all segments of society. Accurate and reliable measurement activities have big impact on scientific, commercial and social dimensions.

106

B. Karaböce

2 Classical/Conventional Metrology A basic tool in ensuring the traceability of a measurement is the calibration of a measuring instrument, measuring system or reference material. Calibration determines the performance characteristics of an instrument, system or reference material. It is usually achieved by means of a direct comparison against measurement standards or certified reference materials. A calibration certificate is issued and, in most cases, a sticker is provided for the instrument. Four main reasons for having an instrument calibrated: • • • •

To establish and demonstrate traceability. To ensure readings from the instrument are consistent with other measurements. To determine the accuracy of the instrument readings. To establish the reliability of the instrument i.e. that it can be trusted [2].

The Joint Committee on Metrology Guidelines (JCGM) of universal constants expressed in SI units defines metrological traceability as ‘the property of a measurement result to which the result can be associated with a reference through a continuous, documented chain of calibrations, each contributing to measurement uncertainty [3]. The traceability of values assigned to calibrators and/or control materials must be assured through available reference measurement procedures and/or available reference materials of a higher order [4] The traceability in measurements are established by means of measurement devices which are more accurate and more resolution devices. A device under test (DUT), micrometer is traceable to SI unit length (m) by means of working standard gauge blocks, reference standard gauge blocks and primary standard laser interferometer system in classical/conventional metrology approach as can be seen in Fig. 1. All intermediate standard measurement systems exist.

Micrometer (DUT)

Gauge blocks (Working standard)

Standard gauge blocks (Reference standard)

Laser Interferometer system (Primary standard)

SI length unit (m)

Uncertainty decrease Fig. 1. The traceability in length measurements in classical/conventional metrology.

3 Medical Metrology It should be said that the measurements in the health field are more important than the applications in classical/conventional metrology due to injury and vital risk. Medical measurements and medical measuring instruments are indispensable for managing the processes in the prevention, diagnosis and treatment of basic diseases that exist in people’s lives. Although the medical doctor relies on his own examination findings, the result of the medical measurement makes an important contribution to the decision making process.

Metrology Versus Medical Metrology

107

Therefore, the accuracy and reliability of medical measurements are crucial to the health of each individual patient. Usually, medical decisions are often based on statistical analysis and conclusions of clinical studies [5–9]. Medical measurements are incorporated within these studies and correlated with other medical findings. Treatment stages and methods are determined by the medical doctor by evaluating the data obtained from medical devices and the information obtained from the examination findings. In some cases, data from medical measurements may be more important than examination findings. For example, a muscle tear that cannot be detected on examination findings can only be seen on Magnetic Resonance Image (MRI). Consequently, each medical decision for an individual may be influenced by the results of previous studies, including the data from medical measurements. Diversity of devices and systems used in the diagnosis and treatment of diseases is increasing day by day. In order to ensure the correct operation of medical devices throughout their lifetime, periodic measurement, verification and calibration, as well as regular maintenance/repairs, is indispensable. Some devices are used for testing, measurement, verification and calibration of medical devices and systems. It is of great importance to ensure that the devices used in institutions and organizations serving in the field of health can be monitored according to national and/or international standards. Primary standards in medical metrology may not always be available. In some cases, the primary standard is human [10], but the fact that the person’s health status is not standard and different people have different characteristics may make this impossible. Measurement that are based on physiological models or assumptions is a special case for medical devices with a measuring function. Non-invasive intraocular pressure measurements are based on a comparative study of direct and indirect measurements on the eyes of approximately 10 deaths and have been used [11–14]. These measurement results have been accepted as the “gold standard” until today. In an example in the field of blood pressure measurements, it is the method of determining the pressure value from the manometer screen at the moment of detecting sound beats with the ear. Noninvasive audiometric blood pressure measurements are based on Korotfoff’s technique which mainly depends on the expertise of medical practitioner. The average value of the equivalent threshold sound pressure levels in a sufficiently large number of ears of otologically normal persons of both sexes, aged 18 to 30 years, refers to the threshold of hearing in a specified acoustical coupler or artificial ear for a specified type of earphone. This is called Reference equivalent threshold sound pressure level (RETSPL) at a specified frequency. The reference zero for the calibration of audiometric equipment for air conduction is defined in ISO 389–1:2017 [15]. Similarly, a reference zero for bone-conduction audiometry in terms of reference equivalent threshold force levels (RETVFLs), has been tabulated by means of vibratory force levels produced by a bone vibrator on a specified mechanical coupler when the vibrator is excited electrically at a level corresponding to the threshold of hearing of young otologically normal persons in ISO 389–3:2016 [16]. Measurements which are based on physiological models or assumptions are used for traceability for medical devices with a measuring function. Metrology for medical devices can be grouped as; • Classical/conventional metrology–a primary standard • Med-Met I–a clinical database used for general application

108

B. Karaböce

• Med-Met II–an individual clinical trial for a specific device is required Approaches for such circumstances can be made as can be seen in Table 2. Table 2. Traceability in metrology for medical devices Traceability in classical/conven- tional metrology

Med-met I

Med-met II

Primary standard

Database of physiological test signals

A specific device with in- dividual data

Working standard

Test device generator (trace-able to primary standards)

Transfer standard (tracea-ble to primary standards

Device under test (DUT)

DUT

DUT

Reference standard

In an example for classical/conventional metrology, Sphygmomanometer is which is DUT, calibrated pressure meters are working standard, pressure balance is reference standard and force and length are traceable primary standards that have been established in an NMI as can be seen in Fig. 2 [17]. Ergometer is DUT, calibrator is working standard, dynamic rotary power standard is reference standard and torque, length and time are traceable primary standards which were established in an NMI as can be seen in Fig. 3.

Fig. 2. Sphygmomanometer is calibrated against pressure balance that is traceable to force and length standards.

In an example for Med-Met I approach, eye tonometer is DUT. Comparative measurement have been realized between reference method and DUT. Reference is clinically tested 10 dead human eyes as can be seen in Fig. 4. The device has to be used according to the descriptions in the user’s manual. The performance of DUT is determined by a comparative measurement procedure with a reference device (eg eye tonometry).

Metrology Versus Medical Metrology

109

Fig. 3. Ergometer is calibrated against dynamic rotary power which is traceable to force, torque and length standards.

Fig. 4. Eye tonometer is calibrated against dynamic rotary power. Comparison of the pressureflow relationship in dead eyes [18].

Another example for Med-Met I can be audiometer that is DUT. Artificial ear and artificial mastoid which are for acoustical and vibrational characterization respectively are working standards. Microphone and impedance head are reference standards and reciprocity calibration of microphones and laser interferometry calibration systems are primary standards as can be seen in Fig. 5 [19]. The data are evaluated and calculated by means of comparing RETSPL and RETVFL which have been determined previously [15, 16]. In an example for Med-Met II approach, in air puff eye tonometry, an eye is DUT. Comparative measurement between reference techniques and DUT is done. Applanation tonometry since Goldmann’s study on the 10 dead human eyes, is applied. A device called “flapper” and/or a rubber eye are used as transfer standard that was clinically tested with enough number of eyes. The reference measurement or procedure has to be performed by well-trained observers, calibrated reference measurements at a time. For Med-Met II approach, the complexity of a physiological measurement process has to be taken into account. This requires a significant amount of medicine competence in addition to the competence in the sector of classical/conventional metrology.

110

B. Karaböce

Audiometer Hearing aid Headphone Other audiology devices

Artificial ear Artificial mastoid

Microphone Accelerometer

Reciprocity calibration system Acceleromeetr calibration system by laser interferometry

Fig. 5. Determination of threshold of SPL of a patient against previously determined RETSPL.

4 Conclusions The definition of four of the seven base units has been renewed. Ancient physical processes; the old references to the international prototype of the Kilogram, the electromagnetic force for the Ampere, the triple point of water for the Kelvin, and the carbon 12 atom for the mole are no longer in use. Medical metrology is an emerging discipline that must be developed awareness for current concepts in measurement practices. Understanding the measurement process and the associated metrological principles enhances the reliability of the result generated in human health. Metrology allows measurement results with required standards norms and accreditation all over the world. Establishment of metrological point of view for measurements involving human parameters should adopt the total quality approach. In order to adapt the procedures to metrology, below must be kept in mind [20, 21]; (i) knowledge of basics of human physiology (male, female, newborn, child, adult etc.), (ii) strategies to define “viability” of data (reproducibility, long-term stability) (iii) harmonization of analytical measurements, assessment of measurement uncertainties and establishment of traceability links. As human characteristics and physiology may differ by region, peoples, age etc., data sets used in audiometry and eye-tonometry as references must be taken according to this differences. A multidisciplinary study is essential in the field of medical metrology. The process and methodology for generating data from a metrological point of view should be determined. Data generation traceability may depend on DUT’s specifications. Therefore, the metrologist must establish a direct link between the quantitative entity to be measured in the study property and the numerical value assigned to it, and make this a transparent, traceable and repeatable process. A psychological measurement methodology should be taken into consideration for a medical metrology view. For this purpose, a significant amount of medical metrology knowledge and know-how with a competent experience in classical metrology is a must.

Metrology Versus Medical Metrology

111

References 1. https://www.bipm.org/en/measurement-units/si-defining-constants 2. Howarth, P., Redgrave F.: Metrology–in short, 3rd edition July 2008. ISBN 978–87- 988154– 5–7 3. [JCGM 2008, 29] 4. Dybkaer.: Metrological traceability for in-vitro diagnostic medical devices-Definitions and ISO/CEN standards 5. Karaboce, B.: Importance of medical metrology. IFMBE Proceed. 73, 443–450 (2020). https:// doi.org/10.1007/978-3-030-17971-7_67 6. Karaboce, B.: Challenges in medical metrology. IEEE Instrum. Meas. Mag. 23(4), 48–55 (2020). https://doi.org/10.1109/MIM.2020.9126071 7. Badnjevi´c, A., Cifrek, M., Magjarevi´c, R., Džemi´c, Z. (Eds).: Inspection of medical devices. Series in Biomedical Engineering. Springer, Singapore (2018) 8. Badnjevi´c, A. et al.: Post-market surveillance of medical devices: a review. Technol Health Care 30(6), 1315–1329 (2022) 9. Badnjevic, A.: Evidence-based maintenance of medical devices: current shortage and pathway towards solution. Technol. Health Care 31(1), 293–305 (2023) 10. Pendrill, L., Petersson, N.: Metrology of human-based measurements. In: 17 International congress of metrology, 10 (2015). https://doi.org/10.1051/201517001 11. Gloor, B.R.P.: Hans Goldmann (1899–1991). Eur. J. Ophthalmol. 20(1):1–11 (2010) 12. Messenio, D., Ferroni, M., Boschetti, F.: Goldmann tonometry and corneal biomechanics. Appl. Sci. 11(9), 4025 (2021). https://doi.org/10.3390/app11094025 13. Troost, R., Vogel, A., Beck, S., Schwenn, O.K., Grus, F., Pfeiffer, N.: Clinical comparison of two intraocular pressure measurement methods: SmartLens dynamic observing tonography versus Goldmann, December 2001. Graefe’s Archive Clin Exp Ophthalmol 239(12), 889– 892.https://doi.org/10.1007/s00417-001-0376-4 14. Sedlák, V., Pražák, D., Schiebl, M., Nawotka, M., Jugo, E., do Céu Ferreira, M., Duffy, A., Rosu, D.M., Pavlásek, P., Geršak, G.: Smart specialisation concept in metrology for blood and intraocular pressure measurements, measurement: sensors, Volume 18, December 2021, 100283 15. ISO 389–1.: Acoustics—reference zero for the calibration of audiometric equipment, Part 1: Reference equivalent threshold sound pressure levels for pure tones and supra-aural earphones (2017) 16. ISO 389–3.: Acoustics—reference zero for the calibration of audiometric equipment: Part 3: Reference equivalent threshold vibratory force levels for pure tones and bone vibrators (2016) 17. Gersak, G., Schiebl, M., Nawotka, M., Jugo, P.H., do C´eu Ferreira, M., Duffy, A., Rosu, D.M., Pavlasek, P., Sedlak, V., Prazak, D.: Physiology-based patient simulator for blood pressure meter testing. Meas. Sens. 18, 100260 (2021) 18. Langham, M.E., Eisenlohr, J.E.: A manometric study of the rate of fall of the intraocular pressure in the living and dead eyes of human subjects. Investig. Ophthalmol. Visual Sci. 2, 72–82 (1963) 19. Karaböce, B., Gülmez, Y., Akgöz, M., Kaykısızlı, H., Yalçınkaya, B., Dorosinskiy, L.: Medical metrology studies at Tübitak UME, 17 International Congress of Metrology, 06011 (20 5). https://doi.org/10.1051/metrology/2015 (http://cfmetrologie.edpsciences.org or https:// doi.org/10.1051/metrology/20150006011) 20. Iyengar, V.: Metrology in physics, chemistry, and biology. Biol. Trace Elem. Res. 116, 1–4 (2007). https://doi.org/10.1007/BF02685914 21. Uher, J.: Measurement in metrology, psychology and social sciences: data generation traceability and numerical traceability as basic methodological principles applicable across sciences. Qual. Quant. 54(3), 975–1004 (2020). https://doi.org/10.1007/s11135-020-00970-2

The Effect of Different Final Irrigant Activation Techniques on the Bond Strength of an MTA-Based Endodontic Sealer- An in Vitro Study Sagar Borse1 , Anita Sanap Tandale1 , Sanjyot Mulay1 , Vini Mehta2(B) , and Karishma Krishnakumar1 1 Department of Conservative Dentistry and Endodontics, Department of Dental Research Cell,

Dr. D. Y. Patil Dental College and Hospital, Dr. D. Y. Patil Vidyapeeth University, Pimpri, Pune 411018, India 2 Department of Dental Research Cell, Dr. D. Y. Patil Dental College and Hospital, Dr. D. Y. Patil Vidyapeeth University, Pimpri, Pune 411018, India [email protected]

Abstract. Aim: We aimed to evaluate and compare the effect of EndoVac™ irrigation, EndoActivator™ irrigation, Laser activated irrigation (LAI), Passive Ultrasonic Irrigation systems (PUI), and Manual irrigation system on push-out bond strength of endoseal Mineral trioxide aggregate (MTA) sealer. Settings and Design: In vitro study. Methods and Materials: Twenty five single rooted anterior teeth were prepared using ProTaper system to size F4, and using 5.25%sodium hypochlorite (NaOCl) and 17% Ethylenediamine tetraacetic acid (EDTA) a final irrigation regimen was done and then divided randomly into 5 groups: Conventional needle, Apical negative pressure using EndoVac™, Sonic Activation using EndoActivator™, PUI, and LAI. All the teeth were dried using paper point and obturated using single cone obturation using Endoseal MTA sealer. A push-out test was used to measure the bond strength. Results: The push out bond strengths of the conventional needle were lower compared to the other groups wherein different irrigation systems were used. However, a substantial increase was noted in EndoVac, EndoActivator, and PUI groups compared to the conventional needle group. The LAI system group showed no significant increase to the conventional group. Conclusion: Different irrigation activation systems increased the bond strength of MTA sealer. Keywords: Bond strength · Endoseal MTA sealer · Irrigation activation

1 Introduction Mechanical preparation with effective irrigation along with intracanal medicament and obturation of root canal (RC) space is the aid that can be used to attain endodontic success [1]. Sodium hypochlorite (NaOCl) and Ethylenediamine tetraacetic acid (EDTA) are © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 112–119, 2024. https://doi.org/10.1007/978-3-031-49068-2_13

The Effect of Different Final Irrigant Activation Techniques

113

used as an irritant for removal of organic as well as inorganic contents of smear layer. It was stated that irrigants are more effective when they are in direct contact with the entire RC wall. Although, it might be difficult using conventional needle irrigation as the anatomy of RC is complex [2]. Several irrigation activation systems had been introduced to improve action of irrigation solutions in the RC system. The systems increase the ability of smear layer removal efficiently, increase the contact time with RC dentine wall, and improve the penetration of irrigating solution in complex RC structures [2]. Mineral trioxide aggregate (MTA) is a portland cement–derived hydraulic material that has been widely used in a variety of endodontic applications. The crystalline deposits produced by the interaction of MTA and physiological fluids positively influence the push-out bond strength of MTA [3]. Endoseal MTA which was recently proposed, is a finely pulverized pozzolan-based MTA sealer. Sealer has a fast setting time (about 12min), higher resistance to washout compared to other MTA’s, alongwith biological effects which include bio-mineralization, bio compatibility, and odontogenic activity which are some of the favorable properties [4]. In spite of having good properties, many studies have shown that MTA-based sealer did not have adequate push-out bond strength and hence we aimed to evaluate pushout bond strength of Endoseal MTA sealer after using different irrigation activation systems.

2 Materials and Methods 2.1 Sample Selection and Preparation Twenty five freshly extracted human permanent single rooted teeth were taken for this study with 75 sections. Tooth was verified radiographically for the single patent canal of curvature less than 5 degrees, fully formed apices, intact root with no cracks, no calcification, no internal resorption, or earlier RC treatment. Specimens were stored in 3% hydrogen peroxide solution for 10 min to remove organic debris which was adhered to external root surface of tooth and to disinfect samples, which were further cleaned up with an ultrasonic scaler to remove remnants of tissue tags. The samples were stored in 0.2% thymol solution until further use. To standardize samples, they were decoronated with the diamond disc in low-speed straight handpiece under a water coolant just enough to protect crown portion as a reservoir for irrigation solution and length was standardized to 21 mm. 2.2 Root Canal Instrumentation and Preparation A size 10 K file was passed 1mm beyond apical foramen ensuring canal patency. The file which snugly fit at the apical foramen was selected for individual tooth and working length was established by subtracting 1mm from this length. Cleaning and Shaping were performed using Protaper Universal rotary File system till F4. In between instrumentation 5.25% NaOCl and 17% EDTA gel was used for lubrication. Teeth were divided randomly into 5 groups, 5 teeth in each group and routine irrigation protocol was applied to all groups with activation of irrigant with respective systems: Group A: Conventional needle irrigation control group (n = 15).

114

S. Borse et al.

Irrigation by conventional needle. Group B: EndoVac™ (n = 15). Irrigation by EndoVac irrigation activation system. Group C: EndoActivator™ (n = 15). Irrigation by EndoActivator irrigation activation system. Group D: Passive Ultrasonic Irrigation (n = 15). Irrigation by Ultrasonic irrigation activation system. Group E: Laser activated Irrigation (n = 15). Irrigation by Laser irrigation activation system. After final irrigation activation canals were irrigated with saline. A master guttapercha of size F4 was selected and apical tug back was checked radiographically. RCs were dried with paper points. Endoseal MTA sealer was used according to the manufactures instruction where it was injected into RC and master cone was positioned into the canal. The tug back was confirmed again. A heated instrument was employed to cut gutta-percha off at RC entrance. The access cavity was sealed using Intermediate Restorative Material (IRM). The radiographs were taken for each tooth to assess the obturation for any voids. All teeth were stored in a humidor at 37°C for 24 h for complete setting of sealer. From each group, 15 samples were made by cutting the tooth perpendicular to their long axis to obtain slices from coronal, middle, and apical portions (2 mm each) with a slow-speed diamond saw. The exact dimension of each disk was measured by a digital calliper to be within the range of 2 ± 0.04 mm. Samples displaying physical deformation signs were discarded immediately. Then sections underwent a push-out test using a universal testing machine, calculating diameter of the canal using image analysis software the plunger sizes of different diameter ranging from 1 mm to 0.6 mm were selected. The maximum load applied to each sample was noted down. The push-out bond strength in MPa was analysed.

3 Statistical Analysis The data were analyzed with Statistical Package for Social Sciences (SPSS) for Windows, version 25.0 (IBM Corp., Armonk, N.Y., USA). Confidence intervals were set at 95% and values of p < 0.05 were interpreted as statistically significant. One way ANOVA was used to compare bond push-out strength for different irrigation activation systems. Tukey’s Post Hoc test was computed to analyze in between-group differences of 5 different irrigating activation systems.

4 Results EndoVac showed the highest bond strength (8.841 ± 1.79) followed by PUI (7.856 ± 1.95), EndoActivator (7.641 ± 1.89), and Laser activation of irrigation (6.016 ± 1.24) (Table 1). The push-out bond strengths of control (conventional needle) were lower than other groups where in different irrigation systems were used. However there was a significant increase in the pushout bond strength only in EndoVac, EndoActivator, and passive ultrasonic groups compared to the conventional needle group. The laser irrigation system group did not show any significant increase than the conventional group (Table 2).

The Effect of Different Final Irrigant Activation Techniques

115

5 Discussion Adhesion of RC filling to dentinal walls has great significance in dynamic and static conditions. During dynamic conditions, resistance to dislodgement of this material during subsequent manipulation is needed. While in a static situation, it must eliminate any space allowing percolation of fluids between the filling and the wall [5]. During chemo-mechanical preparation, a smear layer is generated which can serve as a reservoir or source for microorganisms. Due to this, it can block progression of sealer tags into dentin tubules, such that micromechanical adhesion is decreased [1]. Cleaning and shaping are more effective for the central portion of RC system and not so for anatomical complexities like canal fins, cul-de-sacs, accessory, and lateral canals and isthmi [2]. Activating irrigants facilitates their contact with canal complexities will effectively cause smear layer removal, eventually helping with improvement of obturating material adhesion with dentine wall. Conventional needle irrigation is widely employed and effective to deliver irrigant to root canal before introduction of the various other systems. Various modifications like side vented tip design and smaller gauge of the needle to improve efficacy without extrusion of irrigant are made with conventional needle irrigation. However, these modifications have been ineffective. Limited action achieved with increased risk of periapical extrusion of the irrigating solution are disadvantages that led to the development of new activation systems [2]. EndoVac™ is based on the principle of pressure alteration in which negative pressure is created into the root canal which allows the continuous flow of fresh irrigant till the working length. The EndoVac system is regarded as an apical negative pressure irrigation system composed of three basic components: a Master Delivery Tip (MDT), the Macrocannula, and the Microcannula. The MDT delivers irrigant to the pulp chamber and evacuates the irrigant concomitantly. Both the macrocannula and microcannula are connected via tubing to a syringe of irrigant and the highspeed suction of a dental unit [6]. EndoActivator™ is based on the principle of sonic activation of root canal irrigant. A specialized disposable polymer tip of small,medium,large having yellow, red and blue colors respectively are used which 2–3 vertical strokes creates sonic vibrations with a strong hydrodynamic phenomenon 10,000cycles /min are given to encourage debridement [7]. PUI activation is based on ultrasonic energy, which is transferred from an oscillating file or a smooth wire into RC irrigant through ultrasonic energy that encourage mechanism of acoustic streaming and cavitation of irrigating solution in RC 5 mL of 5.25% NaOCl, then 5 mL of 17% EDTA and 5 mL of distilled water was used for irrigation of root canals. Each irrigant was agitated with an ultrasonic system handpiece, irri safe (Satelac) equipped with a size 25 IRRI Ssmoothwire at 2 mm short of the working length. The irrigation was ultrasonically applied to the root canals for 3 min in total along with 5 mL of 17% EDTA, 5 mL of 5% NaOCl, and 5 mL of distilled water for 1 min (3 cycles of 10s) for each irrigant according to manufactures instructions [3]. In laser-activated disinfection, photomechanical actions like bubble formation and cavitation generate in the irrigation solutions inside root canals when a laser is applied in a pulsative manner in the irrigant filled canal. This photomechanical action induces an effect like a shockwave, which results in Laser activated disinfection for the better activity of the irrigant. 5 mL of 5.25% NaOCl, 5 mL of 17% EDTA and 5 mL of distlled

116

S. Borse et al.

water was used for irrigation of root canals. Each irrigant was fully activated for 1 min (3 cycles of 20 s) with 1064-nm wavelength Neodymium-doped Yttrium Aluminum Garnet (Nd:YAG) laser (Fotona)at 1 W/cm2 Power (20 Hz Frequency, 50 mj/cm2 Energy density) with a pulse duration of 50 µs by a non-cooled handpiece with 300 µm optical fiber [8]. Adhesion is tested using many tests in dentistry among which push-out bond test is the most popular as it has certain advantages like sample preparation is effective and reproducible and also it can be evaluated in low values of bond strengths. The pushout test depends upon shear stresses at junction between cement and dentin, similar to the stress occurring in clinical conditions [9]. Our study showed mean push-out bond strength values to be maximum for EndoVac™ group followed by EndoActivator™, passive ultrasonic irrigation, laser, and conventional needle irrigation. Also, all irrigation activation systems showed a statistically significant difference when compared with conventional syringe irrigation. Table 1. Comparison of bond push-out strength for different irrigation activation systems Groups irrigation activation system

Pushout bond strength (MPa)

F, df

Mean ± SD Group A (Conventional needle irrigation control group)

4.860 ± 0.82

Group B (EndoVac™)

8.841 ± 1.79

Group C (EndoActivator™)

7.641 ± 1.89

Group D (Passive ultrasonic)

7.856 ± 1.95

Group E (Laser)

6.016 ± 1.24

14.428,4

Table 2. Pairwise comparison of bond push-out strength for different irrigation activation systems. Irrigation activation systems

Conventional needle

EndoVac

EndoActivator

Passive ultrasonic

Laser

Conventional needle

–

0,001*

0,001*

0,001*

0,29

EndoVac

0,001*

–

0,21

0,45

0,001*

EndoActivator

0,001*

0,21

–

0,98

0,08

Passive ultrasonic

0,001*

0,45

0,98

–

0,02

Laser

0,29

0,001*

0,08

0,02

–

Tukey’s Post Hoc* Significant.

The Effect of Different Final Irrigant Activation Techniques

117

EndoVac™ group has shown a statistically significant difference when compared with controls and in LAI. The mean value for EndoVac™ group being highest 8.841 ± 1.79, this could be because the negative pressure generated by the macrocannula and microcannula into the canal which effectively delivers irrigant to full working length as well as by suctioning effect pulls irrigant from the apex this establish a constant flow of the irrigant till working length. The continuous flow of irrigant on walls of RC along with apical suction effect creates a very quick turbulent cascade impact as the irrigant is forced to flow between the external surface of microcannula and RC walls. This position of microholes on the microcanula reverses the direction of the fast-flowing stream of irrigant from a full working length as close as 0.2 mm, generates turbulent action creating a current force, which effectively removes smear layer in all parts of RC [10]. The EndoActivator™ group has a mean push-out bond strength value of 7.641 ± 1.89. The greater push-out bond strength of EndoActivator could be attributed to the creation of acoustic streaming, cavitation by a combination of sonic vibration and vertical strokes of the tip. The production of single node and antinode could have increased the irrigant contact time with canal walls facilitating removal of smear layer [11]. For our study, the PUI group has to mean push-out values of 7.856 ± 1.95 and this is attributed to the formation of acoustic micro-streaming and cavitation effect generated by transmitting acoustic energy to an irrigant in the RC through oscillating file or smooth wire, this acoustic energy is in the form of ultrasonic waves. The acoustic microstreaming generates shear flow and shear stresses which remove bacteria and debris from the RC wall. Acoustic cavitation creates new bubbles or contraction, expansion, and/or distortion of previous present bubbles in an irrigation solution which gets imploded on the root canal walls creating a force that helps in the removal of debris [12]. Another important factor in PUI is at the closest point of the tip of the instrument there is rise in temperature into RC from 37 °C to 45 °C and 37 °C distant from the tip when irrigant was continuously activated for 30s with no replenishment which is shown in a paper by Cameron et al. (1988) [13], Ahmed et al. (1990) [14], Cunnigham et al. (1980) [15]. Krishna Prasad et al. in their research concluded that PUI system and EndoVac negative pressure system are more effective than conventional endodontic needles in delivering the irrigant to the working length of root canals. They found no significant difference was seen between the PUI and EndoVac groups (p = 0.06). [16] Narmatha and Sophia in their study found that Passive Ultrasonic agitation produces the cleanest canal walls compared to Manual Dynamic activation and Canal Brush agitation. Canal Brush agitation and Manual Dynamic activation can be used effectively as irrigant agitation techniques compared to conventional syringe irrigation. Canal Brush agitation produced significantly cleaner canal walls compared to Manual Dynamic activation [17]. A systematic review carried out by Carla M. Augusto et al. inferred that most of the final irrigation protocols had a positive impact and promoted an improvement in the dislodgment resistance of root canal sealers to the root dentin [18].

118

S. Borse et al.

6 Conclusion In conclusion, using different irrigation activation system has increased the bond strength of Endoseal MTA sealer as compared to the conventional syringe irrigation. EndoVac™, EndoActivator™, PUI, laser activation have shown improved bond strength of this sealer. Machine assisted irrigation protocol should be incorporated as routine irrigation protocol.

7 Conflict of Interest None. Funding. None.

References 1. Topcuoglu, H.S., Tuncay, O., Demirbuga, S., Dinçer, A.N., Arslan, H.: The effect of different final irrigant activation techniques on the bond strength of an epoxy resin-based endodontic sealer: a preliminary study. J. Endod. 40, 862–866 (2014) 2. Gu, L.S., Kim, J.R., Ling, J., Choi, K.K., Pashley, D.H., Tay, F.R.: Review of contemporary irrigant agitation techniques and devices. J. Endod. 35, 791–804 (2009) 3. Sagsen, B., Ustün, Y., Demirbuga, S., Pala, K.: Push-out bond strength of two new calcium silicate-based endodontic sealers to root canal dentine. Int. Endod. J. 44, 1088–1091 (2011) 4. Yoo, Y.J., Baek, S.H., Kum, K.Y., Shon, W.J., Woo, K.M., Lee, W.: Dynamic intratubular biomineralization following root canal obturation with pozzolan-based mineral trioxide aggregate sealer cement. Scanning 38, 50–56 (2016) 5. Ungor, M., Onay, E.O., Orucoglu, H.: Push-out bond strengths: the Epiphany-Resilon endodontic obturation system compared with different pairings of Epiphany, Resilon, AH Plus and gutta-percha. Int. Endod. J. 39, 643–647 (2006) 6. Mancini, M., Cerroni, L., Iorio, L., Armellin, E., Conte, G., Cianconi, L.: Smear layer removal and canal cleanliness using different irrigation systems (EndoActivator, EndoVac, and passive ultrasonic irrigation): field emission scanning electron microscopic evaluation in an in vitro study. J. Endod. 39, 1456–1460 (2013) 7. Niu, L.N., Luo, X.J., Li, G.H., Bortoluzzi, E.A., Mao, J., Chen, J.H., et al.: Effects of different sonic activation protocols on debridement efficacy in teeth with single-rooted canals. J. Dent. 42, 1001–1009 (2014) 8. Akyuz Ekim, S.N., Erdemir, A.: Effect of different irrigant activation protocols on push out bond strength. Lasers Med. Sci. 30, 2143–2149 (2015) 9. Ureyen Kaya, B., Keçeci, A.D., Orhan, H., Belli, S.: Micropush-out bond strengths of gutta percha versus thermoplastic synthetic polymer-based systems - an ex vivo study. Int. Endod. J. 41, 211–218 (2008) 10. Alkahtani, A., Al Khudhairi, T.D., Anil, S.: A comparative study of the debridement efficacy and apical extrusion of dynamic and passive root canal irrigation systems. BMC Oral Health 14, 12 (2014) 11. de Gregorio, C., Estevez, R., Cisneros, R., Heilborn, C., Cohenca, N.: Effect of EDTA, sonic, and ultrasonic activation on the penetration of sodium hypochlorite into simulated lateral canals: an in vitro study. J. Endod. 35, 891–895 (2009)

The Effect of Different Final Irrigant Activation Techniques

119

12. Van der Sluis, L.W., Versluis, M., Wu, M.K., Wesselink, P.R.: Passive ultrasonic irrigation of the root canal: a review of the literature. Int. Endod. J. 40, 415–426 (2007). Cameron, J.A.: The effect of ultrasonic endodontics on the temperature of the root canal wall. J. Endod. 14, 554–559 (1988) 13. Ahmad, M.: Measurements of temperature generated by ultrasonic file in vitro. Endod. Dent. Traumatol. 6, 230–231 (1990) 14. Cunningham, W.T., Balekjian, A.Y.: Effect of temperature on collagen-dissolving ability of sodium hypochlorite endodontic irrigant. Oral Surg. Oral Med. Oral Pathol. 49, 175–177 (1980) 15. Prasad, K.P., Tiwari, A., Sankhala, A., Madan, G., Parakh, S., Singh, A.: Effect of different irrigant activation techniques on the bond strength of a mineral trioxide aggregate-based sealer: an in vitro study. IJOCR 5(2), 91–96 (2017) 16. Narmatha, V.J., Thakur, S.: Evaluation of manual dynamic activation, passive ultrasonic irrigation and canalbrush on smear layer removal-a scanning electron microscopic study. IJAR 3(3), 390–400 (2015) 17. Augusto, C.M., Cunha Neto, M.A., Pinto, K.P., Barbosa, A.F.A., Silva, E.J.N.L., dos Santos, A.P.P., Sassone, L.M.: 18. Influence of the use of chelating agents as final irrigant on the push-out bond strength of epoxy resin-based root canal sealers: a systematic review. Aust. Endod. J. 48(2), 347–363 (2022)

Verification of Ultrasound Imaging Phantoms: An Evaluation Study Baki Karaböce(B)

and Hüseyin Okan Durmu¸s

TÜB˙ITAK UME, Gebze, Kocaeli 41470, Türkiye {baki.karaboce,huseyinokan.durmus}@tubitak.gov.tr

Abstract. Ultrasonic imaging is one of the most popular soft tissue imaging techniques currently used in modern medicine. It is widely used because it is easy to apply, economical, fast, and gives a snapshot. The absence of side effects of ultrasonic waves, such as X-ray or tomography, which are other commonly used imaging techniques, provides safe use. It can even be used for imaging the fetus in the womb. Non-invasive imaging of soft tissues for diagnostic and therapeutic applications offers great advantages to medical practitioners. Verification of ultrasound imaging devices is done with ultrasonic imaging phantoms. Phantoms are water-based chemical mixtures with agar or Zerdine in their structure. Phantoms are water-based chemical mixtures that are agar or Zerdine in their structure, and their density can change depending on time, and accordingly the sound velocity and sound absorption coefficient change. In this study, acoustic parameters such as sound velocity and acoustic attenuation coefficient of two different ultrasound phantoms, Doppler 403 Flow Phantom and Multi-Purpose Phantom, were evaluated according to manufacturer specifications. The measurement system consists of ultrasonic imaging phantom, ultrasonic transducer (probe), pulser-receiver and oscilloscope. The verification of the measuring system was checked with a stainless steel ladder block. The physical controls of the phantom were primarily made visually and with an ultrasonic imaging device. Ultrasonic transducer was coupled with the surface of the phantom using ultrasonic gel and echo signals reflected from the phantom were evaluated by means of receiving signal from transducer and oscilloscope. Analytical evaluation of verification methods of ultrasonic imaging phantoms will make a valuable contribution to metrological measurements. In addition, in light of this study, a few suggestions for phantom design will be presented to facilitate the measurement process for a better metrological evaluation of phantoms. Keywords: Measurement · Ultrasound imaging phantoms · Sound velocity · Acoustic attenuation coefficient · Medical metrology

1 Introduction Tissue mimicking materials or phantoms are test materials used in many research studies and testing procedures for many diagnostic equipments. Phantoms are used extensively in radiation therapy, imaging and computational fields, especially in medical and health © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 120–131, 2024. https://doi.org/10.1007/978-3-031-49068-2_14

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

121

physics. In the field of radiation therapy; radiation therapy dosimetry phantoms, anthropomorphic phantoms for radiation oncology medical physics, motion phantoms for radiotherapy and phantoms for brachytherapy are used, while x-ray, computed tomography, mammography, ultrasound experimentation and quality control, magnetic resonance imaging (MRI) and nuclear medicine (NM) and positron emission tomography (PET) phantoms are used in the field of imaging. In addition, there are computational phantoms for organ dose in radiation protection and imaging, and computational phantoms’ applications [1]. Ultrasonic imaging is a widely used diagnostic technique that utilizes high-frequency sound waves to produce images of internal organs and tissues within the body [2– 4]. The accuracy and quality of ultrasonic images are essential for effective diagnosis and treatment planning [5, 6]. To ensure the reliability and consistency of ultrasonic imaging, specialized phantoms are used [7, 8]. Applications for ultrasound phantoms might differ greatly. While the majority of phantoms created for sale are employed in quality control procedures, the majority of phantoms generated on-site are utilized for experimental and development purposes by research teams. Furthermore, studies proposing a new method for post-market surveillance of medical devices, evidencebased maintenance of medical devices and conformity assessment tests of therapeutic ultrasounds have attracted attention in recent years [9–11]. These studies are just a few examples of research done to make medical devices safer to use. These studies show that new approaches to the use of medical devices may emerge with the development of technology as well as advances in the field of medicine. The necessity for representations of human anatomy and tissue characteristics gave rise to the phantoms utilized in ultrasonography. The first ultrasonic phantom was built using metal rods placed at certain intervals to give distance calibration and was based on water-filled containers. The necessity for a better material to offer a medium for sound to be delivered at the proper speed and to have reflectors that cause the echo to return to the transducer was realized as ultrasound technology got more sophisticated and advanced. As a result, just like with many imaging modalities, the advancement of ultrasound imaging technology has led to the emergence of ultrasound phantoms. Materials like urethane polymers or soft plastics were initially utilized as possible environments for phantom construction; However, these materials failed to match some of the physical parameters of ultrasound. A team from the University of Wisconsin-Madison presented the first study on materials created to imitate the characteristics of tissue [12, 13]. The manufacturing of the gel used in this study was modeled after the tissue procedures, which served as the foundation for several ultrasonic phantoms in the future. The University of Wisconsin ultrasound group’s early work included designing and constructing ultrasound phantoms. These phantoms served dual purposes: to ensure quality control of ultrasound imaging, and to gain a deeper understanding of how ultrasound physics and tissue propagation impact imaging performance. In later years, the group created anthropomorphic phantoms for use in a quality assurance program for a large breast ultrasound study across multiple centers. The development of ultrasound phantoms has seen contributions from various sources including the FDA and research groups in the UK, Netherlands, France, Germany, and Canada. A significant number of ultrasound phantoms have been created for evaluating the performance of imaging systems. Some

122

B. Karaböce and H. O. Durmu¸s

phantoms are also utilized in experiments to measure properties such as attenuation, backscatter, ultrasound exposure, and bulk material characteristics. These latter phantoms serve as “gold standards” to confirm the accuracy of the ultrasound methods used to determine tissue properties. Aside from laboratory development, there are also numerous commercial companies (Gammex, CIRS, ATS) that produce and sell ultrasound phantoms for quality control and training purposes. With the global ultrasound equipment market reaching nearly $5 billion per year and a steady 5% growth rate, these vendors have a large and expanding market for various quality assurance phantoms. Some of these designs are based on or inspired by standards established by national and international committees relating to ultrasound image quality. For instance, the Technical Standards Committee of the American Institute of Ultrasound in Medicine has published guidelines for commercially available phantoms, and the International Electrotechnical Commission, Group 87 has documented phantoms and testing procedures for diagnostic ultrasound equipment. Most commercial phantoms today, however, are unique in design and created by vendors with a focus on common elements needed for ultrasound quality control, such as a scanning/acoustic window, evenly spaced reflectors, and a background with specified sound speed and attenuation properties. Alongside these phantoms, there have also been various techniques published for making low-cost and easily accessible phantoms for training and verifying positioning equipment. While these “in-house” phantoms are cost-effective and easy to obtain, they lack the absolute validation of their acoustic properties. The uses for ultrasound phantoms vary greatly. Most commercially available phantoms are utilized for quality control purposes, while many of the “in-house” phantoms are produced by research groups to satisfy their experimental and development requirements. Regardless of the purpose, phantom materials are employed to determine sound speed, attenuation, and backscatter [1]. These phantoms are designed to mimic the physical characteristics of human tissue, including its acoustic properties, such as speed of sound, attenuation, and backscatter/reflection. By testing the imaging performance of ultrasonic equipment using phantoms, medical professionals can assess the accuracy and quality of images generated by the equipment and identify any issues that may affect the accuracy of the images. There are several types of ultrasonic imaging phantoms, each designed to test specific aspects of ultrasonic imaging technology. One of the most common types of phantoms is the simple reflector phantom, which is used to evaluate the performance of the ultrasonic transducer. This type of phantom typically consists of a solid block of material with a smooth, flat surface that reflects sound waves back to the transducer. The accuracy and quality of the reflection are used to assess the performance of the transducer. Another type of phantom is the target phantom, which is used to evaluate the spatial resolution of ultrasonic imaging systems. This type of phantom consists of multiple small targets of varying sizes and shapes, placed at different depths within the phantom. By measuring the ability of the ultrasonic imaging system to accurately identify and distinguish the different targets, medical professionals can determine the spatial resolution of the system. In addition to simple reflector and target phantoms, there are also more complex phantoms designed to evaluate the overall performance of ultrasonic imaging systems. These phantoms typically include multiple regions of varying acoustic properties, including

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

123

different tissue types and structures, such as bones and fluid-filled cavities. These phantoms allow medical professionals to evaluate the ability of ultrasonic imaging systems to produce high-quality images of the different tissue types and structures within the phantom. The use of ultrasonic imaging phantoms has numerous benefits. By providing a controlled environment for testing and evaluating ultrasonic imaging equipment, phantoms help ensure that the images produced by the equipment are of the highest quality and accuracy. Additionally, the use of phantoms helps to identify any issues with the equipment, such as transducer degradation or system failures, before they affect the accuracy of images produced during patient examinations. Despite the numerous benefits of ultrasonic imaging phantoms, there are also some limitations and challenges associated with their use. One of the main challenges is the cost of phantoms, which can be quite high, especially for the more complex phantoms designed to evaluate the overall performance of ultrasonic imaging systems. Additionally, the materials used to manufacture phantoms must closely mimic the acoustic properties of human tissue, which can be difficult to achieve. In this study, the aim was to evaluate the acoustic parameters (sound velocity and acoustic attenuation coefficient) of two different ultrasound phantoms, Doppler 403 Flow Phantom and Multi-Purpose Phantom, according to the manufacturer’s specifications. The measurement system consisted of an ultrasonic imaging phantom, an ultrasonic transducer (probe), a pulser-receiver, and an oscilloscope. The accuracy of the measuring system was confirmed using a stainless steel ladder block. The physical characteristics of the phantoms were first visually inspected and then confirmed using an ultrasonic imaging device. The ultrasonic transducer was carefully coupled to the surface of the phantom using ultrasonic gel, and the echo signals reflected from the phantom were evaluated by receiving the signal from the transducer and observing it on the oscilloscope. The analytical evaluation of the verification methods for ultrasonic imaging phantoms will make a valuable contribution to metrological measurements. Furthermore, this study presents several suggestions for phantom design to facilitate the measurement process and improve the metrological evaluation of phantoms. By carefully assessing the acoustic characteristics of the Doppler 403 Flow Phantom and the Multi-Purpose Phantom, the study provides valuable insights into the metrological evaluation of ultrasonic imaging phantoms.

2 Materials and Methods 2.1 Measurement Setup Ultrasound imaging device (Ultrasonix Sonix Touch Q+), pulser / receiver (JSR Ultrasonics DPR300) and oscilloscope (Keysight DSOX1202A InfiniiVision) devices were used in the measurement device used to determine the acoustic parameters of ultrasonic imaging phantoms. Sound velocity and acoustic attenuation coefficient measurements were performed for the determination of acoustic parameters of ultrasonic imaging phantoms as per our laboratory instructions. And all measurements were performed in a laboratory environment with controlled ambient conditions (temperature was (23±3) °C and humidity was (45±15) rh%).

124

B. Karaböce and H. O. Durmu¸s

2.2 The Phantoms Used in the Measurements Acoustic measurements for two different model phantoms were made. One model was doppler 403 flow phantom (Gammex Inc., 1425B-05 model) and the other was multipurpose phantom (ATS Laboratories Inc., AT 539 model) as shown in Fig. 1. Measurements were made according to the specifications given by the device manufacturers [14, 15]. In this context, only the sound velocity for the doppler 403 flow phantom and both the sound velocity and attenuation coefficient parameters for the multipurpose phantom have been specified by the device manufacturers. For this reason, only these parameters were made according to the specified methods in phantom measurements and evaluated according to the expected tolerances.

Fig. 1. Two different model phantoms used in the measurements a) Gammex Inc., 1425B- 05 model, b) ATS Laboratories Inc., AT 539 model

2.3 Sound Velocity Measurements For sound velocity measurements, the depth of the phantom must first be clearly determined. The determination of the depth of the phantom was carried out with an ultrasound imaging device (Ultrasonix “Sonix Touch Q+”). A calibrated probe (NDT SYSTEMS, ICHF016 model) was connected to the ultrasound imaging device, which would reach the maximum depth. For example, the C7-3/50 probe (3-7 MHz, 50 mm, curved array) was such a probe. After the C7-3/50 probe was connected to the device, the “Depth” key was selected on the ultrasound imaging screen and the depth information was measured by selecting the highest depth by means of the mechanical button next to it. The measurement of the depth information was performed as shown in Fig. 2. After obtaining the depth information, an ultrasonic probe with a calibration certificate was connected to a measurement system consisting of a pulser/receiver and

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

125

Fig. 2. Depth measurement performed using an ultrasound imaging device

oscilloscope. Before taking measurements on the phantom, the linearity of the measurement system was checked with a ladder block as shown in Fig. 3. After the verification was made, measurements were later started on the phantoms that will be measured in the same way. For instance, the measurement made on Gammex phantom can be seen in Fig. 4. Connections were made in the following way. The SYNC connection from the Pulser/Receiver was connected to the Ext Trig on the oscilloscope. The OUTPUT connection from the Pulser/Receiver was connected to the channel 1 of oscilloscope and the T/R output from the Pulser/Receiver was connected to the probe. The adjustment was made in the Pulser/Receiver as follows; in the Receiver Echo, Real Gain; 40 and 4, HP Filter at 1.0 MHz (for example, this value can be re-set to 5 MHz if the waves do not come out well), LP Filter at 5.0 MHz, PRF Rate 6, Trigger Int., Pulse Amplitude 8, Pulse Energy 4 (Low) and Dumping 1. On the oscilloscope; X1, X2, Y1 and Y2 were selected with “Push to Select” button, the shape on the oscilloscope was enlarged and reduced with “Push Fine” button, and the screenshot was shifted to the right or left with “Push Zero” button. In the sound velocity measurement, the distance between X1 and X2 (X) gives us the time information in nanoseconds, and the distance between Y1 and Y2 (Y) gives us the amplitude information of the wave required for the attenuation coefficient. Y was calculated for both successive waves in this way. In the Pulse - Echo method, an ultrasonic probe (transducer) was used as both a receiver and a transmitter. The probe was brought into contact with the phantom with the help of the appropriate impedance harmonizing gel and the signals sent and received from the Pulser/Receiver device were visualized on the oscilloscope screen. On the oscilloscope screen, the period value between the observed signal peaks was observed and recorded as indicated in Fig. 5, and the sound velocity was calculated using the formulas (1) and (2) given in the following. The important point here is that the thickness

126

B. Karaböce and H. O. Durmu¸s

Fig. 3. Connecting the probe to the measurement system and a measurement data received with a ladder block before phantom measurements

Fig. 4. The acoustical measurement made on Gammex phantom

(or depth) of the sample being studied were properly determined beforehand. x = c · t x = the distance traveled by sound (2d) (m) c = sound velocity (m/s) t = time difference measured from oscilloscope (s)

(1)

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

127

Fig. 5. Echo patterns observed in the determination of sound velocity by the Pulse - Echo method

Since the wave has traveled the sample twice during the time when the pulse and the first echo were detected in the Pulse - Echo method, distance (x) was taken as x = 2d. Therefore, the formula is used as follows. c = (2 · d)/t

(2)

Since the depth information (d) was detected by the ultrasound device, the sound velocity was determined by measuring the time value from the difference X1 and X2; (X) detected between the two waves from the oscilloscope. The cursors X1 and X2 were set to point to the starting point of the first signal and the starting point of the second signal. These processes were repeated many times to calculate the uncertainty value for the sound velocity. These values were then compared with the reference values. 2.4 Acoustic Attenuation Measurements In order to calculate the attenuation coefficient, it was necessary to determine the peaks of two consecutive waves. First of all, the peak values of each wave in the oscilloscope screen were determined. Then, the attenuation coefficient was calculated with the help of the following formula. A = A0e−αx

(3)

where α is the attenuation coefficient (dB/(cm·MHz)) and x is the distance (cm). Again, x is taken as 2d. The attenuation coefficient has different values for each tissue. This parameter refers to the amplitude drop that occurs in the wave due to factors such as absorption, scattering and mode conversion after the ultrasonic wave enters the tissue.

128

B. Karaböce and H. O. Durmu¸s

2.5 Explanation of Calibration or Verification Process Measurements of both the sound velocity and the acoustic attenuation coefficient of the phantom were made. To determine the acoustic parameters of ultrasonic imaging phantoms, depth data required for the sound velocity were taken first. The determination of the depth of the ultrasonic imaging phantom was carried out with the ultrasound imaging device (Ultrasonix “Sonix Touch Q +”). A probe was connected to the ultrasound imaging device, which would reach the maximum depth. The depth data was then measured with the help of the “Measure” key located on the Ultrasonix “Sonix Touch Q +”. This depth information was used in formula (2). Then, using the pulser/receiver and oscilloscope system, the time differences between the signals (X) were found as nanoseconds and the sound velocity was calculated using the corresponding formula. This process was repeated at least 10 times and the resulting values were compared with the reference values. For acoustic attenuation coefficient measurement, ultrasonic transducer (probe) were connected to the system consists of pulser/receiver and oscilloscope and the amplitude differences (Y) of two consecutive signals in wave clusters were measured in Volts and their acoustic attenuations were calculated using formula (3). This process was repeated at least 10 times and the resulting values were compared with the reference values. 2.6 Standart Requirements for Acoustic Parameters As per IEC TS 62791:2015 (Ultrasonics - Pulse-echo scanners - Low-echo sphere phantoms and method for performance testing of gray-scale medical ultrasound scanners applicable to a broad range of transducer types) [16] and IEC 1390 (Ultrasonics - Realtime pulse-echo systems - Test procedures to determine performance specifications) [17] standards, the acoustic parameters that tissue-like materials used in the quality controls of ultrasonic devices should have are as follows. – Ultrasonic sound velocity must be (1540 ± 10) m/s – Eco targets for low attenuation coefficient must have a value of (0.50 ± 0.04) dB/(cm·MHz) attenuation coefficient and a “background material” must also have a value of (0.70 ± 0.04) dB/(cm·MHz) attenuation coefficient. The first and most important parameter in the phantoms to be developed for use in the performance controls of ultrasonic imaging devices is to ensure that the acoustic parameters of the developed phantom meet the values specified in the standard.

3 Results and Discussion Measurements were made according to the specifications given by the device manufacturers. For the Doppler 403 flow phantom, only the speed of sound was measured due to given single parameter in its specification. As for the Multipurpose Phantom, both the sound velocity and attenuation coefficient parameters were measured. All the sound velocity measurement results are given in Table 1 and the result of attenuation coefficient parameter measurement for the Multi-Purpose Phantom is also given in Table 2.

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

129

Table 1. Sound velocity measurement results for the studied phantoms Name of the phantom

Reference value (m/s) Measured value (m/s) Uncertainty (m/s)

Doppler 403 flow phantom 1550.0 ± 10.0

1549.0

15.5

1450.0 ± 14.5

1444.9

15.0

Multipurpose phantom

Table 2. Acoustic attenuation coefficient measurement results for multipurpose phantom Reference value (dB·cm−1 ·MHz−1 )

Measured value (dB·cm−1 ·MHz−1 )

Uncertainty (dB·cm−1 ·MHz−1 )

0.500 ± 0.025

0.520

0.160

In Tables 1 and 2, the reference values, the measured values and uncertainty values are given. ± values in reference value are given as tolerance values by the device manufacturers. Standart deviations have not used for the measured values, instead of this, uncertainty values are given because it is metrologically more accurate. As can be seen from the results for both of the devices, the values found are within the specification given by the device manufacturer. In addition to this, the results have been also evaluated statistically using separately the two-sample t-test and z-test at the 0.05 alpha level and found statistically significant at 95% confidence level. As for the methods for calculation of uncertainty, certificate uncertainty, resolution and repeatability uncertainty components have been mainly used. The reported expanded uncertainty of measurement is stated as the standard uncertainty of measurement multiplied by the coverage factor k = 2, which for a normal distribution corresponds to a coverage probability of approximately 95%. Tissue mimicking materials or phantoms play a crucial role in many medical research studies and testing procedures for diagnostic equipment. They are used extensively in the fields of radiation therapy, imaging, and computational physics. In the field of ultrasonic imaging, phantoms are used to assess the accuracy and quality of images produced by the equipment, and to identify any issues that may affect the accuracy of these images. The development of ultrasound phantoms has seen contributions from various sources including research groups and commercial companies. There are several types of ultrasonic imaging phantoms, each designed to test specific aspects of ultrasonic imaging technology, including simple reflector phantoms, anthropomorphic phantoms, and phantoms for evaluating properties such as sound speed, attenuation, and backscatter. With the continued growth of the global ultrasound equipment market, the importance of using tissue mimicking materials for testing and evaluating these devices will continue to increase. The following recommendation can be made as a result of this study. For the qualitative and quantitative evaluation of the acoustic parameters of ultrasonic imaging phantoms, setting special measurement targets for making dimensional measurements inside the phantom in future phantom designs and specifying these targets in the specifications will facilitate the metrologists who will make the measurements.

130

B. Karaböce and H. O. Durmu¸s

4 Conclusions In this article, after well-defining the theoretical background of acoustical characterisation of ultrasonic imaging phantoms, acoustical parameter measurements as sound velocity and attenuation coefficient were carried out for two different model ultrasonic imaging phantoms. Medical measurements play a crucial role in the practice of medicine. They help healthcare providers accurately assess a patient’s health, diagnose medical conditions, and monitor the effectiveness of treatment. In this context, ultrasonic imaging phantoms are used to calibrate and test the performance of ultrasonic imaging systems, ensuring that the images produced are accurate and of high quality. In addition to this, ultrasonic imaging is a widely used technology for medical diagnostic applications. The accuracy and efficiency of this technology depends on its correct testing. In conclusion, ultrasonic imaging phantoms play a critical role in ensuring the accuracy and quality of ultrasonic imaging equipment. By providing a controlled environment for testing and evaluating ultrasonic imaging systems, phantoms help medical professionals identify any issues with the equipment and ensure that images produced during patient examinations are of the highest quality and accuracy. The use of phantoms, despite having some limitations and difficulties, is crucial for ultrasonic imaging technology to be utilized effectively. This is because the advantages of using phantoms surpass the drawbacks and make them a necessary tool.

References 1. DeWerd, L.A.: The phantoms of medical and health physics (pp. 127–9). In: Kissick, M. (ed.). Springer, Berlin (2014) 2. Insana, M.F.: Ultrasonic imaging. Wiley encyclopedia of biomedical engineering (2006) 3. Bell III, F.E., Haddad, R.: The basics of ultrasound physics. In Understanding physiology with ultrasound (pp. 11–57). Springer US, New York, NY (2023) 4. Carovac, A., Smajlovic, F., Junuzovic, D.: Application of ultrasound in medicine. Acta Inform Med. 19(3), 168–171 (2011). https://doi.org/10.5455/aim.2011.19.168-171.PMID: 23408755;PMCID:PMC3564184 5. Zheng, Z., Su, T., Wang, Y., et al.: A novel ultrasound image diagnostic method for thyroid nodules. Sci. Rep. 13, 1654 (2023). https://doi.org/10.1038/s41598-023-28932-2 6. Zhao, J., Zhai, H., Liu, X., Song, J., Wang, S.H., Yang, S.: A novel decision support system for capsule segmentation via high frequency ultrasound images. Available at SSRN 4328050 7. Laschke, M.W., Körbel, C., Rudzitis-Auth, J., Gashaw, I., Reinhardt, M., Hauff, P., Zollner, T.M., Menger, M.D.: High-resolution ultrasound imaging: a novel technique for the noninvasive in vivo analysis of endometriotic lesion and cyst formation in small animal models. Am. J. Pathol. 176(2), 585–93 (2010). https://doi.org/10.2353/ajpath.2010.090617. Epub 2009 Dec 30. PMID: 20042678; PMCID: PMC2808067 8. Ommen, M.L., Schou, M., Beers, C., Jensen, J.A., Larsen, N.B., Thomsen, E.V.: 3D printed calibration micro-phantoms for super-resolution ultrasound imaging validation. Ultrasonics 114, 106353 (2021) 9. Badnjevi´c, A., Pokvi´c, L.G., Deumi´c, A., Be´cirovi´c, L.S.: Post-market surveillance of medical devices: a review. Technol. Health Care 30(6), 1315–1329 (2022) 10. Badnjevic, A.: Evidence-based maintenance of medical devices: current shortage and pathway towards solution. Technol. Health Care 31, 293–305 (2023)

Verification of Ultrasound Imaging Phantoms: An Evaluation Study

131

11. Badnjevi´c, A., Deumi´c, A., Ademovi´c, A., Pokvi´c, L.G.: A novel method for conformity assessment testing of therapeutic ultrasounds for post-market surveillance purposes. Technology and Health Care, (Preprint), 1–8 (2022) 12. Madsen, E.L., Zagzebski, J.A., Banjavie, R.A., Jutila, R.E.: Tissue mimicking materials for ultrasound phantoms. Med. Phys. 5, 391–394 (1978) 13. Burlew, M.M., Madsen, E.L., Zagzebski, J.A., Banjavic, R.A., Sum, S.W.: A new ultrasound tissue-equivalent material. Radiology 134, 517–520 (1980) 14. https://www.sunnuclear.com/uploads/documents/datasheets/Diagnostic/Doppler-Flow_P hantoms_113020.pdf 15. https://www.supertechx-ray.com/Ultrasound/QCPhantoms/ATS539.php 16. Commission, I.E.: Ultrasonics-Pulse-echo scanners-Low-echo sphere phantoms and method for performance testing of gray-scale medical ultrasound scanners applicable to a broad range of transducer types. IEC TS 62791, 2015 (2015) 17. CEI IEC 1390 Ultrasonics-real-time pulse-echo systems-test procedures to determine performance specifications

Clinical Engineering and Health Technology Assessment

Development of a Failure Prediction Strategy for Imaging Systems A Study from the Clinical Side of View Dario Léon Merten1(B) , Dubravka Maljevic1 , and Markus Buchgeister2 1 BG Kliniken – Klinikverbund der gesetzlichen Unfallversicherung gGmbH, Leipziger Platz 1,

10117 Berlin, Germany [email protected] 2 Berliner Hochschule für Technik, Luxemburger Straße 10, 13353 Berlin, Germany

Abstract. Predictive maintenance offers great added value for patient safety, availability and safeguarding the clinical process. Through the exchange of relevant data from networked medical devices, increasingly automated decisionmaking and early failure prediction are possible. Using the early condition assessment and the maintenance measures derived it is possible to react preventively to these forecasts within the maintenance process. This can lead to the avoidance of cost-intensive and risky system failures. Various methodological approaches of qualitative and quantitative origin were examined to develop the strategy. Inventory records of the BG hospital group and analyses of the development statuses of various manufacturers for predictive maintenance served as the basis for this. Based on the analysis of the data and various expert interviews, these approaches were evaluated and strategies were designed. The combination of a manufacturer’s solution with in-house measures turned out to be the most promising. The analysis and evaluation are carried out by the manufacturer and can then be used by the clinical engineers. Additional components of the generated error messages are measures for prevention or elimination. As a result, internal hospital expertise and an extended data basis for the evaluation of the system status are initially not absolutely necessary and a failure prediction can be incorporated into the clinical process. This enables an increase in own performance and improved transparency and efficiency. At present, failure prediction for the clinical side is therefore only possible in cooperation. Through additional measures such as the sensitisation of staff, training and the associated increase in specialist expertise, there is the possibility of reacting more quickly and precisely to failure predictions and of operating predictive maintenance partly in-house. Furthermore, with the help of the strategies developed, an optimisation of the existing maintenance strategy and working conditions within the hospitals can be achieved. Keywords: Clinical engineering · Evidence-based · Medical device · Maintenance · Management · Optimisation · Prediction · Strategy hospitals development

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 135–141, 2024. https://doi.org/10.1007/978-3-031-49068-2_15

136

D. L. Merten et al.

1 Introduction Medical technology is one of the most innovative industries and it is impossible to imagine modern healthcare without it [1]. The development of imaging systems in particular enables ever faster and more precise diagnosis and therapy of diseases. The flip side of technical innovation is the increasing complexity of the systems, which is becoming more and more difficult for clinical engineers to master, as a high degree of specific expertise is required to detect malfunctions at an early stage and remedy them immediately [2]. In the course of the digitalisation and Industry 4.0 that has taken place in recent years, Prognostics and Health Management have increasingly become the focus for the maintenance of medical devices. The focus of this special discipline is on the condition assessment of technical devices and forecasts regarding their further condition development. Industry 4.0 devices are characterised by their networking and data exchange and enable new options for maintenance due to the transmission of relevant data. Suitable maintenance measures can be taken to react to these forecasts at an early stage in order to maintain the functional condition and avoid cost-intensive and risky system failures [3, 4]. A study on the topic of sustainable maintenance showed that corrective maintenance oriented towards failure is predominant despite the progress of digitalisation [5]. It follows that, despite the progress in innovation, there are more and more unplannable failures that impair the maintenance goals of sustainability, availability and patient safety in the clinical context [4]. The expansion of the preventive maintenance strategy is therefore the goal of future-oriented maintenance as well as the quality standards of BG hospital group. The objective of this study was to evaluate whether failure prediction for imaging systems is feasible on the part of the clinical side and how strategic implementation can take place in the existing maintenance of the BG hospitals. In doing so, it was to be considered which possibilities and advantages arise within the framework of Industry 4.0 for the maintenance of clinical systems and clinical processes. The initial hypothesis of the study was that the networking and provision of log files and system messages of the medical technology systems lead to a strong gain in information regarding their condition and that a need for timely action for the clinical side can be defined from this. As a result, an increase in the preventive part of maintenance as well as a continuous orientation of the maintenance strategy towards the growing digitalisation can be expected. For this purpose, close cooperation with manufacturers and operators was carried out within the scope of this study.

2 Methods The development of a suitable strategy for the integration of predictive maintenance was based on an empirical research methodology. For this purpose, an inventory and process analysis was first carried out at the BG hospitals and manufacturers. This represented the current status and included the representation of the equipment park of the imaging systems as well as the recording of the maintenance processes. With the help of the quantified equipment pool, one type of equipment was extracted and used representatively for the further process. The selection was based on criteria

Development of a Failure Prediction Strategy for Imaging Systems

137

that take into account economic aspects and maintenance objectives. The processes of the BG hospitals and the manufacturers were elicited and analysed with the help of a qualitative analysis in the medical technology department. The aim was to ensure that existing process elements that could be assigned to predictive maintenance were also considered in the strategy development and that new process elements were added if necessary. In particular, the exchange of information with the manufacturers served to consider further possibilities for implementation within the hospital group. Building on the data analysis and assessment, the strategies were designed. The three approaches to data collection used for this included the use of: (1) internal company data, which were based on the one hand on documented activities of the clinical engineers in the computer-aided facility management (CAFM) software and on the other hand on internally available documents of the services provided by the manufacturers. (2) device-side data of the representatively selected device type. This data included occurring system messages as well as measured values of the sensor technology. To include this data in the study, there was a close exchange with the manufacturers and cooperation with the internal IT department. (3) manufacturer’s solutions to consider possible combinations with the in-house medical technology department and the use of these within the clinical environment.

3 Results and Discussion Based on the inventory, the analysis of the manufacturers and the three approaches, different strategies for the clinical side of view could be extracted. 3.1 Inventory and Data Acquisition The inventory analysis revealed 520 devices in the category of imaging systems. Due to the importance for emergency care, the cost-intensive maintenance and the high degree of digitalisation with corresponding data density, the computer tomograph (CT) was selected as the representative device type. Further selection criteria were ensuring the availability of the equipment and patient safety. The process mapping identified three processes from the activity areas of corrective and preventive maintenance. Corrective maintenance was divided into internal and external fault message processing. Preventive maintenance was subject to the process of fault prevention. The external processing of fault reports outweighed the internal processing, regardless of the type of maintenance. Due to the great importance for patient care, ensuring safe and reliable operation is crucial from the clinical side of view. Downtime is minimised by securing the systems with a full-service contract. The full-service contract always includes automatic data transmission for monitoring the system status. On the one hand, this has the advantage that the data is sent to the manufacturer at an early stage so that the manufacturer can react in time, and on the other hand, it has the disadvantage that the expertise and authority of the clinical engineers on site are kept low. An application of predictive maintenance could not be observed at any of the sites.

138

D. L. Merten et al.

The manufacturer defines error patterns with the help of occurring system messages in combination with technical expertise. The system messages served as a data basis and amounted to approx. 10,000 messages per day for a CT. The recognition of these error patterns was done retrospectively by defining specific trigger points. These trigger points show the repeated occurrence of a system message within a defined time interval in combination with other predefined system messages. In some cases, parameter deviations and trends were also taken into account. The evaluation was partly carried out via a datasupported AI. In both cases, a specific message was created, validated and made available to the clinical side. This message is shown in Fig. 1.

Fig. 1. Structure and content of a predictive message

The approaches listed in Sect. 2 were evaluated based on these findings and showed heterogeneous benefits for strategy development. An analysis of the clinical-internal data for approach (1) was not possible due to the quality of the data. The documented activities as well as the descriptions of failures were insufficient. The data of the CTs for approach (2) were inaccessible. On the one hand, this data was not made available by the manufacturer and on the other hand, it was not possible to access the data from the CT. A high level of device-specific knowledge is required to view or export this data. The manufacturers are very anxious to develop appropriate models from the data, which will give them a competitive advantage in the market. Therefore, only a very small group of people has access to the data on the equipment side. From various interviews, it became clear that the manufacturers are working on the development of predictive maintenance. The remaining approach (3), in which the analysis and evaluation are carried out by the manufacturers, therefore seemed more promising. The messages generated in this way proved to be a useful option and can be presented in a web portal or through software implementation in the CAFM system. The messages include both error descriptions and solution steps for rectification. This represents a strategic approach that partly enables internal failure prediction and can be integrated into the daily clinical routine.

Development of a Failure Prediction Strategy for Imaging Systems

139

3.2 Strategy Development The error messages generated by the manufacturer can be used by the clinical side with approach (3) and offer a possibility to realise predictive maintenance from them. The basis for this is the automatic bidirectional synchronisation between the manufacturer and the CAFM system as well as order transfer for the clinical side. The manufacturer’s predictive error messages can be continuously displayed in the CAFM system. Further advantages for maintenance are simplified order creation, improved transparency and reduction of documentation gaps. If the web portal were used exclusively, the clinical engineers would have to transfer changes from the web portal to the CAFM system. The automatic transfer of orders enables more efficient preparation for upcoming maintenance measures. The timely provision and planning of required resources as well as the early consultation with the clinical department should be emphasised here. With the help of this, the maintenance measure can be planned and integrated into the daily clinical routine. The use of error messages in combination with internal company measures resulted in three strategies for bringing predictive maintenance into the BG hospitals. These are shown in Fig. 2.

Fig. 2. Variants of the strategy developed for the BG hospitals

This study showed that the specific expertise of the clinical engineers was not sufficient to process a proactive report. The training of the clinical engineers is therefore essential in order to be able to use the error messages. Therefore, it is recommended to train the clinical engineers to the level of a manufacturer’s service technician. In addition, the automatically generated error message can be used as a clue for unknown errors. The clinical engineers should also be sensitised to detailed error descriptions. This also applies to the users to additionally promote data quality. The resulting better identifiable error pattern helps to plan maintenance measures more efficiently and to recognise error patterns in the future. Sensitisation can be achieved with the help of regular instruction by clinical engineers to minimise user errors.

140

D. L. Merten et al.

With an appropriate level of awareness, the equipment log data can be used to determine own error patterns, which have been recognised due to the increasing experience. Likewise, a better data scope of internal data would be achieved due to the sensitisations. In combination, this could lead to the competence of own failure prediction in the long run. A mix of internal and external order processing or exclusively external processing would also be conceivable. If the service is provided exclusively externally, no further measures would have to be taken. The implementation would nevertheless provide the advantages mentioned at the beginning, which in turn would lead to better efficiency and transparency. In the case of a mix of external and internal implementation, the respective areas of responsibility must be defined. In this case, too, the clinical engineers should be trained. Here, however, basic equipment training is sufficient. In future, orders with defined solutions would be processed internally, as would clear error patterns or preventive measures. Orders for more complex faults would then be processed by the manufacturer’s service department. The prerequisite for this, however, is a definition of the complexity of the error. The basic prerequisite is a continuous connection to the manufacturer’s service. Monitoring of the interface should be carried out by the manufacturer to ensure data transmission.

4 Conclusion In this work, a strategy for the clinical-side implementation of predictive maintenance was developed. The high demands on reliable failure prediction quickly led to the limits on the clinical side. Identifying faults before they occur and proactively counteracting them cannot be implemented without constant support from the manufacturers. For this, a high level of device-specific expertise and open access to device data are indispensable. It is also clear that for the manufacturer this data is important to the company. It is used for service planning as well as for the development of new equipment and service products, and it reveals a lot about how the system works. It is therefore understandable that manufacturers are not interested in disclosing this internal information. Analyses and evaluations by clinical engineers are not planned at this time. With the help of the developed strategy, which is a combination of a manufacturer’s solution and internal measures, a failure prediction can be carried out as far as possible. It is possible to work in cooperation with the manufacturer and reduce the dependence on the manufacturer. Complete manufacturer independence can be ruled out due to the shortage of skilled workers and the increasing requirements. With the help of the specific expertise that has been built up accordingly, decisions and questions can be analysed and made independently. Staff training and awareness-raising are important in this respect. These measures are aimed at increasing in-house performance so that the outsourcing of services is reduced. The additional functions in the CAFM system also improve working conditions.

Development of a Failure Prediction Strategy for Imaging Systems

141

References 1. VDI Verein Deutscher Ingenieure e.V.: Medizintechnik–trends und perspektiven. VDIGesellschaft Technologies of Sciences, Berlin (2017) 2. Siebold, N., Züchner, K.: Medizintechnischer Service. In: Debatin, J.F., Schulte, B., Ekkernkamp, A., Tecklenburg, A. (eds.) Krankenhausmanagement–Strategien, Konzepte, Methoden, pp. 723–729. MWV Medizinisch Wissenschaftliche Verlagsgesellschaft Berlin, Berlin (2013) 3. Mildner R., Meyer J.-U., Eckardt N., Hartung L, et al.: Medizintechnik 4.0Interoperabilität/Vernetzung, Assistenzsysteme, Wartung & Service, Usability. In: Krankenhaus 4.0, pp. 14–23. UniTransferKlinik Lübeck GmbH, Lübeck (2017) 4. VDI Verein Deutscher Ingenieure e.V.: VDI-Statusreport–prognostics and health management. VDI-Gesellschaft Produkt- und Prozessgestaltung, Berlin (2022) 5. Kuhn, A., Schuh, G., Stahl, B.: Nachhaltige Instandhaltung–Trends. Potenziale und Handlungsfelder Nachhaltiger Instandhaltung. VDMA Verlag GmbH, Frankfurt (2006)

OHIO - Odin Hospital Indoor Compass for Empowering the Management of Hospitals Alessio Luschi(B) and Ernesto Iadanza Department of Medical Biotechnologies, University of Siena, Via A. Moro, 2, 53100 Siena, Italy [email protected], [email protected]

Abstract. OHIO is a project that received funding from the European Union’s Horizon 2020 research and innovation action programme, via the ODIN – Open Call issued and executed under the ODIN project (GA 101017331), focused on the enhancement of hospital safety, productivity, and quality. The main objective of OHIO is to provide a solution to integrate the ODIN Platform with an informative system able to empower the management of hospital facilities in terms of clinical engineering, logistics, and disaster preparedness. It will enhance the existing Computer Aided Facility Management system and Indoor Positioning System mobile application, improving the process of maintenance of medical equipment, streamlining logistics, and supporting the top management in designing effective responses to disasters. OHIO will also be fully integrated within the ODIN Platform exploiting the offered services and features. Keywords: Healthcare · IoT · Clinical engineering · Logistics · Disaster preparedness

1 Introduction The hospital environment is evolving along with technology growth, becoming more and more complex. Today’s healthcare facilities use a variety of tools and technologies to boost the effectiveness and efficiency of medical treatments [1], reduce barriers to accessibility and collaboration, improve safety [2], productivity, and quality of the working environment [3], all while maintaining cost effectiveness. This scenario introduces digital solutions that support resources and services, including the Internet of Things (IoT), robotics, mobile apps [4], sensors, and Artificial Intelligence (AI), which is playing a larger role in nearly all aspects of healthcare. By removing demanding and time-consuming tasks from workers’ workloads, these technologies also strive to make everyday jobs better. Robots and AI significantly contribute to helping human workers perform more effectively in stressful situations by taking over jobs for which their presence is not necessary [5]. Clinical workflows, medical locations, and logistics benefit from the introduction of cutting-edge technologies, their integration into the healthcare environment, and their contact with patients and the local community [6].

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 142–149, 2024. https://doi.org/10.1007/978-3-031-49068-2_16

OHIO - Odin Hospital Indoor Compass for Empowering

143

1.1 The OHIO Project OHIO (Odin Hospital Indoor cOmpass) is a project that received funding from the European Union’s Horizon 2020 research and innovation action programme [7], via the ODIN – Open Call issued and executed under the ODIN project (GA 101017331) [8]. The main objective of the OHIO project, lead by the Department of Medical Biotechnologies of the University of Siena, is to leverage the hospital’s inner informative resources (data, architectures, and legacy applications) and the digital platform provided by ODIN to enhance hospital safety, productivity, and quality by empowering the management of a large-scale pilot (hospital “Le Scotte” in Siena, Italy) in terms of clinical engineering, logistics, and disaster preparedness. 1.2 Unmet Needs Scheduled maintenance and corrective maintenance interventions require the technical staff to access the internal spaces of the departments to find the device they are looking for. The maintenance intervention could fail for several reasons: the device is in use by the healthcare staff at the moment the technician arrives, it cannot be found or it has been loaned to another department, intense health activity makes it difficult for the healthcare staff to interact with the technical staff, etc. There is a need for a wellorganized process, where the devices to be inspected/repaired are made available in a hot-spot collecting area in the department. Besides, hospitals are characterized by high levels of physical activity, with a constant stream of temporary visitors (patients and related visitors), workers, and mobile technical equipment operating in various areas. All large hospitals are subject to continuous rearrangements and restructurings, physiological for complex healthcare structures. They are also served by a multitude of external suppliers and maintainers who often are lacking adequate knowledge of the latest indoor layout of departments, making it difficult to reach their destination without any indoor navigation system. Moreover, effective planning of disaster management needs tools for validating the solutions proposed for the different disasters, which impact the internal routes, changing the time to go from A to B, modifying the proximity of services, etc. A mobile app able to track down routes and timings for ex-post analysis is needed. Indoor wayfinding in hospitals is currently entrusted largely to vertical and horizontal signage. These static signals, sometimes deficient or hard to read, are mainly directed to patients and users, not to technicians. On the other side, disaster preparedness plans are currently “open loop”, meaning that there is a lack of feedback from simulations, to assess the efficacy of the planned solutions for the various catastrophic scenarios, for example by measuring the effects of the reallocation of services in different wards (or buildings), in terms of time changes for equipment management and logistics. In clinical engineering, there is generally no agreed “filter zone” in the wards where the devices to be maintained are placed on the day of the intervention. The technician must interfere with healthcare processes to access the devices. Current Indoor Positioning System (IPS) solutions fail to address the problems of aided logistic support, clinical engineering, being mostly directed to patients and visitors, and lacking connection/interaction with a central platform [9]. The aim of the OHIO project is to actively respond to this set of unmet needs by providing an actual solution to enhance the existing hospital’s Computer

144

A. Luschi and E. Iadanza

Aided Facility Management (CAFM) system named SPOT [10–12], and an IPS mobile application named HiWAY [13, 14]. This will improve the process of maintenance of medical equipment, streamline logistics, and support the top management in designing effective responses to disasters.

2 Materials and Methods The Azienda Ospedaliero-Universitaria Senese (AOUS) in Italy, is equipped with an internally developed custom CAFM informative system, at the pilot hospital “Le Scotte” of Siena, which can output Key Performance Indicators (KPIs) and quantitative parameters about the analysed healthcare facility. The designed system is a self-sufficient application able to manage and analyse digital plans of hospital buildings. It maps the hospital’s inner organisation, destinations of use and environmental comforts providing quantitative, qualitative, and graphical reports. The core database is a Microsoft SQL Server instance linked to other existing hospital databases so that the system can act as a central control cockpit. The suite consists of a main software module named SPOT and extra tools which all refer to the same inner database linked to the Hospital Information System (Fig. 1). A stand-alone main application monitors the status quo of the buildings in terms of beds, square meters, the destination of use, functional areas, and much other info for every room. It currently maps 8,375 rooms in 13 buildings, 155,314 square meters, 43 departments, 220 operative units, and 689 beds. It also maps 3,071 employees and a total of 270,772 assets (44,353 of which are electromedical equipment). Other modules include a custom web search engine, a document manager, and a mobile app for indoor navigation named HiWAY, exploiting the geolocation capabilities offered by an existing third-party indoor navigation framework for determining and displaying the user’s position. The application offers two different navigation modes, depending on the actual position of the device: on- site and off-site. In on-site mode, users are guided to the desired destination, retrieved via SPOT thanks to dedicated Application Programming Interfaces (APIs). Off-site mode is used for path planning. Multiple destinations can be added to obtain complex route maps, as shown in Fig. 2. HiWAY authenticates itself on SPOT using Basic Authentication security protocol, sending a Base64- encoded API key within the request header. This login method avoids users to personally sign-in to the system, in accordance with privacy principles, while assuring a virtual identity confirmation and authorization. SPOT includes a web module for document management. Users can upload different types of technical and administrative documents (images, text files, compressed archives, etc.) directly linked to physical hospital areas (buildings, storeys, or rooms), to be retrieved by authenticated end-users. In OHIO new connectors to the ODIN Platform will be designed and realized, together with new backend modules to organize the maintenance interventions for clinical engineering and to analyse disaster preparedness scenarios and simulations. A new W3C-compliant R2RML (RDB to RDF Mapping Language) connector will also be created, connecting the existing database to the ODIN ontology [15] for enabling access via SPARQL queries. The pilot hospital will be provided with Bluetooth Low Energy (BLE) beacons, exploited by HiWAY for improving positioning accuracy. All the new tools will be registered and integrated into the platform by OHIO as detailed in the next section.

OHIO - Odin Hospital Indoor Compass for Empowering

145

Fig. 1. A screen capture from SPOT.

Fig. 2. HiWAY in off-site mode (left) allows saving custom routes for planning scopes, before reaching the premises. Routes can then be loaded once arrived on site and used to get directions in real-time (middle and right).

The maintenance and corrective maintenance interventions will be scheduled, thanks to a dedicated backend module, by knowing the exact location of devices via SPOT. Specific hot-spot collecting areas will be defined in the pilot hospital wards, where the (movable) devices will be handed over to the technician, thus preventing rescheduling due to missing equipment. Selected devices will be RFID-tagged (Radio-Frequency IDentification), so that the technician will be able to query the platform to know when they are available for him, thanks to the HiWAY IPS. These filter areas internal to the wards will greatly speed up the clinical engineering process, by maximizing the number of successful interventions, while minimizing the interferences with the healthcare operations. Moreover,

146

A. Luschi and E. Iadanza

clinical engineering will benefit from gaining access to provide technical documentation and reports directly on HiWAY, via newly designed Web-APIs which will interface the mobile app to SPOT Document Manager. This feature, together with the possibility to pre-plan both scheduled and corrective maintenance interventions will ease and optimize managing clinical engineering tasks. External suppliers will also benefit from the solution to streamline the logistics: the proposed solution will help both internal and external users to easily reach a destination inside the hospital by knowing any of the possible inputs (points of interest, employee’s names, clinical areas or departments, room codes). Finally, OHIO will also benefit disaster preparedness, by allowing the pilot managers to use SPOT and HiWAY to create simulated scenarios for different disasters and to analyse their impact on routes and timings. Users provided with smartphones will travel routes programmed in the disaster response planning phase, following navigation instructions to simulate different degrees of locomotion disability and assistances (crutches, walking sticks, rollators, wheelchairs). The collected information, routes and timings, will be sources for accurate planning of response to the disaster. The outcomes will be captured in a quantitative way to implement an effective data-driven Evidence Based approach.

3 Integration and Deployment OHIO will register all the currently available resources in SPOT as Key Enabling Resources (KERs) of the local ODIN Platform instance (Fig. 3). Specifically, DWG building floorplans stored in SPOT data storage, as well as files about equipment, personnel and organizations, and technical documents stored on the SPOT Document Manager, will be mapped to the Resource Gateway Layer by developing ad-hoc connectors, which will perform both semantic translations to the ODIN Ontology and syntactic transport across the Enterprise Service Bus (ESB). Three connectors for the SPOT frontend web application, HiWAY mobile application, and SPOT Database Management System will also be developed within the solution provided by OHIO. BLE beacons needed for indoor navigation, together with RFID tags applied on medical devices, will also be available KERs mapped to the ESB via specific connectors through the IoT gateway. All the Web-APIs for external applications will also be connected to the ESB. Among these KERs, the ones exposing APIs to external applications or services will be made visible through the API gateway, to provide a common point of access and authorization. Advanced and specialised features will be available for authorized logins provided with the open-source software Keycloak, to allow Single Sign-On (SSO) with Identity and Access Management through Lightweight Directory Access Protocol (LDAP) integration with the hospital’s personnel database. All the aforementioned KERs will be registered to the ODIN Platform using the Resource Manager and aligned to the ODIN Ontology (Fig. 4) within the OHIO project, providing the desired level of abstraction. Information about the healthcare facility, buildings, storeys, and rooms will be mapped through the Building Topology Ontology, the hospital’s inner organization (departments and operative units) through the Organization Ontology, employees through the NCIT Ontology (for medical and allied health occupation) and the FOAF ontology (for the remaining non-medical personnel). BLE beacons and RFID tags will be mapped through the Web of Things Ontology, while data on medical devices will be mapped through the

OHIO - Odin Hospital Indoor Compass for Empowering

147

Fig. 3. Schema of the ODIN platform.

OdinEMDN ontology. An expansion of the ODIN On-tology with the Dublin Core Ontology (DC), and the LoSeODP ontology, will be proposed and implemented to integrate also digital documents and assets. The former will be used for creating abstractions for digital documents, while the latter will be used for assets and logistic information. Finally, OHIO will ensure that the already existing functionalities of SPOT and HiWAY benefit from the integration with the ODIN Platform services, in particular with the Metric Collection and the Resource Choreographer (RC) components. The former will help retrieve metrics for monitoring Web-APIs via the Prometheus monitoring solution, while the latter will improve the communications and the automation among SPOT modules and services (e.g., an architectural or structural change will trigger a warning for a possible repositioning of nearby BLE beacons for indoor navigation; edits on departments or operative units will trigger new events for updating rooms information; malfunctions on a generic device connected to the IoT gateway will trigger an alarm for maintenance request). OHIO will be tested at the pilot site of the hospital “Le Scotte” in Siena, Italy. The proposed system will take charge of the whole hospital, consisting of more than 8,000 rooms, 155,000 sqm, and a technological park of more than 44,000 pieces of biomedical equipment. The hospital is a reference for about 900,000 citizens in south-east Tuscany, counting more than 3,000 employees, and dispensing more than 21,000 ordinary hospitalizations per year and about 3,200,000 outpatient services per year. Such a similarity in terms of beds, number of hospitalizations per year, outpatients, and employees, will facilitate the technology transfer to some of ODIN’s large-scale pilots (San Carlos Clinical Hospital in Madrid, Spain, University Campus Bio-Medico

148

A. Luschi and E. Iadanza

Fig. 4. The ODIN ontology.

in Rome, Italy, and Medical University of Lodz, Poland). Since the proposed solution can be scaled to any needs, even if an applicant pilot has already its own CAFM system, the disaster preparedness backend module itself (paired with HiWAY) can provide the needed insights for analysis.

4 Conclusions Today’s hospitals are complex, articulated structures providing a wide range of services to both in-patients and out-patients. Both old and new specialities make extensive use of medical technologies that necessitate ongoing intervention for management and maintenance. In Europe, this rapid evolution has not always been accompanied by an equally rapid adaptation of hospital structures. Hospitals are frequently dated structures organized in pavilions, making it difficult for users, particularly patients, to navigate these structures. The main objective of the OHIO project is to empower the management of a large-scale hospital pilot in terms of clinical engineering, logistics, and disaster preparedness, enhancing the existing Computer Aided Facility Management system (SPOT) and Indoor Positioning System mobile application (HiWAY). Once OHIO will be fully deployed, it will deliver concrete value to the ODIN Project, which will benefit from an extended ontology, new APIs and IoT connectors, and will take advantage of the OHIO’s experience in integrating existing tools and databases into the ODIN Platform. At the same time, the proposed solution will impact the large-scale pilots by transferring to them the experience in a large hospital which is already provided with a high level of technological IoT tools, databases, CAFM software (SPOT) and an IPS mobile app (HiWAY) managing more than 155,000 square meters. The authors think that OHIO will have a vast impact on the population, since the pilot is a reference centre for about 900,000 inhabitants, with a tech park of about 100 Me. Improving the logistics, the

OHIO - Odin Hospital Indoor Compass for Empowering

149

device maintenance, and the preparedness for disasters in such a large case study will be of paramount impact on the whole Tuscany region. The outcomes and the impact of the project will be measured thanks to a dedicated set of KPIs based on the analysed data. The output will also be measured in a semi-quantitative way by applying standard usability techniques.

References 1. Luschi, A., Monti, M., Iadanza, E.: Assisted reproductive technology center design with quality function deployment approach. In: IFMBE Proceedings, vol. 51, pp. 1587–1590 (2015). https://doi.org/10.1007/978-3-319-19387-8386 2. Iadanza, E., Luschi, A., Merli, T., Terzaghi, F.: Navigation algorithm for the evacuation of hospitalized patients. In: IFMBE Proceedings, vol. 68, pp. 317–320 (2019) 3. Iadanza, E., Ottaviani, L., Guidi, G., Luschi, A., Terzaghi, F.: License: web application for monitoring and controlling hospitals’ status with respect to legislative standards. In: IFMBE Proceedings, vol. 41, pp. 1887–1890 (2014) 4. Luschi, A., Belardinelli, A., Marzi, L., Frosini, F., Miniati, R., Iadanza, E.: Careggi smart hospital: a mobile app for patients, citizens and healthcare staff. In: 2014 IEEE-EMBS International Conference on Biomedical and Health Informatics, BHI 2014, pp. 125–128 (2014) 5. Pradhan, B., Bharti, D., Chakravarty, S., Ray, S.S., Voinova, V., Bonartsev, A., Pal, K.: Internet of things and robotics in transforming current-day healthcare services. J. Healthc. Eng. pp. 1– 15 (2021) 6. Iadanza, E., Turillazzi, B., Terzaghi, F., Marzi, L., Giuntini, A., Sebastian, R.: The streamer european project. case study: Careggi hospital in florence. In: IFMBE Proceedings, vol. 45, pp. 649–652 (2015) 7. Commission, E.: Leveraging ai based technology to transform the future of health care delivery in leading hospitals in europe. https://cordis.europa.eu/project/id/ 101017331 (2021). Accessed 11 July 2022 8. Odin: Odin-smart hospitals. https://odin-smarthospitals.eu/ (2021). Accessed 11 July 2022 9. Kunhoth, J., Karkar, A., Al-ma’adeed, S., Al-Ali, A.: Indoor positioning and wayfinding systems: a survey. Human-Centric Comput. Inf. Sci. 10 (2020) 10. Luschi, A., Miniati, R., Iadanza, E.: A web based integrated healthcare facility management system. In: IFMBE Proceedings, vol. 45, pp. 633–636 (2015) 11. Iadanza, E., Luschi, A., Gusinu, R., Terzaghi, F.: Designing a healthcare computer aided facility management system: A new approach. In: IFMBE Proceedings, vol. 73, pp. 407–411 (2020) 12. Iadanza, E., Luschi, A.: An integrated custom decision-support computer aided facility management informative system for healthcare facilities and analysis. Heal. Technol. 10(1), pp. 135–145 (2020). https://doi.org/10.1007/s12553-019-00377-6 13. Falleri, N., Luschi, A., Gusinu, R., Terzaghi, F., Iadanza, E.: Designing an indoor real-time location system for healthcare facilities. In: Communications in Computer and Information Science, vol. 1343, pp. 110–125 (2021) 14. Luschi, A., Villa Borsani, E.A., Gherardelli, M., Iadanza, E.: Designing and developing a mobile application for indoor real-time positioning and navigation in healthcare facilities. Technol. Health Care 30(6), pp. 1371–1395 (2022) 15. Luschi, A., Petraccone, C., Fico, G., Pecchia, L., Iadanza, E.: Semantic ontologies for complex healthcare structures: a scoping review. IEEE Access 11, pp. 19228–19246 (2023)

Calculating the Required Mammography Machines for Breast Cancer Screening in Mexican States with High Incidence Rates: A Proposed Model Fabiola M. Martinez-Licona(B) Electrical Engineering Department, Universidad Autonoma Metropolitana Iztapalapa, Av. Ferrocarril San Rafael Atlixco, Núm. 186, Col. Leyes de Reforma 1 A Sección, Alcaldía Iztapalapa, C.P. 09310 Mexico City, Mexico [email protected]

Abstract. Breast cancer is a global health issue, in Mexico it accounts for 28.2% of new cancer cases in women. Early detection through mammography screening is a critical component of social programs aimed at potential high-risk female populations. The availability of mammography machines determines the potential coverage of a region. This paper proposes a model to determine the required number of mammography machines for the States of Colima, Mexico City, Durango, Morelos, and Oaxaca, which had the highest incidence rates for breast cancer in 2019–2021. The time period was chosen for the purpose of analyzing the effect of the pandemic on the burden of disease and the available resource. The model uses socioeconomic and demographic determinants, the screening mammography demand data obtained from official Mexican agencies. The selected model variables are from the target population, epidemiological, and infrastructure categories; this is an adapted model from a previous version applied to the southwest region. Through heuristic assumptions, the analytic hierarchy process is used to select the final set of variables. The Resource Allocation Working Party (RAWP) focus formula is adapted from population to equipment to calculate the Average State Weighted Machines using the selected categories. The findings indicate different scenarios for each State, with enhanced coverage compared to the reported data (an average of 55.71% for Durango, Morelos, and Oaxaca versus 36.21%). The study will continue examining additional regions, the model will be modified to include resource location, further investigation and analysis of factors affecting service’s access and equipment proposals are required. Keywords: Mammography machines · Breast cancer · Infrastructure resource allocation

1 Introduction Breast cancer is a significant health concern worldwide. The International Agency for Research on Cancer reports that 11.7% of new cancer cases globally in 2020 were breast cancer, and 28.2% of new cases were in Mexican women [1]. Early detection plays © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 150–161, 2024. https://doi.org/10.1007/978-3-031-49068-2_17

Calculating the Required Mammography Machines

151

a critical role in breast cancer management; education to promote self-care and early diagnosis and screening are the basis of the action plans towards getting a control level on the disease. Mammography, a radiological study of the breasts with low radiation doses, is a possible screening method used alone or with a physical examination [2]. Screening mammography is typically applied to asymptomatic women to detect tumors and microcalcifications that cannot be felt, potentially reducing the risk of death from breast cancer by around 40% in women aged 50–69 years [3]. This is possible when an organized population-based approach programs the implementation of the screening and allows an operational framework to manage actions and outcomes [4]. Despite their potential impact, screening programs face several challenges that limit breast cancer detection and timely treatment; some of them include the difficulty in accessing and using health services, the lack of clinical infrastructure and specialists needed to operate the equipment, and the restricted access to treatments [5]. The situation becomes more evident when it is analyzed from the coverage perspective. The national cancer control programs must strive to provide universal health coverage, particularly to vulnerable populations [6]. A crucial aspect of successful screening programs is to have safe and effective mammography equipment. Mexico lags behind other countries in the number of mammography machines per million inhabitants, with only 10.4 devices compared to Korea’s 63.4 and the United States’ 66.9 [7]. The acquisition and replacement of equipment to meet the population’s needs require a more specific analysis focusing on the social, demographic, epidemiological, and equipment contexts. Mexico is going through a stage of transformation of its health system concerning coverage; the current administration has implemented a subsystem whose objective is to provide timely services, including consultations, follow-up, treatments, and medications, to the population that, due to their employment status does not have access to health services [8]. The change from the overall scheme (Seguro Popular) to the current one (INSABI Welfare Institute) began to be implemented just before the emergence of the pandemic due to the SARS-Cov2 virus and the impact that this event has had on the health system in general, has been determined [9]. Regarding breast cancer, the consequences are shown in the difficulties in maintaining access and continuity in treatments, with COVID-19 being a negative determinant in this situation [10]. Mammography screening also suffered from problems during the pandemic; prevention services were postponed, reducing the cumulative number of screening mammograms and increasing the disparity in the vulnerable population with a direct consequence on mortality and care costs [11]. Statistics generally show an increasing trend in the incidence of malignant tumors in the breast and a decrease in the performance of mammography studies as a consequence of the pandemic contingency period. Mexico City, Guadalajara, and Monterrey are healthcare attractors for the people of the country’s center, western and northern. Their hospital infrastructure attends not only to the needs of its population but also that of nearby populations that seek specialized care. Although efforts to decentralize health services have had some success, people continue to go to the health facilities in this city. And although the demand for assistance tends to saturate the offer, there are still regions where the problem of locating resources

152

F. M. Martinez-Licona

is of greater importance. This paper shows an adapted model to determine the number of mammography machines required to meet the needs of the five states that have demonstrated the highest breast cancer incidence in 2019–2021 using contextualized information and principles of resource allocation.

2 Methods Identifying the current situation prevailing in the regions is necessary to determine the required equipment demand. An analysis was conducted on the five states with the highest breast cancer incidence previously and during the pandemic’s first two years, Colima, Mexico City, Durango, Morelos, and Oaxaca, considering socioeconomic and demographic factors, with a focus on the female population that is likely to use the screening mammography service. The demand for screening mammography studies was estimated by defining target populations using cohorts defined by WHO [12]. With this information, a model was developed that integrated components to contextualize the need based on the characteristics of the city. This model is adapted to a previous version applied to the southwest region [13] and will determine the number of mammography machines required. The process is described in detail below. 2.1 Breast Cancer Determinants The statistical information on each State’s socioeconomic and demographic determinants, from which I obtained the model variables, was carried out based on [14]. The data were subtracted from the official sites of Statistics and Geography [15], Social Development [16], and the Ministry of Health [17] in the period 2019–2021. After selecting the variables derived from the determinants, I decided on those with the most significant relevance concerning screening mammography to detect breast cancer. I defined the variables related to the profile of the target population, the equipment available in those years, and each State’s situation concerning breast cancer. Using Saaty’s analytic hierarchy process [18], I also incorporated heuristic assumptions to determine the following variables for each State [13]: • • • • • • •

Female population over 50 years old Women’s age who underwent mammography Number of mammography machines Number of medical facilities Mammography machine locations Mammogram outcomes Breast cancer morbidity rate

2.2 Screening Mammography Demand The Ministry of Health in each State provided data on the number of mammographies carried out and their outcomes for recent years. This information was used to complement the variables selected in the previous section, providing insight into breast cancer

Calculating the Required Mammography Machines

153

screening in Colima, Mexico City, Durango, Morelos, and Oaxaca. The data was specifically chosen for women aged 18 to 75 to align with WHO recommendations, but only 50 and over were included in the model. This is because most reports suggest that successfully organized breast cancer screening programs should target this age group [19]. The data on the mammograms developed in the period needed to be completed: more information in 2020 and 2021 regarding age and 2021 regarding the studies’ outputs. A linear trend model was applied for each State and country, considering the data reported since 2013. 2.3 Model The analytic hierarchy process, a multiple criteria decision-making process, was applied to define and test the variables used in the model, which were aimed at determining the required number of medical screening equipment in each State based on the target population, epidemiological indicators, and existing infrastructure. Based on the results of the hierarchical analysis, the variables with the highest weight were selected to propose a model for resource allocation based on the classic principles of population- weighted allocation, just as done in [13]. These principles consider age structure, health needs, and service delivery costs. The weighted population (WP) of an authority is then calculated from these factors as follows: WP = POP ∗ (1 + a) ∗ (1 + n) ∗ (1 + c)

(1)

where POP is the unweighted population, a, n, and c are age, needs, and cost adjustments, respectively [20]. The outcome displays the particular weight for each set of characteristics defined by the adjustments, which is then utilized as a decision-support tool. The model was customized to allocate mammography resources, and the following steps were undertaken. Average State’s weighted mammography capacity (ASWMC): The objective is to determine the required number of machines for the target population. The original model’s focus on the population was shifted to the State’s mammography capacity to serve as the basis for the calculation. The data collected in the study period (2019–2021) were averaged to obtain the adjustments and thus bring the final value. Adjustments: Adjustments were made to reflect the target population’s characteristics, infrastructure, and epidemiological data. The formula for the model is expressed as ASWMC = ASNWMC ∗ (1 + op) ∗ (1 + i) ∗ (1 + ei)

(2)

ASNWMC is the average unweighted State’s current mammography capacity, and op, i, and ei are objective adjustments for population, infrastructure, and epidemiologic information. The number of mammography machines per State was determined using ASWMC. From this information, the number of patients that the equipment and mammography coverage can treat was calculated as Pt = FP50 + /ASWMC

(3)

154

F. M. Martinez-Licona

where Pt represents the number of female 50+ patients that each medical equipment can treat. The % mammography coverage is represented by Mc = 6000/Pt ∗ 100

(4)

here, 6000 mammograms can be performed in a year, considering 20 minutes per studio, 8 hours a day, five days a week, and 50 labor weeks in a year [21].

3 Results The variables applied to the model were obtained for 2019, 2020, and 2021 and were grouped into the categories of the target population, infrastructure, and clinical information. The average value in the three years of each variable was used to calculate the final values and integrate them into the model. 3.1 Target Population The two variables related with this category include the female population aged 50 and above (50+) and the age of women who received mammography. The first variable was considered for women undergoing mammography screening, typically beginning after age 50. The values of the target population (FP50+) and the total State female population (FPTotal) were used to obtain the ratio between both data (PSt) for each State. The final values were obtained by calculating the proportion of the population 50+ of the State to the country’s ratio (Prop. FP50+ St/Co). The data corresponding to this variable are from the last population census carried out in 2020, so there was no averaging in this case. Table 1 shows the results. The average age of the women who underwent mammography and the relationship between the State average (AvSt), the country average (AvCo), were also computed, as presented in Table 2. The values for 2021 and 2021 are estimated with data taken from 2013 to 2019 since they have yet to be reported in the sources consulted. Table 1. Female population 50+. State

FP50+

FPTotal

PSt (50+/Total)

Prop. FP 50+ St/Co

Colima

66 271

370769

0.1787

1.0291

Mexico city

1 083 285

4 805 017

0.2254

1.2980

Durango

153 925

927 784

0.1659

0.9552

Morelos

193 707

1 020 673

0.1898

1.0927 0.9660

Oaxaca

361 944

2 157 305

0.1678

Mexico (Country)

11 209 778

64 540 634

0.1737

Calculating the Required Mammography Machines

155

Table 2. Women´s age who underwent mammography. State

2019

2020

2021

AvSt, AvCo

Av2019–2021

AvSt/AvCo (2019–2021)

Colima

51.07

50.30

50.50

50.62

0.9866

Mexico City

52.42

52.66

52.90

52.66

1.0263

Durango

51.23

51.43

51.65

51.44

1.0025

Morelos

51.52

51.30

51.45

51.42

1.0021 1.0066

Oaxaca

51.61

51.65

51.70

51.65

Mexico (Country)

51.45

51.18

51.30

51.31

3.2 Infrastructure This category pertains to the number of mammography machines and their locations. For each State and country, the number of machines (Mammography Machines MM) and the number of hospital institutions (Medical Facilities MF) were determined. The average equipment-to-facility ratio (MMAv/MFAv) was then calculated for each case, followed by the proportion of this result for each State to the country (MMP/MFP St/Co) as shown in Table 3. The location variable was assessed based on the number of municipalities in each State and the country (MunTot) for the 2020 population census. From this, the number of municipalities equipped with mammography machines (# MunMamm) was determined. The ratio of municipalities with mammograms to the total number of municipalities in each case (AvMunMamm (2019–2021) /MunTot) was then calculated. Furthermore, the proportion of this result in each State to the country (MunMamm/MunTot St/Co) was computed. Table 4 show the values obtained. 3.3 Epidemiological Indicators This category encompasses Mammogram outcomes and breast cancer morbidity rate. The proportion of mammogram studies with negative or benign neoplasia results in the target population (MB&N) was calculated from the reported data on mammograms performed for each State and country (MTot). The ratios (Av MB&N/MTot and St/Co) were then calculated. The Incidence of malignant tumor of the breast (IMTB) average values for women aged 50 and above were obtained for each State and country for 2019– 2021, and the proportional share for each State (Av IMTB St/Co) was calculated. Tables 5 and 6 display these values. 3.4 Number of Mammography Machines Equation 2 determined the number of mammographers per State by averaging the bold values from Tables 1, 2, 3, 4, 5 and 6 and plugging them in. The resulting average un-weighted State’s mammography capacity (ASNWMC) was calculated based on the

156

F. M. Martinez-Licona Table 3. Mammography machines and medical facilities.

State

2019

2020

2021

Mammography Machines MM

MMAv 2019–2021

MMAv/MFAv (2019–2021)

Colima

9

7

6

7

0.0405

Mexico City

142

108

106

119

0.1825

Durango

18

18

19

18

0.0342

Morelos

8

5

6

6

0.0193

Oaxaca

16

12

9

12

0.0077

Mexico (Country)

870

765

819

818

0.0373

MFAv 2019–2021

MMP/MFP St/Co

Medical Facilities MF Colima

173

174

173

173

1.0840

Mexico City

667

660

629

652

4.8895

Durango

525

529

525

526

0.9168

Morelos

304

310

319

311

0.5168

Oaxaca

1 590

1 559

1 546

1 565

0.2054

Mexico (Country)

22 220

21 859

21 662

21 914

Table 4. Mammography machine locations. State

2019

2020

2021

# MunMamm

MunTot 2020

AvMunMamm (2019-2021) /MunTot

MunMamm/MunTot St/Co

Colima

4

4

4

10

0.4000

3.8578

Mexico City

16

16

15

17

0.9412

9.0772

Durango

5

5

7

39

0.1538

1.4838

Morelos

5

4

5

36

0.1515

1.4613

Oaxaca

9

9

7

570

0.0140

0.1354

Mexico (Country)

271

252

244

2 476

0.1037

Specific Action Program for Breast Cancer 2007–2012, which proposes one machine per 60,000 inhabitants aged 50+ [22]. Table 7 presents ASWMC, the number of mammography machines calculated by the model. To obtain the number of patients per machine

Calculating the Required Mammography Machines

157

Table 5. Mammogram outcomes. State

2019

2020

2021

MB&N/MTot

Av 2019–2021 MB&N/MTot

MB&N/MTot St/Co

Colima

0.9114

0.9757

0.9984

0.9618

1.0693

Mexico City

0.9025

0.8613

0.8677

0.8772

0.9752

Durango

0.9143

0.9437

0.9177

0.9252

1.0286

Morelos

0.9040

0.9155

0.8993

0.9063

1.0076 1.0009

Oaxaca

0.9650

0.8175

0.9184

0.9003

Mexico (Country)

0.9057

0.8947

0.8982

0.8995

Table 6. Breast cancer morbidity. State

2019

2020

2021

IMTB

Av 2019–2021 IMTB

Av IMTB St/Co

Colima

372.07

379.83

396.61

382.84

0.1245

Mexico City

117.56

137.42

144.32

133.10

0.0433

Durango

159.69

98.04

118.98

125.57

0.0408

Morelos

806.10

510.21

421.74

579.35

0.1885

Oaxaca

132.32

82.79

130.38

115.16

0.0375

3 626.62

2 553.17

3042.97

3074.25

Mexico (Country)

(Pt) and the coverage percentage of the total computed machines (MC) equations 3 and 4 were applied. The results are shown in Table 8. Table 7. Number of mammography machines per state. State

Op

Inf

Ei

ASNWMC

ASWMC

Colima

1.0078

2.4709

0.5969

11.1287

12.29

Mexico City

1.1622

6.9834

0.5093

26.0518

470.36

Durango

0.9789

1.2003

0.5347

6.6822

17.14

Morelos

1.0474

0.9891

0.5980

6.5078

21.01

Oaxaca

0.9863

0.1704

0.5192

3.5317

21.30

158

F. M. Martinez-Licona Table 8. Mammography coverage.

State

Pt

Mc

Current Mc

Colima

5 522.58

108.64

63.37

Mexico City

2 304.86

260.32

65.91

Durango

9 054.41

66.27

70.16

Morelos

9 224.14

65.05

18.58

Oaxaca

17 235.43

34.81

19.89

4 Discussion Technology as a resource for the care of health problems represents a high-impact component in health systems. Medical devices in isolation or as part of a healthcare system must be cost-effective and safe to contend with the economic and social challenges involved in their introduction, implementation, and management. Breast cancer as a health problem is faced from different fronts; one is aligned to prevention, and screening programs for the target population are key elements. Although the mammography machine as a medical device represents “the state of the art” for this purpose, it is not exempt from controversy by the same target population and some other stakeholders regarding its effectiveness in practice, especially regarding coverage and monitoring. In the model, the female population aged 50 years and over was selected as the basis. This is because it is the age range where the impact of screening programs is most widely recognized, and studies have indicated that screening programs for populations aged 40+ and 65+ years are not cost-effective [23]. An official document from a past administration corresponding to a breast cancer care program was consulted to determine the number of mammography machines per State (ASWMC). This approach was taken to propose a scenario that was closer to an ideal situation. However, it should be noted that this value is kept from other organizations, such as the Pan-American Health Organization (PAHO), which proposed in 2016 to determine the number of machines based on every 100,000 females aged 50–69 years [24]. In contrast to other studies that correlated with the number of radiologists, the female population, or the surface area of the region under study [25], our analysis of the variables determining the number of mammography machines yielded different results. The variable related to morbidity was chosen to incorporate information specifically relevant to the present need for equipment, whether for screening or diagnosis. However, additional processing is necessary to account for these factors in the proposed model. The values entered into the model result from the average of the years consulted. In this sense, it is essential to note that the period before (2019) and the appearance of COVID-19, which monopolized health efforts and resources in 2020 and 2021, were addressed. The effect of the pandemic on the control of noncommunicable diseases is being revealed through reported morbidity and mortality statistics. For breast cancer, the impact is reflected in the economic aspect and timely access to diagnostic and treatment services [26]. The model presents limitations regarding incorporating this factor; some data needed to be reported, so they had to be approximated.

Calculating the Required Mammography Machines

159

On the other hand, the reported information represents only the public health coverage of clinical mammography services; It is known that throughout the country, there are networks of laboratories as well as private hospitals that offer these studies. Another relevant aspect is that the number of mammography machines reported does not necessarily imply that all the equipment is operating or working correctly. In this sense, the importance of having complete and reliable information to improve the estimates by the model is highlighted. The values in Table 8 show the effect of the variables on the outputs. The results show different scenarios for each State to the current coverage and that which would be achieved with the number of machines established by the model. Colima and Mexico City-States would have more excellent coverage than a study per year per individual. However, the number of devices presented is contrasting. It is worth mentioning that Colima is a State of minor proportions, and that Mexico City has a large metropolitan area and a more developed infrastructure. The case of Durango stands out, where the results showed a lower number of mammography machines required than reported and a decrease in coverage. From the results, differences in the courses of action can be inferred in each State concerning the technological resources provision. However, the number of devices does not guarantee better coverage for screening mammography, and further efforts to enhance the model are underway. The resource allocation model was modified to focus on medical equipment rather than population. While the original model must be considered, it is a helpful starting point for developing methods to determine resource allocation. However, this study has some limitations, such as the heuristic nature of certain variables, the quality of the data collected, and the assumptions made to adapt the model.

5 Conclusions A resource allocation model is proposed to determine the number of mammography machines needed to provide screening studies to the target population in the States of Colima, Mexico City, Durango, Morelos, and Oaxaca, which had the highest incidence rates for breast cancer in 2019–2021. The results reveal varying situations for each State, with improved coverage compared to reported data (55.71% average for Durango, Morelos, and Oaxaca versus 36.21%). Further study and analysis of the determinants of access to the service and equipment proposals are necessary. The research will continue for other regions, and modifications to the model will be proposed, including the integration of resource location. Conflict of Interest. “The authors declare that they have no conflict of interest.”

References 1. International Agency for Research on Cancer (IARC).: https://bit.ly/3sBfVsB, https://gco. iarc.fr/today/data/factsheets/populations/484-mexico-fact-sheets.pdf 2. National Cancer Institute at https://www.cancer.gov/types/breast/mammograms-fact-sheet

160

F. M. Martinez-Licona

3. Lauby-Secretan, B., et al.: Breast-cancer screening—viewpoint of the IARC Working Group. New England J. Med. (372.24), 2353–2358 (2015). https://doi.org/10.1056/NEJMsr1504363 4. von Karsa, L., Arrossi, S.: Development and implementation of guidelines for quality assurance in breast cancer screening–the European experience. Salud Publica Mex. 55, 318–328 (2013) 5. Duggan, C., Dvaladze, A., Rositch, A.F., et al.: The breast health global initiative 2018 global summit on improving breast healthcare through resource-stratified phased implementation: methods and overview. Cancer 126, 2339–2352 (2020). https://doi.org/10.1002/cncr.32891 6. National Council for the Evaluation of Social Development Policy (CONEVAL).: Municipal poverty measurement. https://bit.ly/3Mizo9n 7. Organization for Economic Co-operation and Development (OECD).: Mammography machines (indicator). https://data.oecd.org/healtheqt/mammography-machines.htm 8. Ministry of Health.: DECREE that amends, adds, and repeals various provisions of the General Health Law and the Law of the National Institutes of Health (2019). https://www.gob. mx/cms/uploads/attachment/file/521359/2019_11_29_MAT_salud.pdf 9. Doubova, S.V., Leslie, H.H., Kruk, M.E., Pérez-Cuevas, R., Arsenault, C.: Disruption in essential health services in Mexico during COVID-19: an interrupted time series analysis of health information system data. BMJ Glob. Health 6(9), e006204 (2021). https://doi.org/10. 1136/bmjgh-2021-006204 10. Villarreal-Garza, C., Aranda-Gutierrez, A., Ferrigno, A.S., Platas, A., Aloi-Timeus, I., MesaChavez, F., et al.: The challenges of breast cancer care in Mexico during health-care reforms and COVID-19. Lancet Oncol. 22(2), 170–171 (2021).https://doi.org/10.1016/S1470-204 5(20)30609-4 11. Sollozo-Dupont, I., Galván-Espinoza, H.A., Castillo-López, J.P., et al.: Impact of the Covid19 pandemic on breast cancer screening and how to act quickly and safely. Salud Publica Mexico 64(3), 333–339 (2022). https://www.medigraphic.com/pdfs/salpubmex/sal-2022/sal223l. pdf 12. World Health Organization.: WHO position paper on mammography screening (2014). https:// apps.who.int/iris/bitstream/handle/10665/137339/?sequence=1 13. Martinez-Licona, F.M., Martinez-Vazquez, C.M.: A proposed model for calculate the number of mammography machines for the South-West Area of Mexico. IX Latin American Congress on Biomedical Engineering Proceedings, Florianopolis Brazil (in press) 14. Agudelo Botero, M.: Sociodemographic determinants of access to breast cancer screening in Mexico: A review of national surveys. Salud Colectiva 9(1), 79–90 (2013). http://www.sci elo.org.ar/pdf/sc/v9n1/en_v9n1a07.pdf 15. National Institute of Statistics and Geography (INEGI). https://www.inegi.org.mx/programas/ccpv/2020/ 16. National Council for the Evaluation of Social Development Policy (CONEVAL),.https://www. coneval.org.mx/Paginas/principal.aspx 17. Ministry of Health. https://datos.gob.mx/busca/dataset/recursos-en-salud-nivel-central 18. Saaty, T.L.: Decision making with the analytic hierarchy process. Int. J. Serv. Sci. 1(1), 83–98 (2008). https://doi.org/10.1504/IJSSci.2008.01759 19. National Center for Gender Equity and Reproductive Health.: Modelo para la detección, diagnóstico y referencia del cáncer de mama (2011). Secretaría de Salud. https://www.gob. mx/cms/uploads/attachment/file/15174/MODELOCAMA_CNEGSR.pdf 20. Peacock, S., Smith, P.: The resource allocation consequences of the new NHS needs formula. York Centre of Health Economics, University of York (1995) 21. US GAO.: US Government Accountability Office. https://www.gao.gov/assets/gao-06-724. pdf 22. Ministry of Health.: Specific Action Program 2007–2012 Breast Cancer. https://es.calameo. com/read/0009477201394124b17b0

Calculating the Required Mammography Machines

161

23. Mittmann, N., Stout, N.K., Lee, P., Tosteson, A.N., Trentham-Dietz, A., Alagoz, O., Yaffe, M.J.: Total cost-effectiveness of mammography screening strategies. Health Rep. 26(12), 16– 25 (2015). https://www150.statcan.gc.ca/n1/pub/82-003-x/2015012/article/14295-eng.pdf 24. Barr, H., Blanco, S.A., Butler, P., et al.: Mammography services quality assurance: baseline standards for Latin America and the Caribbean (2016). https://iris.paho.org/han-dle/10665. 2/31402 25. Autier, P., Ouakrim, D.A.: Determinants of the number of mammography units in 31 coun-tries with significant mammography screening. Br. J. Cancer 99(7), 1185–1190 (2008). https://doi. org/10.1038/sj.bjc.6604657 26. Figueroa, J., Gray, E., Pashyan, N., et al.: The impact of the Covid-19 pandemic on breast cancer early detection and screening. Prev. Med. 151, 106585 (2021). https://doi.org/10.1016/ j.ypmed.2021.106585

Establishing the Optimal Standard for Preprocessing Head CT Data in Diagnostic Analysis Petra Nemcekova1(B) , Tomas Holecek1,2 , Jiri Chmelik1 , Petr Ourednicek2 , Katerina Valis2 , and Roman Jakubicek1 1 Department of Biomedical Engineering, FEEC, Brno University of Technology, Technicka

3082/12, 616 00 Brno, Czech Republic [email protected] 2 Department of Imaging Methods, St. Anne’s University Hospital Brno, Pekarska 664/53, 656 91 Brno, Czech Republic

Abstract. Acute ischemic stroke (AIS) is a significant cause of mortality and disability in adults. Diagnosis involves patient anamnesis, clinical examination, and brain and imaging. Thrombi on non-contrast CT (NCCT) may manifest as the hyperdense artery sign (HAS), while on CT angiography (CTA), an identification of thrombus can be detected as disruption of the contrast-enhanced artery. The preprocessing steps such as image registration are usually required. However, to the best of our knowledge, there is no definitive standard for the pretreatment of head CT data. To adress this, we used the stroke protocol including a multiphase CTA (mCTA) of the head and neck, and a brain perfusion CT. We conducted the registration process using SmartBrain and Elastix programs. Subsequently, the data were fused utilising a time maximum intensity projection technique (tMIP) over contrast phases. The manual thrombi segmentation was done on tMIP data using MITK software. Keywords: Brain stroke · Thrombus · Multi-phase computed tomography angiography · Image registration

1 Introduction Acute ischemic stroke (AIS) is caused by a blockage or clot in a blood vessel that supplies blood to the brain and is one of the leading causes of death and serious disability in adults. Rapid medical attention is critical to minimize the damage caused by AIS and improve the chances of successful recovery. The diagnosis is established through a combination of patient anamnesis, clinical examination, and initial imaging of the brain and cerebral vessels. Detecting and removing the thrombus is crucial for patient recovery. Based on the examination results, a treatment strategy is chosen, which may include intravenous thrombolysis or endovascular mechanical thrombectomy. Various characteristics of thrombus imaging, such as hyperdense artery sign (HAS) [1], thrombus location, clot © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 162–169, 2024. https://doi.org/10.1007/978-3-031-49068-2_18

Establishing the Optimal Standard for Preprocessing Head CT Data

163

perviousness, length, thrombus attenuation increase (TAI) [2], or clot burden score (CBS) can affect the outcome of AIS treatment [3]. Therefore, further thrombus analysis can be beneficial. The most common and widely utilized imaging technique for AIS diagnosis is computed tomography (CT). Thrombi on non-contrast CT (nCT) scans may manifest as the so-called HAS. On CT angiography (CTA), thrombus can be identified by the termination or disruption of the contrast-enhanced artery, so even non-hyperdense thrombi can be detected. Preprocessing is the crucial step for further image analysis. However, to the best of our knowledge, no definitive standard exists for the preprocessing of head CT data. A large number of brain pathologies can be identified through the comparison of asymmetries between the right and left hemispheres. Achieving rotational alignment of the brain is necessary for this purpose. The Smart Brain software, described in [4], can obtain this alignment. According to authors Yuan et. al. [5], image registration is also an essential part of the brain computed tomography (CT) analysis. As our dataset comprises three contrast phases and a native scan, we performed registration of the phases to the native scan followed by the fusion of all phases into a single fused image.

2 Dataset A CT imaging protocol was performed using a Brilliance iCT 256 scanner (Phillips Medical Systems, Eindhoven, the Netherlands). The protocol was based on the stroke protocol published by Menon et al. [6] and included a non-contrast CT (nCT) of the brain, a multiphase CT angiography (mCTA) of the head and neck, and a brain perfusion CT. The nCT covered the region from the skull base to the vertex using 0.9 mm axial overlapping slices. The first phase of the mCTA scan was timed to the peak arterial filling of the brain using bolus contrast monitoring and was scanned from the aortic arch to the vertex. The second and third CTA phases were performed from the skull base to the vertex as midvenous and late venous phases, respectively, with a time interval of 8 s between each phase. All of the CTA images resulted in 0.9 mm axial overlapping slices. The perfusion scans covered an 8 cm section of the brain with 10 continuous slices. A total of 37 cycles with a cycle time of 1.8 s were collected. Midvenous CTA, late venous CTA, or Timing-invariant CTA (TiCTA) can help assess collateral circulation and add temporal information to thrombus characteristics. [7] For further analysis, preprocessing steps such as image registration are usually required.

3 Methods In order to obtain the physical properties of the thrombus derived from the image features, segmentation of the thrombus is a crucial part of the following analysis. The basic part involves processing the data, such as registration or segmentation. The pipeline depicted in Fig. 1 has been designed and implemented as the goldstandard to facilitate the subsequent extraction of thrombus features. The Matlab source code for the presented preprocessing is available from GitHub repository https://github.com/PetraNemcekova/ Optimal-standard-for-preprocessing-Head-CT-Data.

164

P. Nemcekova et al.

Fig. 1. The proposed pipeline for the preprocessing of data required for the manual segmentation of thrombus properties.

Initially, the alignment of the native and contrast-enhanced scans with the radiological plane was achieved through the use of the SmartBrain [4] program. Usually, in CT data, the values of Hounsfield Units (HU) use to be in the range between −1024 and 3071 [8]. Therefore, the Winsorization of the data using the value of rescale intercept parameter defined in DICOM info was needed. Subsequently, it was necessary to convert all scan types, including the rotated native scans, from DICOM format to NIfTI format for the purpose of registering the contrast-enhanced scans with their corresponding native scans in the standard radiological view. In order to properly register the first phase of the contrast acquisition scan type to the native scan, it was necessary to crop some images due to the presence of the chest and aorta in this phase. This was done to ensure the accuracy of the subsequent data processing. The registration process was performed utilizing the Elastix registration suite [9] with the usage of rigid transformation parameter file. The interpolation was performed using the B-Spline function, with the adaptive stochastic gradient descent serving as the optimizer. Upon completion of the Euler transformation, the transforms were combined through composition. The optimization was performed with a maximum number of 1500 iterations, and a maximum step length of 5 was set. The final stage of the data preprocessing involved the fusion of three contrast scans, resulting in an image displaying the complete enhanced vessel tree. This was achieved by selecting the maximum pixel value from the corresponding position in each scan. With the processed data, the manual segmentation process was then carried out. The process of manual segmentation was carried out using the Medical Imaging Interaction Toolkit (MITK) software [10]. Initially, a radiologist determined the location of the thrombus within the patient. Subsequently, two trained individuals performed segmentation in three-dimensional space by segmenting the vessel perpendicular to its direction, resulting in the segmentation of the vessel’s cross-section. The segmentation process employed a parameter window level of 328 and a window width setting of 878.

Establishing the Optimal Standard for Preprocessing Head CT Data

165

4 Results and Discussion Given that the data obtained from the hospital was not rotated to the radiological plane, the SmartBrain program was utilized to acquire the same. The A part in the Fig. 2 displays the comparison between the rotated and original data, which is a crucial aspect for thrombus detection since the symmetry of hemispheres can be exploited. Given the necessity of registering contrast-enhanced scans with native scans, only transformation of the native scans using the SmartBrain tool was executed.

Fig. 2. Illustration of the results of specific preprocessing steps. Notably, image A displays the disparity between the original data, represented by the red channel, and the data that has been rotated to the radiological plane, shown in the green channel, using the SmartBrain tool. Image B depicts the output of the registration process for the second phase of the contrast-enhanced scan, represented by the green channel, which has been registered to the native data in the radiologic plane, shown in the red channel. Finally, image C exhibits the output of contrast phases fusion using the time maximum intensity projection technique.

The registration process was carried out with the aid of the Elastix tool. Each of the contrast-enhanced phases was registered with the native scan that had been transformed to the radiological plane. As depicted in part B in the Fig. 2, it is apparent that the usage of rigid transformation has led to improper overlay of soft tissues due to variations in head positioning during acquisition. Nonetheless, the skull section of the head in both images shows proper alignment, indicating successful registration. Subsequent to registration, the contrast-enhanced data underwent image fusion utilizing the time maximum intensity projection technique (tMIP). This facilitated the creation of a final image that incorporates all vessels enhanced by the contrast medium during the acquisition process, which is visible in the part C in the Fig. 2. The site of the thrombus is discernible as the absent segment of the vessel. After preprocessing, as defined by the pipeline mentioned earlier, a radiologist wrote the thrombus localization in the processed data. Two trained individuals subsequently applied segmentation in three-dimensional space to the preprocessed data, which was combined by the selection of the part of the disruption of vessel in the tMIP, which can be seen in the Fig. 3, and the selection of the HAS in the NCCT.

166

P. Nemcekova et al.

Fig. 3. Segmentation outputs comprise the following steps. Section A presents the tMIP of the patient’s CT scan, with a focus on the missing segment of the vessel. Section B illustrates the localization of the thrombus, represented by the red colour. Lastly, Section C displays the complete segmented thrombus.

For the purpose of segmentation, three independent raters delineated the thrombus based on a pre-established the golden standard for the selection of the thrombus. The raters included an expert radiologist with several years of practice in the field, a practising radiologist and a trained biomedical engineer. The segmentation was performed perpendicularly to the direction of the vessel to ensure that the lumen of the vessel was accurately selected. The resulting output of the manual segmentation process is the mask of the thrombus, which can be utilized for the extraction of the important thrombus features, such as TAI, CBS or other more abstract features. These features can be crucial in establishing proper treatment strategies. After segmentation, statistical analysis was performed on the segmented thrombi. To evaluate the similarity of the thrombus segmentation, the Søresen-Dice coefficient was calculated for each patient between all three raters. Table 1. Confusion matrix of the mean dice coefficients within raters. Mean dice Rater1

Rater2

Rater3

Rater2_2

Rater1

–

0.39

0.37

0.51

Rater2

0.39

–

0.54

0.61

Rater3

0.37

0.54

–

0.57

Rater2_2

0.51

0.61

0.57

–

The expert radiologist is referred to the Rater1, while the practicing radiologist is identified as Rater2. The third rater, Rater3, was assigned to the trained biomedical engineer. Rater2 performed the thrombus segmentation twice, and the second round of segmentation is denoted as Rater2_2. Table 1 shows that the highest similarity in thrombus segmentations was beween masks from the same rater with a Dice score of 0.61. The Dice scores between the raters in the first round of the segmentation and the

Establishing the Optimal Standard for Preprocessing Head CT Data

167

expert were 0.37 and 0.39. However, the second round of segmentation by the Rater2 increased the mean Dice scores for all other raters compared to the first round. Table 2. Confusion matrices of the maximal and minimal dice coefficients within raters. Max dice

Min dice

Rater1 Rater2 Rater3 Rater2_2

Rater1 Rater2 Rater3 Rater2_2

Rater1

–

0.39

0.37

0.69

Rater1

–

0.39

0.37

0.08

Rater2

0.39

–

0.54

0.80

Rater2

0.39

–

0.54

0.17

Rater3

0.37

0.54

–

0.72

Rater3

0.37

0.54

–

0.344

Rater2_2 0.69

0.80

0.72

–

Rater2_2 0.08

0.17

0.344

–

If the similarity between the Rater2 and the Rater2_2 is not taken into consideration, the maximum similarity between raters was found to be between the Rater3 and the Rater2_2 with a Dice score of 0.72. Additionally, the similarity between the Rater2_2 and the Rater1 was 0.69. The maximal and minimal values of the Dice score are presented in Table 2. High Dice scores were obtained between raters for patients with thrombi located in the linear part of the vessel, where there were no curvatures of the vessel included in the masks. The greatest disparity between raters was observed with a small thrombus size, where even minor differences in the segmented masks resulted in a decrease in the Dice score.

Fig. 4. The figure display histograms of segmented thrombi. Specifically, the red histogram illustrates the values of segmented thrombus obtained by the expert radiologist, while the green histogram depicts the values obtained by the trained radiologist, and the blue histogram represents the values derived from the thrombus segmented by the trained biomedical engineer.

168

P. Nemcekova et al.

To investigate the cause of the low similarities between raters, histograms of the thrombi HU values were plotted. The histograms for a single patient, generated using the masks from all three raters, are shown in the Fig. 4. It can be observed, that the thrombus segmented by the Rater1 was larger in size than those segmented by Rater3 and Rater2. However, the HU values in the segmented regions of the thrombus were consistent across all raters. This indicates that the observed disparity was due to differences in the length and the width of the segmented thrombus, rather than the incorrect localization of the thrombus itself. The varying characteristics of the thrombus, such as its density and clot perviousness, affect its visibility in both native and tMIP scans, making the manual segmentation of the thrombus challenging and inadequate.

5 Conclusion The composition of a thrombus changes its properties, such as density, thrombus attenuation or clot burden score. These properties can aid in determining the appropriate treatment strategy. The segmentation of the thrombus is a crucial step in acquirung its properties. This paper presents the pipeline for the data preprocessing required for the segmentation of the thrombus. The proposed pipeline is defined for NCCT data and also for the mCTA data. The preprocessing is based on the rotation of the data to the radiologic view using SmartBrain and Elastix tool, winsorisation of the HU values and creation of the tMIP. The preprocessed data obtained from the pipeline were used for the manual segmentation by three raters. The results of the segmentations were objectively evaluated using the Dice score and subjectively assessed by comparing the histograms of segmented thrombi. The low values of the Dice score among the raters points out the necessity of standardized and more robust thrombus segmentation. This can be achieved by the creation of appropriately optimized neural networks, which represents the subsequent phase of our future research. Acknowledgment. This paper and the research behind it would not have been possible without the support of Philips Healthcare company.

References 1. Zhou, Y., Jing, Y., Ospel, J., Goyal, M., McDonough, R., Yue, X., Ren, Y., Sun, Y., Li, B., Yu, W., Yang, P., Zhang, Y., Zhang, L., Li, Z., Duan, G., Ye, X., Hong, B., Shi, H., Han, H., Li, S., Liu, S., and, J.L.: CT hyperdense artery sign and the effect of alteplase in endovascular thrombectomy after acute stroke. Radiology 305(2), 410–418 (2022). https://doi.org/10.1148/ radiol.212358 2. Santos, E.M., et al.: Associations of thrombus perviousness derived from entire thrombus segmentation with functional outcome in patients with acute ischemic stroke. J. Biomech. 128, 110700 (2021). https://doi.org/10.1016/j.jbiomech.2021.110700

Establishing the Optimal Standard for Preprocessing Head CT Data

169

3. Huang, S.J., et al.: Value of thrombus imaging in predicting the outcomes of patients with largevessel occlusive strokes after endovascular therapy. Neurol. Sci. 41(6), 1451–1458 (2020). https://doi.org/10.1007/s10072-020-04296-7 4. Chmelik, J., Jakubicek, R., Vicar, T., Walek, P., Ourednicek, P., Jan, J.: Iterative machine learning based rotational alignment of brain 3d ct data. In: 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 4404– 4408 (2019). https://doi.org/10.1109/EMBC.2019.8857858 5. Yuan, H., Yang, M., Qian, S., Wang, W., Jia, X., Huang, F.: Brain CT registration using hybrid supervised convolutional neural network. Biomed. Eng. Online 20(1), 131 (2021). https://doi. org/10.1186/s12938-021-00971-8 6. Menon, B.K., et al.: Multiphase CT angiography: a new tool for the imaging triage of patients with acute ischemic stroke. Radiology 275(2), 510–520 (2015). https://doi.org/10.1148/rad iol.15142256 7. Smit, E.J., et al.: Timing-invariant imaging of collateral vessels in acute ischemic stroke. Stroke 44(8), 2194–2199 (2013). https://doi.org/10.1161/STROKEAHA.111.000675 8. Muschelli, J.: Recommendations for processing head CT data. Frontiers in Neuroinformatics 13 (Sep 2019). https://doi.org/10.3389/fninf.2019.00061 9. Klein, S., Staring, M., Murphy, K., Viergever, M.A., Pluim, J.P.W.: Elastix: A tool- box for intensity-based medical image registration. IEEE Trans. Med. Imaging 29(1), 196–205 (2010). https://doi.org/10.1109/TMI.2009.2035616 10. Wolf, I., Vetter, M., Wegner, I., Nolden, M., Bottger, T., Hastenteufel, M., Schobinger, M., Kunert, T., Meinzer, H.P.: The medical imaging interaction toolkit (MITK): a toolkit facilitating the creation of interactive software by extending VTK and ITK. In: Jr., R.L.G. (ed.) Medical Imaging 2004: Visualization, Image- Guided Procedures, and Display. vol. 5367, pp. 16 – 27. International Society for Optics and Photonics, SPIE (2004). https://doi.org/10. 1117/12.535112

Standardization of Failure Codes and Nomenclature of Medical Devices for Evidence-Based Maintenance Ernesto Iadanza1,2 and Alessio Luschi2(B) 1 International Federation for Medical and Biological Engineering, Council of Societies, Paris,

France [email protected] 2 Department of Medical Biotechnologies, University of Siena, Via A. Moro, 2, 53100 Siena, Italy [email protected]

Abstract. Biomedical technologies should be managed and maintained starting from the evidence provided by data. In this way, Clinical Engineering and Health Technology Management professionals can keep medical equipment safe and reliable with the optimal use of their resources. Evidence-Based Maintenance is grounded on the analysis of Real-World Evidence to monitor the maintenance effectiveness and plan any necessary changes to improve it. The lack of global standards contributes to the scarcity of accessible and shareable data to extract evidence. The purpose of this study is to draw attention to and describe the issues which can emerge during Evidence-Based Maintenance when dealing with the lack of standardization for naming and coding medical devices, with an emphasis on the nomenclature of medical equipment and the standardization of fault codes. Keywords: Evidence-based maintenance · Standardization · Nomenclature · Medical device · Healthcare

1 Introduction The majority of production sectors are impacted by today’s rapid and ongoing technological change, with healthcare playing a central role. Moreover, as it plays a crucial role in patient diagnosis and treatment, healthcare technology has evolved into a key component of the offered services. The volume and diversity of technical assets present in healthcare institutions reflect the complexity of technology management, which must be effective to ensure that the equipment is always used safely and effectively. In this picture, maintenance is an essential step in every medical device’s life cycle. In the last years, we have faced a process of critical analysis of the manufacturer’s maintenance recommendations, urging Clinical Engineering (CE) and Health Technology Management (HTM) professionals to adopt methods based on evidence to keep medical equipment safe and reliable while using their (often limited) resources judiciously. Evidence-Based Maintenance (EBM) starts from the analysis of the causes of equipment failures and uses © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 170–177, 2024. https://doi.org/10.1007/978-3-031-49068-2_19

Standardization of Failure Codes and Nomenclature of Medical

171

these results to continually improve maintenance. EBM involves the use of empirical data and scientific evidence to identify the optimal maintenance strategies for medical devices. This approach aims to improve the efficiency, reliability, and safety of medical equipment, which is critical to ensure the quality of healthcare delivery. EBM allows comparison of different maintenance strategies and provides concrete evidence to prove the safety and effectiveness of the one adopted. EBM begins with the analysis of Real-World Evidence (RWE) to monitor the maintenance effectiveness and plan any necessary changes to improve it. Maintenance reports, stored in Computerized Maintenance Management System (CMMS) software, can be a great source of Real-World Data (RWD), from which RWE can be extracted. Unfortunately, CMMS often only contain a description of the failures, the repair procedures and any spare parts used, thus lacking information about any measures needed to prevent the failure. The EBM approach, like all the approaches which are based on RWE [1–3], requires the availability of a significant amount of data to perform a solid study and extract real evidence of a general nature. The shortage of available and sharable data is worsened by the lack of standardization among countries which makes any sort of parsing nearly unfeasible. For instance, in the specific case of CMMS, the same medical device can be identified with different codes and different nomenclatures from country to country, as well as there is no international standard classification of fault codes. This paper aims to highlight and describe this set of problems, focusing on the need for standardization and interoperability of medical devices’ nomenclatures and faults classification.

2 Standardization of Failure Codes for Maintenance CMMS software became essential for HTM program operations. All modern CMMS software contain the fundamental fields that are needed for basic HTM program operations. Unfortunately, HTM programs differ widely in how they configure or suggest the use of those fields. More broadly, the lack of standardization largely prevents benchmarking between different structures and implementations. Performance metrics from one HTM program often cannot be compared to metrics from another HTM program. As a consequence, the HTM community stays in a weak position when confronting regulatory and accreditation agencies. In response to this challenge, the Association for the Advancement of Medical Instrumentation (AAMI) sponsored a “CMMS Collaborative” project among CMMS suppliers. The project started with the assumption that better use of existing CMMS software would make it easier to feed databases with accurate data and extract useful information from it. The involved suppliers all agreed on proposing a standard for the “Failure Code” field, as its purpose is to document the reason why a piece of medical equipment was unable to achieve its clinical objective of diagnosis, treatment, or monitoring [4]. Table 1 shows the proposed failure codes. A previous study by Iadanza et al. [5] showing the application of the EBM approach to a large hospital fleet of electromedical equipment, proposed a set of fault codes for corrective and predictive maintenance (Table 2), derived by [6–9], with the final goal of calculating 20 Key Performance Indicators (KPI). By analysing the mentioned examples, it surely emerges that the process toward standardization of CMMS failure codes has already begun. On the other hand, a lack of

172

E. Iadanza and A. Luschi

Table 1. Failure Code field options proposed by AAMI’s CMMS Collaborative project [4]. Code

Description

Accessory or disposable failure

Failure of device accessory or disposable, not a failure of the device itself

Calibration failure

Failure of a device to meet calibration parameters, requiring recalibration

Component failure (battery)

Failure of the battery that provides power for device operation

Component failure (not battery)

Failure of a device component other than the battery

Failure caused by maintenance

Failure of a device resulting from maintenance activities

Failure caused by abuse or negligence

Failure of a device resulting from damage caused by intentional misuse or negligent use

Network or connectivity failure

Functional failure external to device from failure of network or connectivity

Software failure

Functional failure of a device resulting from malfunctioning software

Use error (use failure)

Failure of a device to support achievement of a clinical objective

Failure caused by utility system

Functional failure of a device resulting from failure of or access to a utility system

Failure cause by environmental factor

Functional failure of a device resulting from an environmental factor

Failure could not be identified

Reported failure could not be reproduced or identified by testing

Failure not diagnosed—not repaired

Device Reported failure indicated that testing or repair was unwarranted

No failure associated with the work orders There was no failure associated with the work order (included for completeness)

consistency is still missing: the mentioned project by AAMI has been performed with limited consideration of the existing academic literature. A sore point is that the proposed fault codes are mostly focused on hardware failures, leaving it open the challenge of understanding the trends in the faults related to Health Information Technologies (HIT), considering that medical software is becoming more and more pervasive in healthcare, not to mention Artificial Intelligence (AI).

Standardization of Failure Codes and Nomenclature of Medical

173

Table 2. Failure codes for corrective and predictive maintenance proposed by [5]. Code

Description

NPF

No problem found

BATT

Battery failure

ACC

Accessory failure (including supplies)

NET

Failure related to network

USE

Failure induced by use (i.e., abuse, accident, environment conditions)

UPF

Unpreventable failure caused by normal wear and tear

PPF

Predictable and preventable failure

SIF

Induced by service (i.e., caused by a technical intervention not properly completed or premature failures of a part just replaced)

EF

Evident failure (i.e., evident to the user but not reported)

PF

Potential failure (i.e., in process of occurring)

HF

Hidden failure (i.e., not detectable by the user unless special test or measurement equipment)

3 Nomenclature of Medical Devices The nomenclature of medical devices is a coding and naming system used to classify and identify all medical devices and related health products. According to different classification and nomenclature systems, 5,000 to 24,000 different types of medical devices can be identified, ranging from very simple to complex, inexpensive to costly. Figure 1 clearly shows the heterogeneity of the existing nomenclature systems, highlighting that 39% of countries do not use any nomenclature, 8% use more than one system, and 16% have a nationally developed one. The nomenclature systems most widely used for medical devices are the European Medical Device Nomenclature (EMDN), the Global Medical Devices Nomenclature System (GMDN), and the Universal Medical Devices Nomenclature System (UMDNS). The EMDN is based on the Italian “Classificazione Nazionale Dispositivi Medici” (National Classification of Medical Devices - CND) and it is the nomenclature of use by manufacturers when registering their medical devices in the EUDAMED database according to the European Medical Devices Regulation 2017/745 [11]. The GMDN was developed by the European Committee for Standardization and medical device experts from around the world (manufacturers, healthcare authorities and regulators) based on the international standard ISO 15225 [12]. It is managed and maintained by a not-forprofit company, the GMDN Agency, which reports to a Board of Trustees on which medical device regulators and industry are represented. The UMDNS was developed by the Emergency Care Research Institute (ECRI). ECRI is a nongovernmental, nonprofit organization, governed by an Executive Committee and a Board of Trustees. The multiple nomenclatures in existence make it difficult to communicate important information between individuals and organizations, which can result in health, economic

174

E. Iadanza and A. Luschi

and social impact. It complicates interoperability, data extraction, procurement, supply and trade, and tracking of medical devices, negatively affecting patient safety, as well as technology management and maintenance [13]. Having a nomenclature system in place for medical equipment would facilitate its management and regulation by standardizing terms that enable communication despite linguistic and other barriers. Such standardization should be a prerequisite for inventory management and databases for the maintenance of equipment since it would provide a globally accessible, transparent and harmonized nomenclature system [14]. The World Health Organization (WHO) is one of the most significant international entities involved in the effort to establish a universal nomenclature for medical devices. During the 145th WHO Executive Board in 2019, the Director-General emphasized the necessity for a standardized nomenclature of medical devices “as a common language for recording and reporting medical devices across the entire health system at all levels of health care for a full range of purposes [...] The lack of a nomenclature system has hampered the development of the evidence and web-based health technologies database to provide guidance on appropriate medical devices” [15]. Besides, such a lack is actually impeding progress towards access to medical devices, which has a negative impact on efforts to facilitate emergency interventions and achieve universal health coverage [10]. WHO recognizes the availability of multiple systems and offers a platform towards convergence (Fig. 2).

Fig. 1. Distribution of countries based on the implemented nomenclature system [10].

WHO presented the first development of the International Classification and Nomenclature of Medical Devices (ICMD), implemented in the ICD-11 (International Classification of Diseases) platform [16]. The classification and terms generated represent the harmonisation of nomenclature and classes in the form of ontology and it is still under development. During the last 152nd WHO Executive Board in 2022, the DirectorGeneral still focused the attention on the fact that “the goal is to create a standardized

Standardization of Failure Codes and Nomenclature of Medical

175

international classification, coding and nomenclature for medical devices that would be available to all Member States and would support patient safety, access to medical devices for universal health coverage, emergency preparedness and response, efforts to increase the quality of health care”. Moreover, a request for proposals was posted in the United Nations Global Market Place from September 26th to October 20th 2022, with the intention of entering into a contract with the successful bidder for the provision of mapping medical device nomenclature data for integration in WHO platforms. The goals of the request were:

Fig. 2. Graphical overview of the process of mapping across different nomenclature systems leading to the implementation of the ICD [10].

– 1,200 types of medical devices due 3 months. – 4,000 types in 5 months. Contractors from four different countries were among the six offers that were examined. The request will be fulfilled by Symmetric Health Solutions from December 2022 to July 2023 [17].

4 Conclusions The paper performs a brief analysis of the evident lack of an updated world standard for naming and coding medical devices and their fault codes associated with the maintenance work orders. This absence leads to clear issues when trying to collect data from different systems because mapping across different nomenclature is nearly impossible due to the peculiar inner organisation of each nomenclature and Computerized Maintenance Management System software. This set of problems prevents the extraction of harmonized Real-World Data which is restraining, as a consequence, the development

176

E. Iadanza and A. Luschi

of Evidence-Based Maintenance which could otherwise provide guidance on improving the maintenance of medical devices while keeping medical equipment safe and reliable. In regards to the nomenclature of medical devices, something is moving, especially thanks to the efforts by WHO and the development of the International Classification and Nomenclature of Medical Devices. Instead, the standardization of failure codes for maintenance is still in an embryonic phase as involved actors seem more inclined to propose new classifications from scratch rather than making existing methodologies interoperable. Besides, they are also still attached to legacy approaches which, for instance, do not take software failures into account in spite of the ongoing spread of Health Information Technologies.

References 1. Luschi, A., Monti, M., Iadanza, E.: Assisted reproductive technology center design with quality function deployment approach. In: IFMBE Proceedings, vol. 51, pp. 1587–1590 (2015). https://doi.org/10.1007/978-3-319-19387-8386 2. Iadanza, E., Fabbri, R., Luschi, A., Gavazzi, F., Melillo, P., Simonelli, F., Gherardelli, M.: ORÁO: restful cloud-based ophthalmologic medical record for chromatic pupillometry. In: IFMBE Proceedings, vol. 73, pp. 713–720 (2019). https://doi.org/10.1007/978-3-030-179 717106 3. Mascii, L., Luschi, A., Iadanza, E.: Sentiment analysis for performance evaluation of maintenance in healthcare. In: IFMBE Proceedings, vol. 84, pp. 359–367 (2021) 4. Baretich, M.F., Davis-Smith, C.: Optimizing the cmms failure code field. Tech. rep., Association for the Advancement of Medical Instrumentation (2020) 5. Iadanza, E., Gonnelli, V., Satta, F., Gherardelli, M.: Evidence-based medical equipment management: a convenient implementation. Med. Biol. Eng. Compu. 57, 1–16 (2019) 6. Wang, B., Fedele, J., Pridgen, B., Williams, A., Rui, T., Barnett, L., Granade, C., Helfrich, R., Stephenson, B., Lesueur, D., Huffman, T., Wakefield, J., Hertzler, L., Poplin, B.: Evidencebased maintenance: Part i: Measuring maintenance effectiveness with failure codes. J. Clinic. Eng. 35(3), pp. 132–144 (2010) 7. Wang, B., et al.: Evidence-based maintenance: part ii: Comparing maintenance strategies using failure codes. J. Clin. Eng. 35(4), pp. 223–230 (2010) 8. Wang, B., et al.: Evidence-based maintenance: part iii: enhancing patient safety using failure code analysis. J. Clin. Eng. 36(2), pp. 72–84 (2011) 9. Wang, B., et al.: Evidence-based maintenance: part iv: comparison of scheduled inspection procedures. J. Clin. Eng. 38(3), pp. 108–116 (2010) 10. World Health Organization.: Nomenclature of medical devices. https://www.who.int/teams/ health-product-policy-and-standards/assistive-and-medical-technology/medical-devices/ nomenclature (2023). Accessed 23 March 2023 11. European Union: Medical devices regulation (eu) 2017/745. https://eur-lex.europa.eu/legalcontent/IT/ALL/?uri=celex:32017R0745 (2017). Accessed 23 March 2023 12. ISO 15225:2016.: Medical devices—Quality management—Medical device nomenclature data structure. Standard, International Organization for Standardization (2016) 13. Iadanza, E., Cerofolini, S., Lombardo, C., Satta, F., Gherardelli, M.: Medical devices nomenclature systems: a scoping review. Heal. Technol. 11, pp. 1–12 (2021) 14. Dezi, M., Luschi, A., Iadanza, E.: Creation of a system for the coding of medical devices. In: IFMBE Proceedings, vol. 51, pp. 1501–1503 (2015). https://doi.org/10.1007/978-3-31919387-8364

Standardization of Failure Codes and Nomenclature of Medical

177

15. World Health Organization.: Standardization of medical devices nomenclature: International classification, coding and nomenclature of medical devices: report by the director-general (2019) 16. Harrison, J., Weber, S., Jakob, R., Chute, C.: Icd-11: an international classification of diseases for the twenty-first century. BMC Med. Inform. Decis. Mak. 21 (2021) 17. World Health Organization.: Standardization of medical devices nomenclature: International classification, coding and nomenclature of medical devices: report by the director-general (2022)

Advanced Diabetes Technology for Better Glucoregulation, Opportunities and Cost Benefit (“Review on the Reality of a Developing Country”) ˇ ˇ Alma Badnjevi´c-Cengi´ c(B) , Amila Cerim-Aldobaši´ c, Mubina Hodži´c, and Davorka Dautbegovi´c-Stevanovi´c Department of internal medicine with hemodialysis, Cantonal hospital Zenica, Zenica, Bosnia and Herzegovina [email protected]

Abstract. The management of diabetes has been revolutionized by the introduction of novel technological treatments and modalities of care, such as continuous glucose monitoring, insulin pump therapy and artificial pancreas (closed-loop). Increased digitalization, and particularly the use of digital medicine products such as connected devices and digital applications, offers many potential benefits such as improved health outcomes, increased care efficiency, and improved quality of life for people with diabetes. However, these devices are unfortunately not universally available largely for economic reasons in developing countries. Treatment of diabetes is aimed at maintaining normoglycemia. That means maintenance of the fasting glucose level at patient with individual goals by achieving through glycated hemoglobin A1c and time in range. However, a significant number of patients with type 1 diabetes and type 2 diabetes use home self-monitors for glucose control. While self-monitoring of blood glucose is useful for measuring blood glucose levels, patients do not regularly check and self-monitoring of blood glucose does not enable many to adequately manage blood glucose levels or capture marked and sustained hyperglycemic excursions. This article presents the advantages of new technologies in the treatment of diabetes, as well as the possibilities, cost-effectiveness and availability for patients with diabetes in developing countries. Keywords: Advanced devices for continuous glucose monitoring · Diabetes type 1 · Diabetes type 2 · Self-monitoring of blood glucose · Advantages of diabetes technology

1 Introduction The prevalence of diabetes is increasing worldwide [1, 2]. Type 1 diabetes (T1D) is a chronic autoimmune disease characterized by loss of pancreatic beta cells and subsequent insulin deficiency. Lifelong intensified insulin treatment is currently the only recommended treatment modality for T1D, with the aim to reduce hyperglycemia and thereby prevent diabetes-related complications and premature mortality [3]. Healthcare © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 178–190, 2024. https://doi.org/10.1007/978-3-031-49068-2_20

Advanced Diabetes Technology for Better Glucoregulation

179

expenditure associated with T1D include the cost of therapy and treatment of diabetesrelated complications, such as retinopathy, kidney failure, neuropathy and foot ulceration, amputation, cardiovascular disease and stroke [4]. The International Diabetes Federation (IDF) estimates that 536.6 million people are living with diabetes (diagnosed or undiagnosed) in 2021. This number predicted to rise to 643 million by 2030 and 783 million by 2045. Diabetes caused at least USD 966 billion dollars in health expenditure-a 316% increase over the last 15 years [1, 5]. Despite significant therapeutic advancements, a person with diabetes routinely experiences physiological, cognitive, pragmatic, and psychological burdens. To achieve and maintain optimal glycemic control, those who require insulin generally must engage in time-consuming behaviors such as frequent glucose monitoring and quantifying carbohydrate intake, while also taking into consideration variables such as noncarbohydrate food content, exercise, illness, menstruation, stress, and other life events to adjust their medication doses [6, 7]. Newer insulin delivery systems are in development that seek to mitigate both hyperglycemia and hypoglycemia and increase time in range. Information systems now exist that may be leveraged to merge data from previously discrete systems into new models of connected care [8, 9]. Self-monitoring of blood glucose (SMBG) levels became the standard of care for type 1 diabetes (T1D) after the development of the first glucose meter for home use in 1970 [10]. The size, speed and accuracy of glucose meters have improved over time, and the volume of blood required for testing has decreased substantially. Despite these improvements in performance characteristics of glucose meters, little changed in home self-monitoring strategies [11]. Self- monitoring of blood glucose is universally considered to be an integral part of T1D management and crucial for optimizing the safety of patients with T2D, especially at patients in developing countries like Bosnia and Herzegovina. However, SMBG cannot detect nocturnal hypoglycemia and asymptomatic hypoglycemia, and it needs multiple finger-stick blood samples throughout the day. Therefore, SMBG may cause pain, discomfort, and compliance issues [12]. In addition, SMBG cannot provide any information on glucose trends, which may miss important glucose fluctuations. Continuous glucose monitoring (CGM) and continuous subcutaneous insulin infusion (CSII) are two diabetes technologies that have revolutionized diabetes care, with both technologies associated with improved HbA1c levels, increased TIR, and fewer hypoglycemic events compared with multiple daily doses of insulin (MDI) [13]. CGM is achieved through a compact medical device that monitors blood glucose levels in real time, which can send alerts to either a dedicated monitoring device or a smartphone app. CGM thereby reduces the need for frequent SMBG testing (and associated finger-pricking). The latest guidelines published by the Endocrine Society in the USA recommended use of CGM devices in adult populations with T1D who are willing to utilize the devices on a near-daily basis, irrespective of glycemic control status [14, 15]. The advancement of technology in the field of glycemic control has led to the widespread use of continuous glucose monitoring (CGM), which can be nowadays obtained from wearable devices equipped with a minimally invasive sensor, that is, transcutaneous needle type or implantable, and a transmitter that sends information to a receiver or smart device for data storage and display. In the 2000s, real-time interstitial CGM (rt-iCGM)

180

ˇ A. Badnjevi´c-Cengi´ c et al.

was introduced, although it still requires regular calibration (except for Dexcom G6 or G5 devices) [16, 17]. A recent advance in commercial use is interstitial glucose monitoring through flash glucose monitoring (FGM) technology in the form of devices such as Abbott’s FreeStyle Libre. Real- time interstitial CGM and FGM often find early clinical implementation in patients at high risk for complications, such patients with T1D and pregnancy. More recent studies have supported rt-iCGM and FGM using cost-benefit models compared to self-monitoring glucometers in type 1diabetes [18]. Furthermore, rt-iCGM and FGM provide additional information in the form of comprehensive data on the 24-hour glucose profile, current glucose trend, glucose variability, detection of periods of hypoglycemia and hyperglycemia, and estimated HbA1c. FGM has the additional advantage of factory calibration and interstitial blood sampling, thus avoiding the risk and discomfort of frequent subcutaneous sampling, significantly increasing its utility. CGM or FGM usually consists of 3 components: a wearable sensor, a transmitter that wirelessly transmits glycemic data, and a receiver nearby that displays such readings to the user [19]. This is further augmented by mobile health (mHealth) diabetic management systems. The World Health Organization defines mHealth as a medical and public health practice supported by mobile devices, such as mobile phones, patient monitoring devices, personal digital assistants, and other wireless devices [20]. There are significant diferences between self-monitoring of blood glucosae and flash continuous glucose monitoring (Table 1). Table 1. Key characteristics of different glucose monitoring methods Parameter

SMBG

Traditional CGM

Flash CGM

Fluid tested

Blood

Interstitial fluid

Interstitial fluid

User calibration required

No

Twice daily

No

Maximum duration of sensor use, days

Not applicable

7

14

Number of tests

Limited

Virtually unlimited

Virtually unlimited

Report interpretation

Variable (hardware-and software-dependent)

Can be complex (software-dependent(

Relatively easy (based on GP)

Operator training

Simple

Complex

Simple

Numerous large randomized studies (DIAMOND, HypoDE, REPLACE) have shown improvement of HbA1c, reduced time spent in hypoglycemia and hyperglycemia and reduction in the number of moderate and severe hypoglycemia in patients who used rtCGM or isCGM in relation to SMBG. The beneficial effect is particularly visible in people who have insensitivity to symptoms of hypoglycemia [21–24]. Freestyle Libre 3 continuous glucose monitor (CGM) from Abott was approved by the Food and drug administration (FDA) in may 2022 (Table 2.). Like the Freestyle Libre

Advanced Diabetes Technology for Better Glucoregulation

181

2, it is a 14-day system and has an easy-to-apply one-piece applicator. The biggest change with the new system is that it no longer requires the user to scan the sensor. The Libre 3 sensor automatically sends blood glucose readings to the user’s smartphone or receiver once every minute and has optional alerts for high and low readings. The Freestyle Libre 3 was FDA-approved for use in patients 4 years of age and older with type 1 or type 2 diabetes. The Libre 3 is being promoted as the world’s smallest, thinnest, and most accurate 14-day CGM system. Accuracy of blood glucose readings has also improved from the Freestyle Libre 2. CGMs use the metric Mean Absolute Relative Difference (MARD), which is the comparison between CGM readings and capillary blood glucose measurements. A lower percent difference reflects a more accurate reading from the CGM. The Freestyle Libre 2 system had a MARD of 9.3%, which improved to 7.9% with the Libre 3 system and is the first CGM system to have a MARD 39 and 31% for more than 3–4 h with increased frequency >Thrice a day.46% responded to question that if online schooling added more hours to total duration of usage of electronic devices. A significant correlation was found between (p < 0.05) Online schooling and total increased duration of use of electronic devices. To restrict this increased usage of device parents reduced easy access of children to devices and 66% of them even used Parenting control tools. A significant correlation was found between (p < 0.05) easy access of devices to children’s and use of parental control tool by parents. Parents feel that there is significant increase in device use buy children’s i.e., up to 57% of them (Fig. 4) and 54% of parents positively responded to its impact on sleep patterns and daily activities (Fig. 5). So, a significant correlation was found between (p < 0.05) increased device usage amongst children and its impact on their sleep patterns.

5 Discussion COVID-19 is a disease caused by a new strain of coronavirus.Individuals can also be infected from and touching surfaces contaminated with the infected person. As the crisis spreads around the world, it is transforming parents and children’s day-to-day lives. So, government ensured complete lockdown to interrupt virus from further transmission.

196

S. S. Shah et al.

Fig. 2. 100% of parents agreed to the think that their child’s sleep is not affected due to fear of covid-19 situation.

Fig. 3. 79% parents do not agree with tiredness of their child during daytime activities due to lack sleep and 21% of parents observed tired child while daytime activities.

Prevalence of Electronic Gadgets Usage and Its Impact on Sleep Patterns

197

Fig. 4. Total of hundred parents 57 percentage of parents observe increased device usage during COVID-19 pandemic by their child while 43% of parents think that they have not observed in any increase in the usage of devices

So, it is important that parents, teacher’s, pediatricians, pediatric dentists and people around them should provide information in an honest, age-appropriate manner and should help children cope up with situation. To this crisis children may respond in different ways. Common responses include having difficulties sleeping, bedwetting, having pain in the stomach or head, and being anxious, withdrawn, angry, clingy or afraid to be left alone. These changes can impact daily routine, activities as well as the sleep/wake pattern (circadian rhythmicity) [7]. Sleep is a periodic resting behavior which is synchronized in circadian rhythms [8]. Sleep is crucial for child and parent’s health and wellbeing. so, the potential for sleep related problems to emerge or worsen during lockdown was high. According to similar study done by author Liu Z et al. [9] there was big change in sleep/wake schedule of children in all age groups. Younger groups mainly represented by difficulty falling asleep that changed in all age groups (from 16.8 to 29.5% in 1- to 3-year-old children, from 13.3 to 25.9% in the 4- to 5-year-old group, from 11.6 to 26.5% in 6- to 12-year-old subjects and from 12.3 to 21.9%). There was daytime sleepiness while activities was also seen with a higher percentage of children They also noted the Children Sleep habits Questionnaire (CSHQ) reporting even a decrease of overall sleep disturbance, from 55.6 to 77.7% as compared to data previously collected by the same authors in 2018. Di Giorgio et al. [10] in their study evaluated about change in children’s behavioral and psychological factors during pandemic. These authors reported that the proportion of preschoolers with some sleep difficulties was stable from 41.5% before the lockdown to 44.7% during the lockdown as per responses parents.

198

S. S. Shah et al.

Fig. 5. Represents that 54 percentage of parents think that think that excessive use of device affects their sleep pattern and daily activities and 40% of the parents think that their child sleep is not affected

Children adapt to new behaviors due to disturbed psychological and social aspects in terms they tend to behave differently in public places and even in dental clinics. All the work, socialization, schools and business shifted to digital platforms. Due to all the restrictions and unable to go out and do activities children’s try to find more comfort and reduce boredom by playing video games, watching cartoons etc. using electronic devices [11]. Online learning led to an increase of the screen time, resulting in turn to a greater possibility of students’ distraction by social media or games.During the pandemic, children and adolescents significantly increased their average daily smartphone use as per reported by study by Drouin et al. [12] in 2020. This is unsurprising because the potential tasks of smartphones are used as source of communication, information and entertainment. All of the following adverse effects (clinical, psychological/behavioral, social) related to smartphone use were significantly more frequently reported in the study population during the COVID-19 pandemic, compared to the pre-epidemic period. Exposure to the light from screens in electronic devices which can lead to unfavorable outcomes such as more exposure to artificial light, irritation of eyes, reduced melanin secretion due to distorted circadian cycle and eventually disturbed sleep patterns. Pediatric dentist should help parents in recognizing all this changes and associated health risks. It is impossible to completely detach children from electronic devices. Parental attitude towards smartphone use should be known to help them further in decreasing or giving alternatives for overuse of electronic devices. So, along with the parents we should carry out all possible interventions to avoid impact of increase device

Prevalence of Electronic Gadgets Usage and Its Impact on Sleep Patterns

199

usage on sleep pattern. Parents should Listen to child’s concerns and give them affection, reassure them about their safety and should praise them frequently to encourage them in other activities. If possible, create opportunities for children to play and relax engage them in physical activities. Keep regular routines and schedules as much as possible, especially regarding sleep or rise timings, or help create a new environment. Try to use parental control tools to avoid maximum use of electronic devices and use of UV light protecting goggles or other protective system while device usage.etc. This study was constructed to capture some key issues related to overuse of electronic gadgets and propose early steps that parents can take to mitigate negative consequences for children.

6 Conclusion The results of study solidify some well-established data based on responses given by parents about disturbed sleep patterns of participants. Parents observed increase in the use of electronic device by children during this pandemic is significantly associated with effect of usage of electronic devices on the sleep pattern and the daily activities of children aged 7–10 years old in Pimpri-Chinchwad city. Altogether, as we know lockdown and crisis shifted all the work, socialization and schooling to digital platforms. Along with all the benefits of electronic devices, however comes repercussions. It is important for us and parents to be aware of impact of overuse of electronic devices on children’s health as well as sleep patterns. We should with parents try to eliminate those risks and help parents to restrict and control use of device usage.

7 Limitations 1. This is a cross-sectional questionnaire study, thus we only speculated about a causal relationship between sleep pattern and related risk factors due to increased device usage. 2. Other recent studies on the COVID-19 pandemic, used retrospective questions with the risk of pitfalls and biases although data which are quite consistent. 3. The parental-report nature of the study may affect our findings and it is possible that the parental perception of their child behaviour was heightened during the lockdown, with a potential inadvertent bias in their responses. 4. Our sample cannot be Considered as a representative for all the parents of Pimpri Chinchwad population and results may not be generalized to other cities.

8 Strengths 1. Our study, differently from the others, used a standardized questionnaire for evaluating Sleep pattern disturbances and there is a casual relationship between increase device usage among 7- to 10-year-old. 2. It can be used as a guide for parents to recognize and reduce increased device usage.

200

S. S. Shah et al.

9 Future Scope 1. One can conduct study regarding Children’s behavior during covid-19 consequent waves and there adjusting to new daily routine and its casual relation to Overuse of Electronic devices. 2. Furthermore, children’s behavior in Dental clinics during Covid-19 and their sedimentary behavior changes due to overuse of devices.

References 1. Dalton, L., Rapa, E., Ziebland, S., Rochat, T., Kelly, B., Hanington, L., et al.: Communication Expert Group. Communication with children and adolescents about the diagnosis of a lifethreatening condition in their parent. Lancet 393, 1164–1176 (2019) 2. Singh, S., Roy, D., Sinha, K., Parveen, S., Sharma, G., Joshi, G.: Impact of COVID19 and lockdown on mental health of children and adolescents: a narrative review with recommendations. Psychiatry Res. 293, 113429 (2020) 3. Interagency Standing Committee (IASC): Addressing mental health and psychosocial aspects of covid-19 outbreak. (2020) 4. Kozaki, T., Kubokawa, A., Taketomi, R., et al.: Effects of day-time exposure to different light intensities on light-induced melatonin suppression at night. J. Physiol. Anthropol. 34, 27 (2015) 5. Štefan, L., Horvatin, M., Bai´c, M.: Are sedentary behaviors associated with sleep duration? a cross-sectional case from Croatia. Int. J. Environ. Res. Public Health 16, 200 (2019) 6. AAP Council on Communications and Media: Media and young minds. Pediatrics 138(5), e20162591 (2016) 7. Altena, E., Baglioni, C., Espie, C.A.: Dealing with sleep problems during home confinement due to the COVID-19 outbreak: practical recommendations from a task force of the European CBT-I Academy. J. Sleep Res. 29 (2020). https://doi.org/10.1111/jsr.13052 8. Bruni, O., Malorgio, E., Doria, M., Finotti, E., Spruyt, K., Melegari, M.G., Villa, M.P., Ferri, R.: 9. Liu, Z., Tang, H., Jin, Q.: Sleep of preschoolers during the coronavirus disease 2019 (COVID 19) outbreak. J. Sleep Res. (2020). https://doi.org/10.1111/jsr.13142 10. Di Giorgio, E., Di Riso, D., Mioni, G.: The interplay between mothers’ and children behavioral and psychological factors during COVID-19: an Italian study. Eur. Child Adolesc. Psychiatr. (2020). https://doi.org/10.1007/s00787-020-01631-3 11. Dontre, A. J.: The influence of technology on academic distraction: a review. Hum. Behav. Emerg. Tech. 1–12 (2020) 12. Drouin, M., McDaniel, B.T., Pater, J., Toscos, T.: How parents and their children used social media and technology at the beginning of the COVID-19 pandemic and associations with anxiety. Cyberpsychol. Behav. Soc. Netw.. Behav. Soc. Netw. 23(11), 727–736 (2020)

Gender Impact Assessment for Medical Devices: A Compass to Find the Way in the Gender-Technology Reciprocity Manuela Appendino1(B) , Maria Agnese Pirozzi1 , Rossella Tomaiuolo2 Veronica Moi1 , and Luca Radice1

,

1 WeWomEngineers ETS, Turin, Italy

[email protected] 2 Faculty of Medicine and Surgery, Vita-Salute San Raffaele University, Milan, Italy

Abstract. In literature, data shows that medical devices generally do not intercept gender - referenced elements as well as the regulatory framework for their design. This perspectival study aimed to emphasize the features of gender, introducing health determinants to the common male-female identification, as also required by European Regulation. We conducted an evaluation on the presence of the term "gender" in the Regulations, identifying a first model approach to the reciprocity between gender and technology, accordingly, integrating gender, not only in the clinical dimension but also in the technological dimension. For the evaluation of the actual efficiency and safety of gender-related technology design, the importance of a common language to be able to report characteristics for the population as broad as detailed is presented in our study. This leads to a refinement of expertise through a combined approach involving physician and biomedical engineer in a synergic manner aimed at optimizing health economic resources, reducing side effects and potential clinical complications resulting from design imbalance. The presented model, based on Gender Impact Assessment (GIA) tool by European Institute for Gender Equality (EIGE), allows for refining the design process of medical devices, highlighting the real but so far undetected gender gaps. Keywords: Medical devices · Gender impact assessment · Gender-technology reciprocity

1 Introduction Integrating sex (refers to biological attributes) and gender (the social construction of femininity and masculinity, which includes sociocultural and psychological aspects) analysis into medical research design can improve its effectiveness. Indeed, peer-reviewed journals have implemented editorial guidelines to evaluate the rigor of sex and gender analysis, as well as major national and international funding agencies (Canadian Institutes of Health Research, European Commission, US National Institutes of Health, German Research Foundation, among others) adjusted funding policies considering gender determinants. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 201–207, 2024. https://doi.org/10.1007/978-3-031-49068-2_22

202

M. Appendino et al.

The same cannot be said for the criteria governing the design of the medical device and for the novel EU Medical Device Regulation (MDR) adopted by the European Parliament and Council in 2017, which will officially enter into operation on May 2024. This MDR was reimagined to improve the degree of public health protection and control operations over technical documentation; it clarifies the management and verification of medical devices, improving traceability through the UDI code and the European EUDAMED database. The new EU-MDR fits international agreements on medical devices as the European Union covers 30% of the global market, with a substantial increase in sales expected by 2027. Based upon manufacturer prices, the European medical device market is estimated to make up 27.3% of the world market. It is the second largest medical device market after the US (43.5%) [1]. The development of new medical devices benefits from positive factors such as the broad market and the novel MDR created to unify their intrinsic treatment, which, however, collides with rapid technological evolution (which could be reflected in a short life cycle of medical technologies if the products are placed on the market in the absence of an accurate health technology assessment) and the lack of gender in the regulatory context. The purpose of this paper is, through a multidisciplinary team, to combine medical and bioengineering skills to structure an approach capable of intercepting the elements related to sex and gender that can lead to medical device shortage (MDR art. 2.59) due to an indeterminate basic characterization of the patient, losing effectiveness (MDR art. 2.64).

2 Sex and Gender Determinants for Medical Devices 2.1 Medical Devices in Clinical Pathways First of all, it is necessary to define that the term medical device is an umbrella term: according to Regulation EU 2017/745 (MDR), Article 2, a “medical device” is “any instrument, apparatus, appliance, software, implant, reagent, material or other article intended by the manufacturer to be used, alone or in combination, on humans, for medical purposes and which does not exert in or on the human body the principal activities for which it is intended by pharmacological, immunological or metabolic means but whose function may be assisted by such means. The medical devices are also considered for conception control or support; the “accessory” necessary to specifically enable or directly assist their medical functionality [...]”. 2.2 Gender and Technology in the Current Scenario Despite these differences, one of the characteristics of medical devices is that they are usually part of a complex clinical pathway that includes a variety of medical actions and processes. Medical devices can potentially improve patient outcomes if technological improvements translate into improved patient outcomes. Evaluating the efficacy of medical devices has significant outstanding challenges, not least the need to incorporate the determinant of gender.

Gender Impact Assessment for Medical Devices

203

Even though according to standard IEC 62366-1, the patient population (sex and gender determinant, age, ethnicity, physique, etc.) shall be defined in the preparation of use specification as an entry point of the medical device design, these important data are introduced in the clinical investigation phase and is aimed at gathering helpful information to optimize the efficiency of healthcare resources and the effectiveness of healthcare interventions. The introduction of sex and gender determinants can favorably contribute to improving the effectiveness and safety of healthcare interventions and promote gender and health equity goals since the physiological aspect and the pathological course of the diseases are influenced by these factors. Hence the need to promote the integration of sex and gender as standard practice in the medical device development process. Opting for a direct relationship between gender and technology requires in-depth study and specific training to determine the same two-way relationship between physicians, designers, and other actors who, in the professional dimension, make their contributions with the goal of placing medical devices on the market that meet the main potential objectives of the new regulation. In order to consider a medical device safe and effective, it is not sufficient to assume a theoretical solution, but it is necessary to define a plan of activities that includes within it the organization of clinical investigations aimed at making consistent the intended use of our medical device, its function, and usability of it. Where clinical investigations should not consider the element of “gender” as a determinant of the type of investigation to be carried out, we may induce biases that would not allow us to identify a complete data scenario useful for the actual placement of the medical device.

3 Gender-Technology Reciprocity Gender-technology reciprocity theory refers to the inherent responsibility of the relationship between gender and technology as we consider both elements complement of design. Reciprocity intersects with sex, ethnicity, and comorbidity elements to be considered in the design and identification of clinical trials, even though clinical trials for medical devices may not have contributed to comparing data of gender understood as a “Condition” versus sex understood as a “biological condition (Fig. 1). The FDA filed a draft of “Evaluation of Sex-Specific Data in Medical Device Clinical Studies” in 2014 [2]. The document points out the discrepancy of sex-balanced data and describes specific potential actions to improve data monitoring and evaluation. First, to certify technology in a safe and balanced manner by encouraging further investigation of other related information that may emerge from sex-specific evaluations, secondly, enrolling patients considering demographics of disease distribution in an appropriate manner. Data disaggregation is the prerequisite for identifying gender relevance, as analyzing experimental results by sex and/or gender is critical for improving accuracy and avoiding misinterpretation of data. The common practice of pooling the response of females and males or women and men can mask sex differences. Data collection tools do not adequately incorporate these concepts. An important strategy for successful medical device development is to consider how datasets can be expanded to collect information beyond clinical indicators and alternative methods and mixed methods analyses to obtain different data types. This approach is especially salient given that

204

M. Appendino et al.

Fig. 1. Gender-technology reciprocity scheme

paying attention to the influence of gender requires asking multiple and multifaceted questions. This perspective paper explores how integrating sex and gender analysis into the medical device design.

4 Gender Impact Assessment The Gender Impact Assessment (GIA) method is an ex-ante or ex-post assessment of law, policy, social, or public health issues aimed at analyzing the influence of gender and planning policies suitable to prevent a negative effect on gender equality and improve gender equality through gender-oriented strategies. GIA was proposed by the European Institute for Gender Equality (EIGE) and allows for a structured (i.e., systematic, analytical, and documented) assessment of all these aspects following macro-steps [3], as exemplified in Table 1. Thus, medical device developers would benefit from assessment tools such as the GIA, to promote the adoption of sex and gender in healthcare research in support of better evidence. The primary objective is to plan medical device development to avoid negative effects, ensure technological safety and efficacy and set gender equality goals. Analyzing experimental results by sex and/or gender is critical for improving accuracy and avoiding misinterpretation of data. Therefore, some simplification and non-exhaustive questions are proposed as a guideline for the gender impact assessment applied to medical devices: • How can sex and/or gender impact the design of the medical device? • How can we verify sex and/or gender regarding existing medical devices and clinical investigations already prepared? • How can we consider “gender” as a distinguishing feature valid? • How do we ensure technological safety and efficacy of gender and sex are not considered? • How do we make design equitable if I do not contemplate gender-technology reciprocity?

Gender Impact Assessment for Medical Devices

205

Table 1. Table captions should be placed above the tables. Step 1

To define context, objectives and indicators

Context Objectives: provide gender-based recommendations for medical devices Indicators: sex and gender-specific anthropometric and psychological characteristics

Step 2

To explicate the relevance for gender Gender dynamics impacts Direct impacts of gender dynamics Indirect impacts of gender dynamics

Step 3

To identify gender impacts

Gender stereotypes Hierarchical positioning Unequal condition Unfair and unbalanced representation

Step 4

To evaluate gender impacts

Harmful impacts of gender bias aspects that reduce inequalities Aspects that promote equality over the status quo

Step 5

To provide recommendations for adjustments of gender gap

Suggestions for reducing inequalities Development of strategies to transform negative impacts of gender-gap into positive ones

• How can they democratize care if I do not start from design in gender? According to “Evaluation of gender-specific data in clinical trials of medical devices” [2], the points to consider are the gender-specific prevalence, the gender-specific diagnostic and therapeutic models, and any already known clinically significant sex or genderspecific differences in outcomes with respect to safety or efficacy.

5 Patterns of Gender-Technology Reciprocity Approach • Level 1: USER NEED LIKE gender-technology reciprocity 1. Comprehension and differentiation of the definition of gender and genre: YES/NO? 2. Macro definition of the medical device idea and intended purpose (MDR, art. 2.12) 3. Identification of the type of concept: does it come from outside due to market needs?/Does it come from specific needs inside the hospital facility? 4. Specific identification of the clinical performance of the technology: diagnosis, prevention, monitoring, prediction, prognosis, treatment or alleviation of disease, compensation for an injury or disability, investigation, replacement, or modification of the anatomy or of a physiological or pathological process or state, providing information by means of in vitro examination of specimens, control or support of conception (MDR, art. 2.1)

206

M. Appendino et al.

5. Within the detected function requirements and clinical performance (MDR art. 2.52) co-exists elements related to sex and gender that may influence? 6. Multiparametric clinical-specific identification 7. Identification of the pathology/anatomic district involved 8. Identification of gender-related elements and/or gender-related specificities 9. Identification in the pathology of clinical elements that can be differentiated according to sex and gender 10. Identification for the anatomical district involved clinical elements that can be differentiated according to sex and gender. • Level 2: DESIGN INPUT LIKE patient population (MDR Annex II, point 1.1 (c)) and gender-technology reciprocity. 1. Identification of the function and performance of the proof of concept: dimensions, materials of which it will be composed, and specific technological features 2. Identification of related gender elements and/or gender-related specificities on dimensions (i.e., knee prosthesis, stent) 3. Identification of related gender elements and/or gender-related specificities on materials (i.e., knee prosthesis, stent) 4. Identification of specific anatomical features that can be differentiated by sex and gender. • Level 3: DESIGN PROCESS LIKE gender-technology reciprocity Development and implementation of the prototype with specific features of level 1 and level 2 • Level 4: DESIGN OUTPUT LIKE gender-technology reciprocity Do the outputs take into account Level 2 in step 7? • Level 5: DESIGN VERIFICATION LIKE gender-technology reciprocity related to performance attributes 1. Does the verification take into account level 4? 2. Do the measured parameters consider the analysis performed following gendertechnology reciprocity? See all levels on Fig. 2.

6 Discussion and Conclusion The summarized model, which involves physicians and biomedical engineers through a combined approach, intercepts the elements related to sex and gender that can lead to a shortage of medical devices (MDR art. 2.59), due to an indeterminate basic characterization of the patient. The fallout of the proposed approach is to improve performance and reduce the risks associated with the non- recognition of specific conditions relating to sex and gender (MDR art. 2.64). As also referred by GIA, the aim of this research is to consider and distinguish the significant variables that discriminate men and women, to define specific parameters that differentiate the populations toward which the devices are intended, opting for a synergistic alignment of gender-referenced clinical information

Gender Impact Assessment for Medical Devices

207

Fig. 2. Flowchart of the proposed gender-technology reciprocity approach

with the medical device reference information. An alignment of language, combining skills and knowledge for a common interest, is sought to enrich and populate the genderreferenced clinical data pool by verifying and evaluating the type of data. The results should be contextualized by gender to determinate gender-disaggregated outcomes, in order to improve physician or designer understanding, and to optimize the funds available towards gender-disaggregated population health services.

References 1. The European medical technology industry in figures, MedTech Europe (medtecheurope.org) (2022). Accessed 1 March 2023 2. Evaluation of Sex-Specific Data in Medical Device Clinical Studies - Guidance for Industry and Food and Drug Administration Staff | FDA. https://www.fda.gov/regulatory-inform ation/search-fda-guidance-documents/evaluation-sex-specific-data-medical-device-clinicalstudies-guidance-industry-and-food-and-drug. Accessed 16 March 2023 3. European Commission. Toolkit for gender in research: checklist for gender in research. European Commission. 2011. https://op.europa.eu/en/publication-detail/-/publication/c17a4eba49ab-40f1-bb7b-bb6faaf8dec8. Accessed 16 March 2023

Open-Source Medical Device for in Vitro Diagnosis of Malaria Florinda Coro1,2(B) , Andrea Arcangeli2 , Carmelo De Maria1,2 Valentina Mangano3 , and Arti Ahluwalia1,2

,

1 Research Centre ‘E. Piaggio’, University of Pisa, Pisa, Italy [email protected], {carmelo.demaria, arti.ahluwalia}@unipi.it 2 Department of Information Engineering, University of Pisa, Pisa, Italy [email protected] 3 Department of Translational Research in New Technologies and Medicine, University of Pisa, Pisa, Italy [email protected]

Abstract. Malaria remains one of the greatest public health challenges at a global level. Early diagnosis can substantially reduce the number of victims. However, in low-middle income countries, which are mostly affected, the preponderance of low resource settings (LRS), such as community clinics, does not favour this practice. Microscopy is the gold standard in the diagnosis of malaria as it allows direct visualization of the parasite. Notwithstanding its potential, microscopy requires particular care in sample preparation, equipment and reagents, and highly qualified personnel, all of which are critical. Here we describe a prototype device to produce high-quality thin blood smears which can be stained and analysed using a microscope to diagnose malaria. The device was fabricated using additive manufacturing techniques that enable local production and fit appropriately into LRS. It was tested at the Pisa University Hospital, in Italy, comparing it with gold standards. Keywords: Malaria diagnosis · Open-source medical devices · Image analysis

1 Introduction Malaria is a parasitic disease caused by protozoans of the Plasmodium genus and transmitted by female Anopheles mosquitoes. According to the World Health Organization (WHO) 2022 World Malaria Report, only in 2021 there were 247 million clinical cases of malaria with about 619 000 victims, of whom 76% were children under 5 years old. It is currently endemic in 84 countries and the higher incidence is registered in SubSaharan African countries. Nigeria, the Democratic Republic of the Congo, Mozambique, Uganda, and Niger report almost half of global cases [1]. In these particularly troubled regions, malaria disproportionally affects poor people with limited access to health facilities and prevention tools [2]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 208–214, 2024. https://doi.org/10.1007/978-3-031-49068-2_23

Open-Source Medical Device for in Vitro Diagnosis of Malaria

209

Although there has been a decline in both incidence and casualties since the 2000s, the last few years have been marked by a stall in reduction and in 2020–2021 by an increase due to the COVID 19 pandemic and the consequent disruption of health services including malaria prevention strategies [3]. Different methods are used today to diagnose malaria: rapid diagnostic tests (RDTs), microscopy, quantitative buffy coat (QBC), polymerase chain reaction (PCR), and loopmediated isothermal amplification (LAMP) [4]. Apart from rapid antigenic tests, all technologies require qualified personnel and expensive equipment that are not available in low-resource settings (LRS), which are nevertheless the most affected [5]. On the other hand, RDTs show limited sensitivity and specificity [6]. Despite technological developments, microscopy remains the gold standard technique for diagnosis to date as it allows both qualitative and quantitative diagnosis. In addition to expensive laboratory equipment (such as the microscope itself), this procedure requires both preparation and analysis to be performed by experienced personnel. A badly stained smear is challenging to read, and an inadequately analyzed smear leads to a false diagnosis (either a false positive, which would result in unnecessary treatment favouring the development of drug resistance, or a false negative which would result in lack of treatment and therefore high chances of severe disease and death) [7]. In this context, the aim of this work is the design and fabrication of an open-source automated smearing system that can make the procedure independent of staff expertise and that is specifically designed for LRS in which it will be employed.

2 Materials and Methods 2.1 Device Design and Fabrication The thin blood smear procedure was analysed with the help of expert personnel at the Dep. of Translational Research in New Technologies and Medicine and at Pisa University Hospital (AOUP), identifying the bottlenecks or the steps more sensitive to errors. The operating procedure to prepare the samples for microscopy involves the use of 2 laboratory slides. A drop containing 5 µl of blood is placed on the first slide. The second slide, settled at an angle of approximately 45° by a sliding movement in the two different directions, first widens and then swipes the blood drop along the entire length of the first slide (Fig. 1a–c). The sample is then stained and analysed by an operator (Fig. 1d–e). At the end of the smear, a good quality sample will have a monolayer of red blood cells. A device prototype was developed involving the use of two slides as in the standard procedure (Fig. 1f). Its design was optimized to be 3D printed in ABS (acrylonitrile butadiene styrene) through the fused deposition modelling technique (Stratasys™ F370 3D printer available at the Research Centre E. Piaggio of the University of Pisa but feasible employing any 3D printer capable of working with ABS), thus enabling local manufacturing. The slide with the smear is placed in the slide housing while the second slide is held at 45° using a sliding slot. Once this procedure is complete, the sliding slot rotates to retrieve the smeared slide ready to be stained and examined under the microscope.

210

F. Coro et al.

Device detailed information is available on UBORA, a virtual platform to co-design open-source medical devices [8].

Fig. 1. (a) Deposition of the blood drop on the laboratory slide [7]; (b) Widening of the blood drop [7]; (c) Spreading of the drop on the laboratory slide [7]; (d) Thin blood smear; (e) Thin blood smear after staining; (f) CAD model of the device in Autodesk Fusion 360®

2.2 Usability Testing of the Device The device was tested in the Hematology Unit, Department of Laboratory Medicine of the AOUP. The tests were carried out using 5 blood samples from consenting patients who did not suffer from malaria infection as the aim of the experiment was to assess the quality of the smear, not of the diagnosis. The thin smears were performed by non-expert personnel under the guidance of expert staff. 2.3 Smear Quality Assessment The quality assessment of the produced smear was carried through both an image analysis procedure, and through a professional smear analyser. Images of the smears were acquired using a Leica DM6M microscope equipped with a 20× optical objective. In particular, images were acquired in the “tail” of the smear, where the blood is thinnest and therefore the single red blood cell layer can be assessed. The aim of the analysis is to evaluate the quality of the thin smear and the following quantitative parameters can be used to assess the quality:

Open-Source Medical Device for in Vitro Diagnosis of Malaria

211

• Number of red blood cells (RBCs) and density expressed as the number of RBCs per 1000 µm2 ; The two parameters give macroscopic information about the quality of the slide and the distribution of RBCs. Ideally the distribution should be homogeneous and the RBCs should not overlap; • Average area of RBCs expressed in µm2 ; • Average number of adjacent RBCs; average number of adjacent RBCs gives information on their overlapping; • Clark-Evans aggregation index. Clark Evans index measures the degree of clusteredness (if less than 1) or regularity (if greater than 1 up to the maximum value equal to 2.15). The images acquired by Leica DM6M were initially imported in the open-source image analysis software Cell Profiler. The Cell profiler analysis pipeline is shown in Fig. 2a. The image is uploaded to the software and converted first to grayscale and then to a black and white image. It is then inverted to exploit the “identify primary objects” function which requires that the objects of interest (RBCs in this case) be brighter than the background. Once a clear image is obtained, it is converted to gray-scale and for each identified RBC, information such as area and position is saved in a .csv file. The information is processed in pixels and then converted into microns using a scale bar. The.csv Cell Profiler output file is then analysed using a Python script in order to extract the five characteristic parameters described above. Parameters were obtained over 9 fields of view (FOV) acquired for each of the 5 thin smears (from 5 different blood samples) produced with the device. For further confirming the quality of the smears, the slides were analysed by the image processing module of an automatic smearer (Dasit Sysmex SP-10) present at AOUP to evaluate whether commercial equipment can analyse the smear obtained using the device without reporting any errors.

Fig. 2. (a) CellProfiler Pipeline; (b) microscope image; (c) grayscale image (d) output image of the fill objects function; (e) mask image

212

F. Coro et al.

3 Results and Discussion 3.1 Thin Smear Evaluation and Device Assessment The device was easy-to-use even for non-expert personnel. However, the ABS used for manufacturing was difficult to clean after the various tests due to its high porosity. As the device is hand-operated, a significant contribution dependent on the operator remains since the result is conditioned by the pressure of the slide and the sliding speed.

Fig. 3. Device quality parameter values. Each colour refers to a different blood sample. Mean and standard deviation refers to 9 fields of view for each patient sample. (a) Mean RBC count per FOV. (b) Mean RBC density [RBCs/1000 µm2 ]. (c) Mean Clark-Evans aggregation index per FOV. (d) Mean area per RBC [µm2 ]. (e) Mean adjacent neighbours per RBC.

The Clark-Evans aggregation index has an average value (averaged across all samples for each patient) of 1.418 (Fig. 3), far from zero representing the worst case and tending towards the optimum value of 2.15: this indicates low RBC aggregation [9]. The average number of adjacent neighbours is 0.376; this value, close to zero, indicates a low probability of RBCs overlapping.

Open-Source Medical Device for in Vitro Diagnosis of Malaria

213

In summary, the parameter values extracted from the images for the evaluation of the thin smear indicate that the specimens produced with the device are reproducible. The performance of the device in creating the thin smear under the hands of an unskilled operator is comparable to that obtained by an expert using traditional manual procedures [7] as confirmed by data in Table 1. The values in the ‘optimal values’ column (in Table 1) are mathematical indices which refer to the desired values. The values in the ‘skilled operator’ column (in Table 1), on the other hand, are taken from a study conducted in [7]. Density and RBC count per field of view are not shown as they are not comparable since they depend on the experimental set up (microscope magnification, area of the slide analysed). Mean index of aggregation value and mean adjacent neighbours per RBC value are close to the values of thin smears performed by expert personnel. It is therefore possible to state that the device is capable of obtaining smears of a high quality even if it is used by untrained operators. Table 1. Comparison between blood smear goodness parameters obtained by the device operated by a non-expert, a skilled operator, and standards parameters. Smear quality measurements

Device

Skilled operator [7]

Optimal values

Average RBCs count per field of view

1178

–

–

Average density (RBCs/1000 µ2 )

7.29

–

–

Mean index of aggregation

1.42

1.11

2.15

Mean adjacent neighbours per RBC

0.38

0.17

0

Mean area per RBC (µm2 )

50.85

–

–

The quality of the smears was confirmed by the commercial image analyser, which provided a complete blood count recognizing the slides as if they were produced by the automatic machine.

4 Conclusions The objective of this work was the fabrication of a device capable of facilitating the preparation of slides for the diagnosis of malaria independent of qualified personnel, to allow their routine use in LRS. The device still needs improvements to make its performance user independent. One possible approach would be to incorporate a motor to set the speed of motion and the design of a system for the controlled pressure of the slide. In addition, the material used to fabricate the device needs further optimisation. Although there are still some aspects to improve, the preliminary results are promising and suggest that the open-source device described can be appropriate for LRS. Acknowledgements. This work was supported by the University of Pisa project PRA mOSAIc: Open Source as key enabling approach for AI in healthcare (PRA_2020_38). The authors thank the Hematology Unit of the Department of Laboratory Medicine of the Azienda Ospedaliera Universitaria Pisana for their kind support in device testing.

214

F. Coro et al.

References 1. World Health Organization.: World malaria report 2022. World Health Organization (2022) 2. Worrall, E., Basu, S., Hanson, K.: Is malaria a disease of poverty? A review of the literature. Tropical Med. Int. Health 10(10), 1047–1059 (2005) 3. Weiss, D.J., et al.: Indirect effects of the COVID-19 pandemic on malaria intervention coverage, morbidity, and mortality in Africa: a geospatial modelling analysis. Lancet. Infect. Dis. 21(1), 59–69 (2021) 4. Tangpukdee, N., Duangdee, C., Wilairatana, P., Krudsood, S.: Malaria diagnosis: a brief review. Korean J. Parasitol. 47(2), 93 (2009) 5. Boadu, N.Y., Amuasi, J., Ansong, D., Einsiedel, E., Menon, D., Yanow, S.K.: Challenges with implementing malaria rapid diagnostic tests at primary care facilities in a Ghanaian district: a qualitative study. Malar. J. 15(1), 1–12 (2016) 6. Mwesigwa, J., Slater, H., Bradley, J., Saidy, B., Ceesay, F., Whittaker, C., D’Ales-sandro, U., et al.: Field performance of the malaria highly sensitive rapid diagnostic test in a setting of varying malaria transmission. Malaria J. 18, 1–13 (2019) 7. McDermott, S., Kim, J., Leledaki, A.A., Parry, D., Lee, L., Kabla, A., Cicuta, P. et al.: Autohaem: 3D printed devices for automated preparation of blood smears. Rev. Sci. Instrum. 93(1), 014104 (2022) 8. Malaria in vitro diagnostic tool, UBORA platform.: https://platform.ubora-biomedi-cal.org/ projects/44e4439b-f682-4bf1-afb1-cf2aa561ef9d. Last accessed Feb 2023 9. Clark, P.J., Evans, F.C.: Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 35(4), 445–453 (1954)

Assessing the Baseline Impact of Agile Human Resource Management in the Healthcare Systems of Western Balkan Countries Sandra Tinaj, Milica Vukotic, Bojana Malisic(B) , and Lidija Lukovac University of Donja Gorica (UDG), Donja Gorica Bb, 81000 Podgorica, Montenegro {Sandra.tinaj,milica.vukotic,Bojana.malisic, Lidija.lukovac}@udg.edu.me

Abstract. In many developed countries, human resources are managed on the basis of the HRM and human resources are highly trained and well paid. However, in the Western Balkans, human resources management in the healthcare sector are limited or non existent, and health personnel are often underpaid and undertrained. Work in this paper is based on desktop research of public and private health care organizations, telephone calls made to those we recognized as leaders in the Western Balkans in the public and private healthcare system, as well on relevant studies and scientific papers we gathered on this topic. The aim of this paper is to describe the health sector in the Western Balkans, to identify the situation in terms of human resources management and other related elements, both on a micro and macro level, as one of the basic conditions for solving the existing situation. Also, in paper were identified certain guidelines based on examples of more developed countries, and explained the necessary circular and agile approach to HRM as a tailor-made solution for WB. Keywords: Healthcare systems · Agile human resource management · Circular approach

1 Introduction and Background One of the main differences between the Western Balkans (WB) and other developed countries when it comes to human resource management (HRM) in healthcare is the level of investment and way of managing in human resources. Given that the entire data set for Western Balkan was not available, in some parts we used analysis that refer to developing countries. In many developed countries, human resources are managed on the basis of the HRM and human resources are highly trained and well paid. However, in the Western Balkans, human resources management in the healthcare sector are limited or non existent, and health personnel are often underpaid and undertrained. This lack of human resource management in health personnel can lead to lower levels of quality and access to health services and in its indirect context affects the entire population, social and economic development. As such, investing and managing of human resource © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 215–224, 2024. https://doi.org/10.1007/978-3-031-49068-2_24

216

S. Tinaj et al.

management in health can help improve the quality and access to health services in the Western Balkans. The countries of the Western Balkans continue to have some of the lowest public expenditure and investment in health systems in Europe and absence of the human resource management. This situation is continuous, it has implications in the present moment, while it will undoubtedly have implications in the future as well. Based on this, the measures taken in this area, even if they are to be taken now, only in 10–15% of the initiatives taken can have positive implications in the period of 1–2 years, while all other measures could only give results in a period of 5–7 years.

2 Method An analysis of the application of human resources management methods in health care in the Western Balkans, has not been done so far. Official data on whether there is an organized macro and micro level of human resource management in the health care sector in WB countries is not available. We based this work on desktop research of public and private health care organizations, telephone calls made to those we recognized as leaders in the Western Balkans in the public and private healthcare system, as well on relevant studies and scientific papers we gathered on this topic. This kind of content analysis identified the main themes, and resulted in a more negative feedback than we expected. We have conceived this paper in such a way as to first describe the health sector in the Western Balkans. After that, we identified the situation in terms of human resources management and other related elements, both on a micro and macro level, as one of the basic conditions for solving the existing situation. At the very end, instead of a conclusion, we identified certain guidelines based on examples of more developed countries, and we explain the necessary circular and agile approach to HRM as a tailormade solution for WB.

3 Current Status Analysis The health sector in the Western Balkans records absolutely all indicators significantly below the European average. These are facts that we can find in all studies and analyses. The list of indicators would be very long, but for these purposes we have singled out only some of the significant ones that speak in favor of the fact that urgent measures must be taken. Most of the measures, in addition to a higher level of investment, refer to the improvement, or better said, the establishment (since in all Western Balkan countries the existence of centralized units for the management of human resources1 at the macro level is absent, as well as departments at the micro level), organization and management of human resources, which has as its goal not only the investment of professional means, but more efficient development and management of human resources in the health system. 1 When we use the term human resources in our paper, we most often mean in addition to doctors,

physiotherapists and nurses, according to the Global Health Workforce Statistics Database, we also mean the seven other broad categories of the health workforce as defined by the WHO i.e. dentistry, pharmacy, laboratory, environment and public health, community and traditional health, health management and support, and all other health workforce categories.

Assessing the Baseline Impact of Agile

217

Fig. 1. Summary of shortage/surplus occupation in health workforce

Figure 1 gives summary of shortage/surplus skills occupation in health workforce in 2019 for WB countries according to the Occupation shortage index (OSI) [1].2 Positive values indicate shortages; negative values indicate surpluses. According to provided datas Albania and Montenegro are countries with both surplus in both health professionals and health associate professional, it refers to a situation where the supply of skills exceeds the demand of skills, while Bosnia and Herzegovina, Serbia and North Macedonia are in shortage (hard to find). This refers to a situation when the demand for skills exceeds the supply of skills in mentioned countries. Croatia is in surplus when it comes to health professionals while is in shortage in health associate professionals. These datas from OECD [2], based on the re-weighted occupational imbalance indicator shows that already experienced shortages of health professionals and/or health associate professionals prior to the COVID-19 pandemic, suggesting that it was hard to find sufficient numbers of workers with the right skills in health services occupations. The way in which they differ across countries, highlight the challenge of aligning the future supply of health workers to long-term demand correctly. For all countries, reforming training and employment strategies to better respond to changing skill and health needs can help meet demand. 2 Similar to the approach taken by OECD, the sub-indicators are linearly combined based on pre

defined weights to form the occupational shortage index (OSI). The employment sub- indicator was assigned a lower weight due to the ambiguity in its influence on skill shortages. The remaining sub-indicators were weighted equally. In the event one or more sub-indicators were missing, the OSI was computed using the remaining sub-indicators and re-weighted accordingly. Each sub-indicator was standardized by dividing it by the economy-wide standard deviation weighted by employment shares. The OSI is available for each reference area, occupation group and year (henceforth level-of-detail), allowing for year-on-year comparisons to be made. [1].

218

S. Tinaj et al.

The alarming situation in the countries of the Western Balkans, which requires action, that is, the application of HRM, is data that indicates that the healthcare system in the region has a low level of personnel and health equipment. The number of beds per doctor and bed occupancy rates between hospitals can vary significantly, which further underlies potential inefficiencies in the geographic dispersion of health services [3, 4]. Doctors and nurses per 1,000 inhabitants are below the EU average, and long waiting times for medical services are common (for example if we look at Montenegro it has 2.6 doctors per 1000 inhabitants compared to 3.4 which is the EU average). Moreover, the quality of medical training in the region is largely unable to meet EU medical training standards, as indicated by medical degrees from the region that are not always recognized abroad. In 2020, the satisfaction rate of health personnel in the Western Balkans was around 75%, which is lower than the global average of 85%. The countries of the Western Balkans continue to have some of the lowest public expenditure and investment in health systems in Europe. Despite the relatively low level of healthcare expenditure, healthcare costs impose a heavy financial burden on many people in the region—in particular those that are socioeconomically disadvantaged. In many Western Balkan countries, the share of out-of-pocket expenditure in current health expenditure is among the highest in Europe, with a regional average of 39% in 2018, compared to the EU average of 22% [5]. In 2020, the satisfaction rate of health personnel in the Western Balkans was around 75%, which is lower than the global average of 85%. If we start a concrete analysis of human resources, we will see similar data, as with other indicators. We find the topic of HR in healthcare in Western Balkan countries (WB6) highly intriguing because of the culture of living in those countries. Many studies have revealed that healthcare organizations do not have an HR department and have no plan of establishing one. The HR function is primarily led by employee specialists, or it is part of the legal and financial sectors. However, to be autonomous, there are only a few organizations that have HR with all of its functions. The desktop research and discussion with more than 30 leading hospitals form Western Balkans revealed that healthcare institutions in WB have the following institutional structure: CEO, Board of Directors, and may (or may not) have a specialist in charge of employees. WB6 healthcare institutes and organizations are still primarily organized centrally. The working environment and working conditions of healthcare employees are prioritized in all future plans and that is desirable, but nowhere do we find a setting in which it is very important to monitor the efficiency of the works, the evaluation of the effect. The disadvantage is that the working environment will not deliver efficient and effective work, high quality service, and required services to its clients and patients, if at the same time we do not have a more efficient organization and management of human resources in that environment. Existing data indicating that the Western Balkans needs HR management on both macro and micro level is that in the health sector the unemployment rate in the field of health in the region has increased by almost 10% in the last 10 years. This is an indicator that the need for improvement and more effective management of human resources in the field of healthcare in the Western Balkans clearly exists, and that more effective HR strategies should be implemented. In the Western Balkans and in the field of healthcare, there is a high unemployment rate, which is a consequence of the lack of appropriate HR strategies that would be adapted to the needs of the local labor market. Furthermore,

Assessing the Baseline Impact of Agile

219

there is a shortage of health workers in the Western Balkans, which makes the problem even more significant. Some additional factors include lack of funding and insufficiently trained staff. As is stated that it is necessary to introduce agile management of human resources as a response to the challenges faced by the WB health system, we must mention a couple of elements related to the introduction of an information system, which was based on a more efficient organization of human resources and resource management in the health system. It is unquestionable to talk about more efficient use of resources, and not to mention digitization processes, although based on qualitative analysis, it seems that although a number of WB countries have introduced this system, it is still not functional or is functional with a lot of shortcomings. Several countries have yet to fully implement centralized health information and communication systems at the state level (and those implemented must be further improved). These systems are vital for efficient delivery of health services and avoiding duplication of services, especially cancer screening for which current surveillance and response capacities are limited [6]. The increased use of private health care providers by those who can afford it further signals dissatisfaction with public health care.

4 Practice and Research that Indicate the Necessity of HRM By working on a comparative analysis of the practices and research activities of certain countries, we received confirmation of the thesis that organized and efficiently oriented management of human resources and the existence of human resources professionals really improves the condition and treatment in the entire health care system, and in an indirect sense has effect on the overall condition of the society. In the analysis, we chose those countries in which we fully recognized the elements of HRM, especially in the part of the recognized HRM processes that we identified as attracting, maintenance and development of HRM. Also, one part refers to the analysis of human resource professionals in the part of competences and treatment. Here we will mention a very interesting analysis related to the impact of HRM on patient mortality, where of course this impact is observed through the quality of work of health workers, in an indirect context. Although the conclusions were not entirely clear, and we do not have updated research after 2002. All findings indicate that this thesis is correct. Furthermore, when you evaluate the degree of reliability and quality of health systems, the number of health workers, the way of organization and the mortality rate at the level of individual countries, there is also a correlation that indicates that this is all extremely connected. A large study of 61 UK hospitals found a strong association between the quality of HRM and patient mortality. In particular, the adequacy of in-house training, the extent and sophistication of staff assessment and the extent of teamwork have been identified as factors influencing patient health [7, 8]. A recent international study highlighted how patient outcomes are mediated by key attributes of professional nurses, including nurse staffing, nurse-physician relationships and nurse autonomy. In addition to the above mentioned, we performed an analysis of the German health policy. Germany could definitely be the subject of a more detailed analysis and comparative practice, because it applies HRM policies in a large number of hospitals. The analysis in the case of

220

S. Tinaj et al.

Germany is also carried out in the next part, where a paper that analyzes five parameters, number composition and distribution of health care workers, workforce training issues, migration of health workers, economic development in a country and group of factors is taken into consideration. For instance, in the context of employee development, we can look at Germany as a model that provides its healthcare staff with opportunities for training and development. The healthcare system in Germany provides a diverse range of training options both on and off the job. Healthcare workers are also motivated to take part in continuing education programs and workshops, which help them stay up-to-date with the latest advancements in their field. Furthermore, Germany provides financial incentives to healthcare workers who engage in training and development activities. As an example of comparative analysis, Cyprus [9] is also very interesting. Cyprus has a highly organized and focused HRM both at macro and micro levels. Exceptional HRM policy in terms of employment as well as training and development of employees. In Cyprus, physical and human resources are split between government hospitals and health care centers and private hospitals, clinics, and polyclinics. Cyprus has the best healthcare system in the world, with the highest quality. Nicosia General Hospital (NGH) is the largest hospital and serves as a referral center for specialized care not available elsewhere in the country. In the public health sector, the recruitment process for all sorts of health care professionals is the same. The Civil Service Committee is in charge of the recruitment procedure. Hospitals will have more management authority over health worker recruitment under the proposed GHIS. In addition to the analysis of certain policies, the contribution to the understanding of the situation when it comes to the management of health workers from the aspect of the principles and functioning of HRM can be found in the paper by authors Kabene et al. [10] which provides an overview that illuminates some of the implications for health care professionals with regard to the five main issues raised in the article which vary from country to country. For example, number composition and distribution of health care workers needs to be assessed in Canada and the US at the state and local levels, and in Germany there is a need to address oversupply in certain regions. Developing countries face significant challenges in all areas, and as they increase the number of workers, HRP (Human resources professionals) will have an active and important role to play in managing healthcare resources efficiently and effectively. As for workforce training issues in Canada, HRP will help create a culture that encourages interdisciplinary health care delivery. In the United States, HRP will assist in developing a culture that ensures effective and efficient health care delivery and in Germany, HRP has an opportunity to provide input into planning for future training to reflect current and anticipated needs, and support the move towards interdisciplinary health care delivery. In developing countries with limited resources, HRP will have to develop sustainable strategies, including increased use of technology and a broader role for different health workers, and work with existing health workers to integrate new approaches to training. When talking about migration of health workers there are existing challenges of meeting the health needs of remote areas in Canada and the US, and the need for HRP to work with the provinces and state officials to develop programs and incentives to encourage health workers to consider moving to these areas. In Germany, HRP can work on strategic planning to

Assessing the Baseline Impact of Agile

221

better match supply and demand and reduce the migration of workers. However, developing countries face extreme differences between rural and urban areas and increasing pressure from other countries to poach health care workers, and HRP can help reduce the migration of health workers by developing policies and strategies to address workforce training issues. When talking about the level of economic development in a country HRP plays an important role in addressing financial pressures on health spending in Canada and other countries at all levels of economic development. In the United States, the issue is not lack of resources, but rather access to them, as shown by the large number of uninsured Americans. In Germany, HRP can contribute to a more efficient health care system. Developing countries face the challenge of limited resources, and HRP can help them maximize their resources by making the health system more efficient and effective. HRP should be involved at a strategic level in health planning to influence spending priorities and address economic realities. What is also discussed are the unique challenges faced by Canada, the United States, and Germany in terms of sociodemographic, geographical, and cultural factors, as well as their economic development. While Canada and the United States share many similarities, such as an ageing population and financial pressures in healthcare spending, Germany faces its own challenges. However, the developed nature of these countries means that approaches that work in one country can be adapted and applied to others. Developing countries, on the other hand, require careful consideration of their specific differences, and HRP can play a crucial role in identifying and addressing these factors. The importance of HRP in addressing the identified factors that impact healthcare cannot be overstated, as solutions to healthcare issues extend beyond medical interventions. It is important to note that in the paper is not analyze migration as one of the major problems of the healthcare system, although the introduction of HRM undoubtedly provides an answer to this problem as well. Understanding why healthcare personnel emigrate from the Western Balkans and Croatia and which are the consequences of this process are key to enabling state agencies and the government to develop optimal intervention strategies to retain medical staff. The benefit of this method is reliable estimates that can enable state agencies and the government to prepare and better respond to a possible shortage of healthcare workers and to protect the functioning of the health system [11].

5 HRM Circular Model, Agile Perspective of Human Resources Management In this paper, we singled out agile HR management with the aim of making a distinction related to traditional HRM, which is often associated only with a planning approach to human resources, and an HRM approach that primarily includes the entire set of activities and then continuous circular activity. The need for agile management of human resources in the health care system is unquestionable, and it is completely absent when we talk about WB. All the systems that we have analyzed in developed countries have developed specific functions or entire systems of HRM on a micro and macro level, while WB countries, as we stated in the section of desktop research and discussions with more than 30 hospitals, have absence of almost all elements of HRM policies and

222

S. Tinaj et al.

approaches. We especially want to emphasize the absence of elements such as training and development and performance management and performance appraisal in the health care system, i.e. the circular HRM system which is the basis for an agile approach and which can provide answers even in situations of imbalance (we mentioned that on the one hand we have an increase in the unemployment of health workers and on the other a great need, for example). With agile HR, you can apply an agile way of thinking and different work methods within your own teams and projects. Agile for HR has the potential to design an operating model to simply perform adequate maintenance of the existing staff, to direct them to much more efficient and high-quality work, on the one hand, while on the other hand, direct forces to attract adequate resources in the part of enrollment in secondary schools for medical staff, but also at faculties and colleges in the area of doctors and highly educated people. What was not mentioned in the paper and should be mentioned is that the HRM system adequately provides answers related to the economic analysis of human resources. The economic analysis of all the mentioned elements is very important, especially in the context of sustainability. Economic analysis is often avoided in the health care system, especially in the area of human resources, while both theory and practice have already shown “The basic idea that we developed is that the study of healthcare and health contains essential elements that require quality economic analyses [12]. HRM in WB, in addition to requiring a systemic approach, the absence of which is evident, also requires a continuous and circular approach. The circular approach of agile HRM with key elements is given in the following Fig. 2, which was created with the aim of a clearer presentation of the model.

Fig. 2. HRM circular model in healthcare macro and micro level

Assessing the Baseline Impact of Agile

223

6 Conclusion The healthcare system has an unquestionably huge impact on the development of the entire society. Human resources are the basis of every system, in health this role is still dominant despite the beginning of the application of Al, ML and DS. The analysis of successful systems and WB speaks in favor of the fact that this situation must be changed. The extremely bad and unsustainable situation in terms of the number of skilled employees in the healthcare system supports the fact that a circular approach to human resources management is necessary, which will include the systematic and model attraction, maintenance and development of human resources. The role of HRM in improving spending and investment in health systems in the Western Balkans can be seen from the perspective that HRM will ensure that health personnel are adequately attracted especially in the system of formal education, maintenance and development. This will help to create sustainable achievements. In addition, HRM can help reduce healthcare costs by improving efficiency and reducing waste time. HRM can also help motivate health care staff, which can lead to increased productivity and better patient outcomes. Acknowledgement. This research was carried out within the IPA project CFCU/MNE/212 Empower HR4Inno—Reinforce and connect absorption capacity in companies for new research and innovation solution, through the process of empowerment of the human resource management models and tools based on identified gap—IPA II—Annual Action Programme for Montenegro for the year 2020 Reference: EuropeAid/172–351/ID/ACT/ME.

References 1. International Labour Organization (ILO): Homepage. https://ilostat.ilo.org/. Last accessed 10 March 2023 2. OECD Homepage. https://www.oecd.org/. Last accessed 1 March 2023 3. World Bank Report: Homepage. https://www.worldbank.org/en/home. Last accessed 1 March 2023 4. World Bank Report: Homepage (2015). https://www.worldbank.org/en/home. Last accessed 1 March 2023 5. OECD Homepage. https://www.oecd-ilibrary.org/sites/ced731a9-en/index.html?itemId=/con tent/component/ced731a9-en. Last accessed 13 March 2023 6. European Commission Homepage. https://commission.europa.eu/index_en. Last accessed 1 March 2023 7. West, M.A., et al.: The link between the management of employees and patient mortality in acute hospitals. Int J Hum Resour Manag 13(8), 1299–1310 (2002) 8. Baldwin, S., George, J.: Qualitative study of UK health professionals’ experiences of working at the point of care during the COVID-19 pandemic. BMJ Open 11(9), e054377 (2021). https:// doi.org/10.1136/bmjopen-2021-054377 9. OECD Homepage, Cyprus: Country Health Profile (2021). https://www.oecd.org/. Last accessed 5 March 2023 10. Kabene, S., Orchard, C., Howard, J., M., Soriano, M., Leduc, R.: The importance of human resources management in health care: a global context. Hum. Resour. Health. 4(20) (2006)

224

S. Tinaj et al.

11. Juric, T.: Medical brain drain from Western Balkan and Croatia to Germany and Austria—an approach to the digital demography. MedRxiv. (2021) https://doi.org/10.1101/2021.05.26. 21257893 12. Vukotic, M., Tinaj, S.: Fundamentals of health economics. Health Information Management: Empowering Public Health, , p. 53. IOS press (2020)

Genetic Engineering

Review Paper: Autism Spectrum Disorder—Molecular Mechanisms and Diagnosis Selma Cifri´c Mujezinovi´c and Dado Latinovi´c(B) International Burch University, Francuske Revolucije Bb, , 71 210 Ilidža, Bosnia and Herzegovina [email protected], [email protected]

Abstract. Autism spectrum disorder (ASD) is a complex disorder characterized by: deficits in social communication and interaction; restricted, repetitive patterns of behavior, interests, or activities. In 10–20% of cases of ASD, there was a connection between chromosome aberrations and gene defects. Since there are many factors contributing to this condition, the focus of this research is on the molecular mechanisms associated with ASD and proper diagnosis by applying genomic tools. Rearrangements found among ASD patients are rare de novo and inherited CNVs. These variants hit genes that encode proteins responsible for normal neural development. Nowadays, researchers focus on these mechanisms to explain the dysregulation of neural development in ASD individuals. Gastrointestinal problems that are associated with most ASD cases suggest that it is not just a psychiatric disorder, it also has a physiological base. Rearrangements detected by MLPA should always be confirmed by other methods since some probe signals are more sensitive to sample purity or alterations in experimental conditions. Other genomic tools used for diagnostic purposes are CMA and WES/WGS. The development of novel methods to interpret the potential causes of ASD will further enable patients and their physicians to realize the maximum benefits of genetic testing for clinical care. Keywords: ASD · Neurodevelopment · Microbiome · CMA · WGS · WES

1 Introduction Autism Spectrum Disorder (ASD) is a group name for a wide range of neurodevelopmental disorders that can be described by poor social interaction, a deficit in a child’s communication, and unusual behavioral skills. ASD is a heterogenous, complex, and broad neural disorder characterized by sensory abnormalities and specific repetitive behavior [1]. The exact etiology of ASD is still unknown since it is considered a multifactorial disorder. Family studies have shown that ASD is both a familial and heritable condition [2, 3]. Atypical regulation in both innate and adaptive immunity is considered one of the indications of ASD [4]. Therefore, as a consequence of the multifactorial pattern of ASD, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 227–235, 2024. https://doi.org/10.1007/978-3-031-49068-2_25

228

S. C. Mujezinovi´c and D. Latinovi´c

a wide range of genes have been studied to make connections with ASD outcomes. The studies of genes include genes controlling the central nervous system (CNS), genes guiding biochemical pathways, those responsible for the formation of axons, dendrites, and synapses, and, also, genes that are correlated with autoimmune disorders and the immune system [5]. Many studies have proven that ASD is usually caused by combinations of recurrent and non-recurrent mutations. The chromosomal structural variations in the form of copy number variations (CNV) account for 10% of cases of autism with no obvious symptoms [6]. Rearrangements found among ASD patients are rare de novo and inherited CNVs. These variants hit genes that encode proteins responsible for normal neural development; the formation of synapses [7, 8]. ASD typically involves a comprehensive evaluation of behavior and development, including observation of social interaction, communication, and repetitive behaviors, which is one of the major challenges when applying diagnostics. The following are some of the diagnostic techniques that are used in the evaluation of individuals with autism: developmental screening, behavioral assessments, psychological evaluations, speech and language evaluations, and medical evaluations. Not only are the chemical pathways needed to be observed and the control of genes, but the medical observation has to be analyzed. It is important to note that there is no single test that can diagnose autism and a diagnosis is typically made by a team of healthcare professionals, including a pediatrician, psychologist, and speech-language pathologist [1, 2, 4, 9–13].

2 Methods and Materials This review paper aims to summarize the literature on the correlation of ASD with other disorders/diseases and present diagnostic techniques. Relevant papers were selected based on specific criteria, including English language, and keywords such as ASD; neurodevelopment; microbiome; CMA; WGS; WES. Databases like NCBI, Google Scholar, ResearchGate, and Web of Science were searched for relevant papers. Schematic illustrations, tables, and figures were employed to effectively summarize and present the information on this topic.

3 Results ASD affects many systems and processes in the body. Below are the covered the once with the greatest impact. Studies suggest that there is a strong correlation between the immune system and the outcome of ASD. Atypical number and ratios of lymphocytic cells and sufficiency or deficit of complements, cytokines, and immunoglobulins. Neural dysfunction can result from atypical immune activity during brain development. These include maternal immune abnormalities during pregnancy and a family history of autoimmune disorders [9, 14]. As shown in Table 1. ASD patients have a decreased level of naive T cells, an increased number of monocytes and macrophages, reduced activity of natural killer cells, and changes in interleukins and TNF-α. These results show the correlation between ASD and dysregulation of the immune system.

Review Paper: Autism Spectrum Disorder

229

Table 1. Alterations in the immune system in ASD individuals [9, 15–19]. Molecules of the immune system

Findings

Sources

Naïve T cells

Decreased levels of naïve T cells

[15]

Monocytes

Increase in the number of circulating monocytes

[16]

Natural killer cells

Reduced NK cell activity

[17]

Cytokines

Increased TNF-α and IL-12 Increased IL-6 and IL-10

[9, 18, 19]

Poor regulation of the immune system can lead to inflammation of neurons and the release of immune molecules. Studies have shown that there is a correlation between immune dysfunction and ASD and that treatment with immunomodulatory therapies can improve symptoms in some individuals with ASD [9]. Around 20–30% of ASD patients go through the loss of behavioral and language skills that are acquired before. This occurs around one or two years old. The brain structures that behave atypically in ASD are parts of the brain that develop later. ASD patients cannot express or detect emotions easily, the limbic system is one of the major interests in ASD studies. It is shown that many disrupted genes/chromosome regions in ASD are also found in other neurological disorders. Such variants cause autism alone, but they also may cause a wide variety of phenotypes with or without autism present, for example, intellectual disability, epilepsy, psychosis, etc. [20–22]. Below is the list of genes, a form of genetic risk variant, and clinical conditions where the mutation has been observed. Abbreviations used in the table are SZ, schizophrenia; ID, intellectual disability, DD, developmental delay, TS, Tourette syndrome, LD, language delay, SLI, specific language impairment, PM, point mutation, CP, and common polymorphism (Table 2) [23]. Table 2. Major genes/mutations associated with ASD and other neurodevelopmental disorders [23–26]. Gene or region

Mechanism

Disorders

NRXN1

CNV, PM

ASD, SZ

CNTNAP2

CNV, PM, CP

ASD, ID, epilepsy, LD/SLI, TS

16pdel

CNV

ASD, SZ, DD, LD, normal carrier

15q13, 3del

CNV

SZ, epilepsy, ASD, normal carrier

22q11

CNV

ASD, ADHD, SZ, ID, epilepsy

1q21

CNV

ASD, SZ, ID, epilepsy

230

S. C. Mujezinovi´c and D. Latinovi´c

A microbiome is a complete set of microorganisms that inhabit a specific area in the human body. Some of the experiments showed that the maternal microbiome during pregnancy plays a very important role in developing risks for ASD. Researchers believe that GI infection contributes to neurodegeneration. Using an animal model, it was observed that the presence of IL-1 and proinflammatory response can result in abnormal neural development. This is related to the disruption of the placenta by the immune system components. The prevalence of gastrointestinal symptoms such as abdominal pain, diarrhea, constipation, etc., is 23–70% in individuals with ASD [27, 28]. It is demonstrated that the infant’s gut is colonized by the microbiome from the maternal vagina and skin during the delivery of newborns through the birth canal. The results of studies have shown that the mother’s vaginal microbiota is the first exposure of the newborns to microbes. In babies born by C-section, the first microbial communities correspond to those found on the surface of the skin, for example, Staphylococcus spp. Changes in the gut microbiota can modulate the peripheral and CNS in animals, resulting in altered brain behavior. A pathway between the gut and the brain communication is known as the gut-brain axis. The relationship between ASD and the gut microbiota is represented by the increased permeability of the intestinal tract, known as the “leaky gut”. Previous experiments done on ASD animals had shown that defects of the GI tract, like “leaky gut”, help the entry of bacterial products and toxins into the bloodstream. These toxins influence brain function and behavior [29].

4 Discussion Besides the fact that physicians play an important role in the early recognition of ASD, genomic approaches need to be considered for appropriate and detailed diagnosis. MLPA (multiplex ligation-dependent probe amplification) detects abnormal copy numbers of different genomic sequences, which can distinguish sequence variation in only one nucleotide. It has been suggested that specific microdeletions and microduplications which are associated with cognitive impairment are also present in ASD individuals (also detectable by MLPA). Rearrangements detected by MLPA should always be confirmed by other methods since some probe signals are more sensitive to sample purity or alterations in experimental conditions [30]. Chromosomal microarray analysis (CMA), which screens the entire genome for tiny alterations in the sequence, should be the first genetic test performed when diagnosing autism. Researchers can uncover differentially expressed genes and pathways that may be involved in the development of ASD by comparing gene expression profiles between people with ASD and neurotypical controls [31, 32]. Most clinical approaches rely either on karyotyping—which involves looking at structural abnormalities of chromosomes under a microscope or testing for fragile X syndrome. Because these measures detect only the most obvious rearrangements, they may produce false results. CMA uses gene chips to pick up tiny genetic mutations across the genome, making it more likely that abnormalities will be found. The disadvantage of the CMA becoming the main diagnostic tool is the high cost of the procedure [22, 33]. All three tests (karyotyping, FXS test, and CMA) were performed on 933 children with autism. Karyotyping identified abnormalities in 2.23% of cases, whereas the FSX test

Review Paper: Autism Spectrum Disorder

231

gave positive results in 0.46% of cases. In contrast, CMA tests detected abnormalities in 7.3% of cases. It is easy to conclude that the CMA test shows the highest accuracy compared to other tests [34, 35]. CNVs (copy number variations) detected by CMA significantly contribute to understanding the etiology of ASD and other related conditions. Detection of CNVs by using microarrays is an error-prone process, therefore procedures must be followed strictly. A workflow of this procedure for the detection of de novo mutations is presented in Fig. 1. CNV regions were detected in individuals if they had an overall likelihood measure >0.9. CNVs were ignored if they were 60% similar in content to a variant detected in parents. If the variant appears as de novo, present only in the child, not in either parent, parents are undergoing testing using multiple informative markers. In the end, the additional validation of the suspect de novo lesion is conducted, in parents and subjects [36].

Fig. 1. Step-by-step procedure for the detection of de novo CNVs (present in a child and not in the parents) [36].

232

S. C. Mujezinovi´c and D. Latinovi´c

WES (whole genome sequencing) is a genomic procedure of sequencing all protein— coding regions of genes in the genome. Patients with intellectual disabilities, seizure disorders, and other conditions are usually good candidates for WES. The workflow of WES/WGS is shown in Fig. 2. However, there are several issues with experimental designs regarding the clinical application of WES in ASDs. The sequencing of trios, i.e. proband and both parents, has been the common experimental design in most research WES studies in ASDs. Some clinical laboratories vary regarding the sequencing of the parents. The advantage of the proband-only model is lower cost. The difference between these two designs is apparent because the sensitivity to detect de novo sequence variants is predicted to be higher in the trio design. All laboratories have an equal opportunity to access public reference genomes, also, each laboratory may have its internal reference genomes. In addition, different array platforms may vary in terms of their coverage, resolution, and cost, which can affect their suitability for ASD research. For example, whole-genome arrays can provide comprehensive coverage of the genome, but may be more expensive and may generate large amounts of data that require sophisticated analysis. In contrast, targeted arrays that focus on specific genomic regions may be more cost-effective and generate more manageable data but may miss important genetic variations outside of the targeted regions. Technically, not all coding exons can be captured for sequencing. Therefore, many differences in each laboratory may influence the sensitivity or success rate of WES [37, 38].

Fig. 2. The workflow of WES/WGS in clinical applications [37].

One of the first steps in improving this technique is the development of a platform and database for information sharing and the curation of clinical phenotypes and exome data worldwide [37, 38]. Compared to WES, WGS offers many specific advantages, even though it is considered a more expensive and complex process in terms of data storage and analysis. The DNA sample and library preparations are quick and straightforward since no DNA enrichment step is needed. The non-coding regions that are not covered by WES are included in WGS. WGS will also allow the detection of CNVs in high resolution.

Review Paper: Autism Spectrum Disorder

233

For these reasons, it is easy to predict that WGS will become a routine molecular test replacing WES and CMA as soon as the costs are reduced [13, 37].

5 Conclusion ASD is a complex disorder influenced by genetic and environmental factors. However, the primary cause and target for treatment remain unknown. Further research is necessary for a better understanding of ASD. While numerous genes associated with ASD have been identified, most contribute to genetic predisposition for various conditions. Dysregulation of the immune system can lead to neuronal inflammation and neurodevelopmental changes. High-resolution CMA has proven valuable in assessing neurodevelopmental disorders and identifying causal genetic variants. Future challenges for genetic research in ASD include performing WGS to compare susceptible individuals, investigating genotype-phenotype relationships for personalized treatment approaches, and improving diagnostic methods to account for the variability and complexity of the condition. Further study is crucial to comprehend the interplay of genetic, environmental, and neurological factors in ASD emergence.

References 1. Yochum, A.: Autism spectrum/pervasive developmental disorder. Prim. Care: Clin. Off. Pract. 43(2), 285–300 (2016). https://doi.org/10.1016/j.pop.2016.01.010 2. Guerra, D.J.: The molecular genetics of autism spectrum disorders: genomic mechanisms, neuroimmunopathology, and clinical implications. Autism Res. Treat. 2011, 1–16 (2011). https://doi.org/10.1155/2011/398636 3. Woodbury-Smith, M., Scherer, S.W.: Progress in the genetics of autism spectrum disorder. Dev. Med. Child Neurol. 60(5), 445–451 (2018). https://doi.org/10.1111/dmcn.13717 4. Meltzer, A., Van de Water, J.: The role of the immune system in autism spectrum disorder. Neuropsychopharmacology 42(1), 284–298 (2016). https://doi.org/10.1038/npp.2016.158 5. Needham, B.D., Tang, W., Wu, W.-L.: Searching for the gut microbial contributing factors to social behavior in rodent models of autism spectrum disorder. Dev. Neurobiol. 78(5), 474–499 (2018). https://doi.org/10.1002/dneu.22581 6. Sebat, J., et al.: Strong association of De Novo copy number mutations with autism. Science 316(5823), 445–449 (2007). https://doi.org/10.1126/science.1138659 7. Cook, E.H., Jr., Scherer, S.W.: Copy-number variations associated with neuropsychiatric conditions. Nature 455(7215), 919–923 (2008). https://doi.org/10.1038/nature07458 8. Brian, J.A., Zwaigenbaum, L., Ip, A.: Standards of diagnostic assessment for autism spectrum disorder. Paediatr. Child Health 24(7), 444–451 (2019). https://doi.org/10.1093/pch/pxz117 9. Lordan, R., Storni, C., De Benedictis, C.A.: Autism spectrum disorders: diagnosis and treatment. PubMed. (2021). https://www.ncbi.nlm.nih.gov/books/NBK573609/ 10. Chahin, S.S., Apple, R.W., Kuo, K.H., Dickson, C.A.: Autism spectrum disorder: psychological and functional assessment, and behavioral treatment approaches. Transl. Pediatr. 9(S1), S66–S75 (2020). https://doi.org/10.21037/tp.2019.11.06 11. Broome, K., McCabe, P., Docking, K., Doble, M.: A Systematic review of speech assessments for children with autism spectrum disorder: recommendations for best practice. Am. J. Speech Lang. Pathol. 26(3), 1011–1029 (2017). https://doi.org/10.1044/2017_ajslp-16-0014

234

S. C. Mujezinovi´c and D. Latinovi´c

12. Geschwind, D.H., Levitt, P.: Autism spectrum disorders: developmental disconnection syndromes. Curr. Opin. Neurobiol. 17(1), 103–111 (2007). https://doi.org/10.1016/j.conb.2007. 01.009 13. Lord, C., Risi, S., DiLavore, P.S., Shulman, C., Thurm, A., Pickles, A.: Autism from 2 to 9 years of age. Arch. Gen. Psychiatry 63(6), 694–701 (2006). https://doi.org/10.1001/archpsyc. 63.6.694 14. Huerta, M., Lord, C.: Diagnostic evaluation of autism spectrum disorders. Pediatr. Clin. North Am. 59(1), 103–111 (2012). https://doi.org/10.1016/j.pcl.2011.10.018 15. Sweeten, T.L., Posey, D.J., Shankar, S., McDougle, C.J.: High nitric oxide production in autistic disorder: a possible role for interferon-γ. Biol. Psychiat. 55(4), 434–437 (2004). https://doi.org/10.1016/j.biopsych.2003.09.001 16. Bennabi, M., et al.: Persistence of dysfunctional natural killer cells in adults with high— functioning autism spectrum disorders: stigma/consequence of unresolved early infectious events?. Mol. Autism 10(1), (2019). https://doi.org/10.1186/s13229-019-0269-1 17. Jyonouchi, H., Sun, S., Le, H.: Proinflammatory and regulatory cytokine production associated with innate and adaptive immune responses in children with autism spectrum disorders and developmental regression. J. Neuroimmunol. 120(1–2), 170–179 (2001). https://doi.org/10. 1016/s0165-5728(01)00421-0 18. Zimmerman, A.W., et al.: Cerebrospinal fluid and serum markers of inflammation in autism. Pediatr. Neurol. 33(3), 195–201 (2005). https://doi.org/10.1016/j.pediatrneurol.2005.03.014 19. Acosta, M.T., Pearl, P.L.: Imaging data in autism: from structure to malfunction. Semin. Pediatr. Neurol. 11(3), 205–213 (2004). https://doi.org/10.1016/j.spen.2004.07.004 20. Betancur, C.: Etiological heterogeneity in autism spectrum disorders: more than 100 genetic and genomic disorders and still counting. Brain Res. 1380, 42–77 (2011). https://doi.org/10. 1016/j.brainres.2010.11.078 21. Geschwind, D.H.: Genetics of autism spectrum disorders. Trends Cogn. Sci. 15(9), 409–416 (2011). https://doi.org/10.1016/j.tics.2011.07.003 22. de Hamilton, A.F.C.: Reflecting on the mirror neuron system in autism: A systematic review of current theories. Dev. Cogn. Neurosci. 3, 91–105 (2013). https://doi.org/10.1016/j.dcn. 2012.09.008 23. Bucan, M., et al.: Genome-wide analyses of exonic copy number variants in a family-based study point to novel autism susceptibility genes. PLoS Genet. 5(6), e1000536 (2009). https:// doi.org/10.1371/journal.pgen.1000536 24. Levinson, D.F., et al.: Copy number variants in schizophrenia: confirmation of five previous findings and new evidence for 3q29 microdeletions and VIPR2 duplications. Am. J. Psychiatry 168(3), 302–316 (2011). https://doi.org/10.1176/appi.ajp.2010.10060876 25. Vassos, E., et al.: Penetrance for copy number variants associated with schizophrenia. Hum. Mol. Genet. 19(17), 3477–3481 (2010). https://doi.org/10.1093/hmg/ddq259 26. Girard, S., Tremblay, L., Lepage, M., Sébire, G.: IL-1 receptor antagonist protects against placental and neurodevelopmental defects induced by maternal inflammation. J. Immunol. (Baltimore, Md.: 1950) 184(7), 3997–4005 (2010). https://doi.org/10.4049/jimmunol.090 3349 27. Chaidez, V., Hansen, R.L., Hertz-Picciotto, I.: Gastrointestinal problems in children with autism, developmental delays or typical development. J. Autism Dev. Disord. 44(5), 1117– 1127 (2013). https://doi.org/10.1007/s10803-013-1973-x 28. Dominguez-Bello, M.G., et al.: Delivery mode shapes the acquisition and structure of the initial microbiota across multiple body habitats in newborns. Proc. Natl. Acad. Sci. USA 107(26), 11971–11975 (2010). https://doi.org/10.1073/pnas.1002601107 29. Mayer, E.A., Tillisch, K., Gupta, A.: Gut/brain axis and the microbiota. J. Clin. Investig. 125(3), 926–938 (2015). https://doi.org/10.1172/jci76304

Review Paper: Autism Spectrum Disorder

235

30. Miller, D.T., et al.: Consensus statement: chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies. Am. J. Hum. Genet. 86(5), 749–764 (2010). https://doi.org/10.1016/j.ajhg.2010.04.006 31. Ashwood, P., Wills, S., Van de Water, J.: The immune response in autism: a new frontier for autism research. J. Leukoc. Biol. 80(1), 1–15 (2006). https://doi.org/10.1189/jlb.1205707 32. Ho, K., et al.: Chromosomal microarray analysis of consecutive individuals with autism spectrum disorders using an ultra-high resolution chromosomal microarray optimized for neurodevelopmental disorders. Int. J. Mol. Sci. 17(12), 2070 (2016). https://doi.org/10.3390/ ijms17122070 33. Shen, Y., et al.: Clinical genetic testing for patients with autism spectrum disorders. Pediatrics 125(4), e727–e735 (2010). https://doi.org/10.1542/peds.2009-1684 34. Levy, D., et al.: Rare de novo and transmitted copy-number variation in autistic spectrum disorders. Neuron 70(5), 886–897 (2011). https://doi.org/10.1016/j.neuron.2011.05.015 35. Verlouw, J.A.M., Clemens, E., de Vries, J.H., Zolk, O., Verkerk, A. J. M. H., am ZehnhoffDinnesen, A., Medina-Gomez, C., Lanvers-Kaminsky, C., Rivadeneira, F., Langer, T., van Meurs, J.B.J., van den Heuvel-Eibrink, M.M., Uitterlinden, A.G., Broer, L.: A comparison of genotyping arrays. Eur. J. Hum. Genet. (2021). https://doi.org/10.1038/s41431-021-00917-7 36. Yang, Y., et al.: Clinical whole-exome sequencing for the diagnosis of mendelian disorders. N. Engl. J. Med. 369(16), 1502–1511 (2013). https://doi.org/10.1056/NEJMoa1306555 37. Alonso-Esteban, Y., Alcantud-Marín, F.: Screening, diagnosis and early intervention in autism spectrum disorders. Children 9(2), 153 (2022). https://doi.org/10.3390/children9020153 38. Belkadi, A., et al.: Whole-genome sequencing is more powerful than whole-exome sequencing for detecting exome variants. Proc. Natl. Acad. Sci. 112(17), 5473–5478 (2015). https://doi. org/10.1073/pnas.1418631112

Analysis of MicroRNAs in Correlation to Astrocytoma Lejla Kadri´c1(B)

, Dina Neiroukh2 , Johannes Wagner1 and Aida Hajdarpaši´c1

,

1 University Sarajevo School of Science and Technology, Hrasniˇcka Cesta 3a, Sarajevo, Bosnia

and Herzegovina [email protected] 2 University of Galway, University Road, Galway H91 TKK33, Ireland

Abstract. Brain tumors, particularly astrocytomas, are a leading cause of cancerrelated deaths. MicroRNAs (miRNAs) are small, non-coding RNA molecules that control post-transcriptional gene expression. Since they are mostly extracted from body fluids and require a minimally invasive approach, miRNAs are significant biomarkers for the detection of cancer. This research aimed to elucidate miRNAs that are potential biomarkers for the detection of primary brain tumors, specifically astrocytomas. We searched for target genes involved in astrocytomas and then selected 11 genes that were most frequently mentioned in the research. Through an in-silico search, 741 miRNAs of target genes were found and further analyzed. Since there are very few common miRNAs between these genes, genemania software was used to see how the genes interacted so that they could be grouped for network analysis. MiRNAs of the following genes were selected for in-depth analysis: EGFR, BRAF, TP53, TERT, IDH1, and IDH2. Five miRNAs were classified as unique miRNAs for all genes (miR-181c- 3p, miR-423-5p, miR-200c-5p, microRNA let-7a-3, miR-744-5p), and five more as miRNAs with high confidence of interaction with mRNAs of the gene signature (miR-30c-1-3p, miR-152-3p, microRNA 30a, microRNA 30c-1, and microRNA 150). Dysfunction of some or all of these miRNAs is associated with astrocytoma tumors. Keywords: Astrocytoma · Glioblastoma · miRNA

1 Introduction Gliomas are the most common primary brain tumors of the central nervous system (CNS) arising from neuroepithelial cells like glial cells or precursor cells [1]. They include astrocytic tumors like astrocytoma, anaplastic astrocytoma, glioblastoma, as well as oligodendrogliomas and ependymomas, and are divided into two main subgroups: diffuse gliomas and non-diffuse gliomas [1, 2]. The most prevalent type of glioma known as an astrocytoma develops in the star-shaped glial cells known as astrocytes in the cerebral cortex. It often affects the brain and occasionally the spinal cord [3]. Glioblastoma Multiforme (GBM) is the most frequent astrocytoma and an aggressive malignant brain © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 236–245, 2024. https://doi.org/10.1007/978-3-031-49068-2_26

Analysis of MicroRNAs in Correlation to Astrocytoma

237

tumor that can develop as a primary glioblastoma or through progression from low-grade astrocytomas as secondary glioblastomas [4]. The diagnostic and prognostic evaluation of diffuse gliomas is primarily based on clinical and histopathologic evaluation, neuroimaging, and molecular genetic characteristics [5]. Investigation of molecular genetics has led to a re-evaluation of diagnostic criteria where biomarkers have become a component of the neuropathologic diagnosis of gliomas. Among the well-known biomarkers in glioma is the methylation status of the promoter for MGMT (O6-methylguanine-DNA methyltransferase) [4], as well as the mutational status of IDH (isocitrate dehydrogenase) 1 and IDH2 [5]. IDH mutations are commonly found in 85% of secondary GBM, while 7% in primary GBM, indicating a significant role in gliomagenesis [6]. Epidermal Growth Factor Receptor (EGFR) codes for a tyrosine kinase receptor, making it a hallmark of GBM prognosis [7]. Another associated gene is tumor protein 53 (TP53), a widely known tumor suppressor protein that has been observed at a higher rate in secondary GBM, implying that mutations occur at an early stage of glioma and later progress to GBM [7]. MicroRNAs (miRNAs) can generate significant advantages to better define both tumor grade and histological subtype of the tumor due to their marker stability. miRNAs are short, noncoding RNA molecules that post-transcriptionally regulate gene expression by base-pairing to 3’ untranslated region (3’UTR) of the targeted messenger RNA (mRNA) and thereby triggering mRNA degradation or translation inhibition [7]. Dysregulation of their expression and function can implicate pathogenesis. These molecular markers can be used in a less invasive manner as they can be collected from bodily fluids such as blood or saliva, specifically from the cerebrospinal fluid (CSF) for more accurate profiling of primary brain tumors [8]. MiR-30c, miR-152, miR-301a, miR-222, and miR-7 are among miRNAs that have been implicated in the development and progression of glioma [9–14]. MiR-30c and miR-152 have been found to be downregulated in glioma cells [9, 10]. Zheng et al. found indicators of miR-152 to inhibit brain tumor invasion via deactivation of the MEK/ERK pathway [10]. On the other hand, miR-301a and miR-222 are upregulated in glioma cells [11, 12]. Lan et al. found that miR-301a may be involved in the AKT and FAK signaling pathways by down-regulating PTEN [11], while miR-222 has been found to be a potential regulator in the AKT and IFN-α signaling pathways [12]. MiR-7 is another miRNA that downregulates the AKT pathway, as well as EGFR pathway, making it an important regulator of critical genes and malignant processes in glioblastoma [13]. In this study, miRNAs that are potential biomarkers for the detection of primary brain tumors, specifically astrocytomas, were elucidated. Target genes involved in astrocytomas were identified through in silico analysis and further analyzed. Based on the analysis of gene interactions, the most significant ones were grouped for network analysis of their miRNAs. Dysregulation of some or all miRNAs is connected to astrocytoma or other gliomas.

238

L. Kadri´c et al.

2 Materials and Methods The genes involved in astrocytoma were investigated by searching the PubMed database [14]. Astrocytoma and genetic mutations were used as search terms, and only studies published within the last five years, written in English and involving people who were 19 or older, were selected for further analysis (Fig. 1).

Fig. 1. PRISMA research process flow diagram

Using the genemania database [15], we identified the functions and interactions of the discovered genes with each other and with other relevant genes. The main interactions analyzed were Physical interactions, Co-expression, Predicted networks, Shared protein domains, Genetic interactions, Co-localization, and Pathways. miRNAs targeting the genes involved in astrocytoma were predicted using the mirWALK database [16]. Those miRNAs with a target score of 1, position 3 UTR, and validated, were selected. MetaCore software [17] was used to analyze the interactions between these miRNAs, i.e. network analysis was performed on all unique miRNAs and miRNAs with high confidence of interaction (predicted by both TargetScan and miRDb). Furthermore, based on genemania results, genes that showed significant interactions were clustered for network analysis of their miRNAs. Network clusters of miRNAs with a z-score larger than 60 were analyzed further.

Analysis of MicroRNAs in Correlation to Astrocytoma

239

3 Results By searching the PubMed database, 11 target genes involved in astrocytoma and glioblastoma were identified: TP53, EGFR, ATRX, IDH1, IDH2, TERT, MGMT, PTEN, BRAF, CDKN2A, and CDKN2B. EGFR and BRAF, EGFR and TP53, and TERT and TP53 have been shown to have genetic interactions. According to our results, there is an overall 82.01% physical interaction between the analyzed genes. Co-expression was 5.43%, while the percentage of predicted networks was 4.21%. There are 2.50% shared protein domains and 2.46% genetic interactions. The colocalization was 2.43%, while the percentage for the pathway was 0.96%. Based on genemania results, the genes selected for network analysis of their miRNAs were clustered as follows: miRNAs targeting EGFR and BRAF, miRNAs targeting EGFR, TP53, TERT, and BRAF, and miRNAs targeting IDH1 and IDH2. Genes selected for an in-depth analysis were: EGFR, BRAF, TP53, TERT, IDH1, and IDH2. The number of miRNAs for these genes was as follows: 11 miRNAs for EGFR, 6 miRNAs for BRAF, 293 miRNAs for the TP53 gene, 4 miRNAs for TERT, 368 miRNAs for IDH1, and 4 miRNAs for IDH2. miRNAs targeting more than one gene within the signature were miR-200c and miR222. The most significant cluster based on unique miRNAs for all genes included the following miRNAs: miR-181c-3p, miR-423-5p, miR-200c-5p, microRNA let-7a-3, miR744-5p. The processes controlled by these miRNAs are positive regulation of transcription initiation from RNA polymerase II promoter (20.8%), histone acetylation (25.0%), histone H3 acetylation (20.8%), regulation of transcription initiation from RNA polymerase II promoter (20.8%), internal peptidyl-lysine acetylation (25.0%). miRNAs with high confidence of interaction with mRNAs of the gene signature are: miR-30c-1-3p, miR-152-3p, microRNA 30a, microRNA 30c-1, and microRNA 150. The processes controlled by these miRNAs are cellular response to vascular endothelial growth factor stimulus (23.1%), positive regulation of Schwann cell differentiation (12.8%), acetaldehyde metabolic process (12.8%), response to insulin (30.8%), response to endogenous stimulus (56.4%). Based on miRNAs that target EGFR and BRAF, the most important miRNA cluster contained miR-145-5p, miR-27a-3p, miR-200a-3p, microRNA 200a, and miR-let-7a- 23p. These miRNAs regulate peptidyl-serine phosphorylation, protein modification, protein metabolic process, peptidyl-serine modification, and macromolecule modification. miR-19a-3p, miR-222-3p, miR-let-7a-2-3p, miR-19a-5p, and microRNA 1249 constitute a significant network of miRNAs targeting EGFR, TP53, TERT, and BRAF genes. These miRNAs regulate a variety of biological processes, including the regulation of DNA-templated transcription, nucleic acid-templated transcription, RNA biosynthetic process, negative regulation of nitrogen compound metabolism, and negative regulation of macromolecule metabolism.

240

L. Kadri´c et al.

4 Discussion Gliomas are classified as primary central nervous system tumors and cause significant levels of morbidity and death due to the location they originate and how aggressively they spread locally [17]. MiRNAs are endogenous, non-coding short RNAs that suppress translation or degrade particular mRNA to negatively regulate gene expression [18]. Additionally, certain miRNAs affect the formation and development of gliomas [19]. When compared to healthy cells, miRNA expression patterns in cancer cells are different, so miRNA profiles can identify the original metastatic brain tumor and distinguish between primary and secondary brain cancers. These findings suggest that miRNA analysis can be used to identify the origin of cancer cells and how they respond to treatment [20]. The most significant cluster based on unique miRNAs for all genes included the following miRNAs: miR-181c-3p, miR-423-5p, miR-200c-5p, microRNA let-7a-3, miR-744-5p. miR-181c targets the Krüppel-like factor family, which regulates a variety of biological and metabolic processes, including the cell cycle and proliferation of cancer stem cells [19]. miR-181c-3p is found to be expressed in breast cancer, epithelial ovarian cancer, endometrial carcinoma cells, liver cancer, osteosarcoma [21–24], etc. It is also found that miR-181c-3p is differentially expressed in glioblastoma [25]. Using astrocytes collected from rats, Song et al. [26] explored the contribution of exosomal miR-181c-3p and found that it is able to inhibit neuroinflammation by reducing the expression of CXCL1 and the production of inflammatory factors in astrocytes [26]. Cellular metabolism, amino acid metabolism, and transcription are the primary pathways whose activity is impacted by the miR-423-5p expression [27]. The evidence that miR-423-5p controls cellular metabolism in cancer cells was presented by Luce et al. [27]. miR-423-5p is involved in the development of glioblastomas, hepatocarcinoma, colorectal, gastric, prostate cancer, and ovarian malignancies [28–31]. A study performed by Li et al. [32] showed that miR-423-5p is overexpressed in glioblastoma. Through a variety of mechanisms including the stimulation of tumor cell proliferation, invasion, and angiogenesis as well as the activation of the AKT and ERK1/2 signaling pathways, miR-423-5p promotes the formation of tumors [32]. Many studies have revealed the function of miR-200c in tumorigenic pathways. Regulating multiple gene targets, the miR-200 family was wildly reported to suppress tumor progression and related to the prognosis of cancer patients, such as bladder cancer, prostate cancer, ductal carcinoma, and hepatocellular carcinoma [33–35]. There is no literature specifically investigating miR-200c-5p in relation to brain tumors. However, it was discovered that miR-200c-5p is expressed in hepatocellular carcinoma [35], renal cell carcinoma [36], colorectal cancer [37], etc. The let-7 microRNA, which is usually reduced in malignancies, is a tumor suppressor that prevents cellular proliferation and stimulates differentiation [38]. Let-7 microRNAs are dysregulated in several cancers, including hepatocellular carcinoma, gastric adenocarcinoma, brain, pancreatic, ovarian, breast, and prostate cancer, Burkitt lymphoma, renal cell carcinoma, and melanoma [39, 40]. There is very little research specifically addressing let-7a-3 in tumors. So far, this miRNA has been found to be involved in lung cancer [41], acute myeloid leukemia [42], and ovarian cancer [43].

Analysis of MicroRNAs in Correlation to Astrocytoma

241

miR-744-5p is involved in the negative regulation of many cancers, such as ovarian cancer, non-small cell lung cancer, colorectal cancer, cervical cancer, and glioblastoma [44, 45], while its overexpression can prevent cells from proliferating, migrating, and invading [50]. MiR-744 was strongly downregulated in GBM tissues and cell lines, according to Deng et al. [46]. In vitro, miR-744 overexpression prevented GBM cells from proliferating, forming colonies, migrating, invading, and causing apoptosis [46]. Moreover, Fan et al. [47] found that miR-744-5p, which directly targets RFC2, controls how RFC2 affects GBM cell proliferation, migration, adhesion, and death. miRNAs with high confidence of interaction with mRNAs of the gene signature found in this research are miR-30c-1-3p, miR-152-3p, microRNA 30a, microRNA 30c1, and microRNA 150. Three members of the miR-30 family were found in our research, including miR-30a, miR30c-1, and miR-30c-1-3p. Recent research has demonstrated that the miRNA- 30 family plays a crucial regulatory role in the development of tissues and organs and the pathogenesis of associated clinical disorders, as well as in the development of various cancers [47]. MiR-30a has been shown to modulate autophagy pathways in glioblastoma and medulloblastoma, indicating its potential as a therapeutic strategy for brain tumor treatment [48]. MiR-30c is expressed in various types of cancer, including non-small cell lung cancer, breast cancer, prostate cancer, colon cancer, gastric cancer, multiple myeloma, medulloblastoma, and glioblastoma [49, 50]. It is strongly associated with TNF-related apoptosis-inducing ligand during glioblastoma development [50]. Our study revealed that miR-30c-1 and miR-30c-1-3p are both involved in astrocytoma. However, there is no literature confirming the presence of these miR-30c family members in astrocytoma or other brain cancers. microRNA-152 (miR-152) acts as a tumor suppressor microRNA in a number of human cancers, such as ovarian cancer, breast cancer, cervical cancer, prostate cancer, and gastric cancer [51]. A study performed by Ma et al. [52] proved that miR-152 is downregulated in human glioblastoma stem cells. Increasing miR-152 expression significantly reduces cell migration, invasion, and proliferation while also triggering death [52]. miR-152-3p, which was discovered in this study as a significant miRNA involved in astrocytoma, was previously associated with other types of cancers such as breast cancer, colon cancer, prostate cancer, leukemia, etc. [53, 54]. A study performed by Sun et al. [55] revealed that glioblastoma tissues and cells had lower miR-1523p expression levels than non-tumor samples and normal cells. They also suggested that the regeneration of miR-152-3p may have therapeutic benefit in the treatment of glioblastoma. Studies have shown that miR-150 overexpression is directly linked to carcinogenesis, the formation of cancer, and malignant behavior. Also, it may have therapeutic effects through altering oncogenes and/or tumor suppressor genes [56, 57]. miR-150-5p is believed to function as a tumor suppressor by preventing the proliferation and migration of glioma cells [58]. Tian et al. [59] confirmed that miR- 150-5p is downregulated in gliomas and that overexpression of this miRNA may prevent colony formation, cell division, and tumor progression. Our study revealed an interesting function, positive regulation of Schwann cell differentiation, demonstrated by miRNAs with high confidence of interaction with mRNAs

242

L. Kadri´c et al.

of the gene signature. According to Deborde et al. [60], Schwann cells directly interact with cancer cells, leading to disease progression. The identified miRNAs regulate the formation of astrocytoma tumors by targeting genes related to the disease. However, there is little scientific evidence linking the identified miRNAs to astrocytomas. These miRNAs may serve as potential biomarkers for early diagnosis and may be useful in the treatment of brain tumors, especially astrocytoma, as their upregulation or downregulation is associated with the disease.

5 Conclusion An in silico analysis of astrocytoma miRNAs and their target genes was conducted. A significant amount of evidence confirmed that these miRNAs indeed control many genes and cellular processes involved in the development of astrocytoma and brain tumors in general. Accordingly, their discovery can help establish a way of developing brain tumor biomarkers, specifically astrocytoma. New findings in this area will help clinical professionals to improve existing diagnostic procedures and develop appropriate treatment methods.

References 1. Nizamutdinov, D., et al.: Prognostication of survival outcomes in patients diagnosed with glioblastoma. World Neurosurg. 1(109), e67-74 (2018) 2. Petrescu, G.E., Sabo, A.A., Torsin, L.I., Calin, G.A., Dragomir, M.P.: MicroRNA based theranostics for brain cancer: basic principles. J. Exp. Clin. Cancer Res. 38(1), 1–21 (2019) 3. Kapoor, M., Gupta, V.: Astrocytoma. InStatPearls [Internet] (2021) Nov 20. StatPearls Publishing 4. Oronsky, B., Reid, T.R., Oronsky, A., Sandhu, N., Knox, S.J.: A review of newly diagnosed glioblastoma. Front. Oncol. 5(10), 574012 (2021) 5. Appin, C.L., Brat, D.J.: Biomarker-driven diagnosis of diffuse gliomas. Mol. Aspects Med. 1(45), 87–96 (2015) 6. Sasmita, A.O., Wong, Y.P., Ling, A.P.: Biomarkers and therapeutic advances in glioblastoma multiforme. Asia Pac. J. Clin. Oncol. 14(1), 40–51 (2018) 7. Sun, E., Shi, Y.: MicroRNAs: small molecules with big roles in neurodevelopment and diseases. Exp. Neurol. 1(268), 46–53 (2015) 8. Saenz-Antoñanzas, A., et al.: Liquid biopsy in glioblastoma: opportunities, applications and challenges. Cancers 11(7), 950 (2019) 9. Liu, S., Li, X., Zhuang, S.: MiR-30c impedes glioblastoma cell proliferation and migration by targeting SOX9. Oncol. Res. Featur. Preclin. Clin. Cancer Ther. 27(2), 165–171 (2019) 10. Zheng, X., Chopp, M., Lu, Y., Buller, B., Jiang, F.: MiR-15b and miR-152 reduce glioma cell invasion and angiogenesis via NRP-2 and MMP-3. Cancer Lett. 329(2), 146–154 (2013) 11. Lan, F., Qing, Q., Pan, Q., Hu, M., Yu, H., Yue, X.: Serum exosomal miR-301a as a potential diagnostic and prognostic biomarker for human glioma. Cell. Oncol. 41, 25–33 (2018) 12. Sana, J., Hajduch, M., Michalek, J., Vyzula, R., Slaby, O.: MicroRNAs and glioblastoma: roles in core signalling pathways and potential clinical implications. J. Cell Mol. Med. 15(8), 1636–1644 (2011) 13. Zhang, C., et al.: Global changes of mRNA expression reveals an increased activity of the interferon-induced signal transducer and activator of transcription (STAT) pathway by repression of miR-221/222 in glioblastoma U251 cells. Int. J. Oncol. 36(6), 1503–1512 (2010)

Analysis of MicroRNAs in Correlation to Astrocytoma

243

14. PubMed [Internet]. Bethesda (MD): National Library of Medicine (US), National Center for Biotechnology Information; (2004). [cited 2023 Feb 20]. https://pubmed.ncbi.nlm.nih.gov/ 15. Warde-Farley, D., Donaldson, S.L., Comes, O., Zuberi, K., Badrawi, R., Chao, P., Franz, M., Grouios, C., Kazi, F., Lopes, C.T., Maitland, A.: The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucl. Acids Res. 38(suppl_2), W214-20 (2010) 16. Sticht, C., De La Torre, C., Parveen, A., Gretz, N.: MiRWalk: an online resource for prediction of microRNA binding sites. PLoS ONE 13(10), e0206239 (2018) 17. Mafi, A., et al.: Recent insights into the microRNA-dependent modulation of gliomas from pathogenesis to diagnosis and treatment. Cell. Mol. Biol. Lett. 27(1), 1–32 (2022) 18. Rao, S.A., Santosh, V., Somasundaram, K.: Genome-wide expression profiling identifies deregulated miRNAs in malignant astrocytoma. Mod. Pathol. 23(10), 1404–1417 (2010) 19. Shen, X., Li, Y., Sun, G., Guo, D., Bai, X.: MiR-181c-3p and-5p promotes high-glucoseinduced dysfunction in human umbilical vein endothelial cells by regulating leukemia inhibitory factor. Int. J. Biol. Macromol. 1(115), 509–517 (2018) 20. Sun, G., et al.: MiR-423-5p in brain metastasis: potential role in diagnostics and molecular biology. Cell Death Dis. 9(10), 936 (2018) 21. Rajarajan, D., Natesh, J., Penta, D., Meeran, S.M.: Dietary piperine suppresses obesityassociated breast cancer growth and metastasis by regulating the miR-181c-3p/PPARα Axis. J. Agric. Food Chem. 69(51), 15562–15574 (2021) 22. Rawoof, A., Au Yeung, C.L., Cheng, R.S, Ferri-Borgogno, S., Mok, S.C.: The role of exosomal miR-181c-3p within the ovarian tumor microenvironment 23. Livingstone, M.C., Johnson, N.M., Roebuck, B.D., Kensler, T.W., Groopman, J.D.: Profound changes in miRNA expression during cancer initiation by aflatoxin B1 and their abrogation by the chemopreventive triterpenoid CDDO-Im. Mol. Carcinog. 56(11), 2382–2390 (2017) 24. Leichter, A.L., Sullivan, M.J., Eccles, M.R., Chatterjee, A.: MicroRNA expression patterns and signalling pathways in the development and progression of childhood solid tumours. Mol. Cancer 16(1), 1–7 (2017) 25. Khalil, S., et al.: MiRNA array screening reveals cooperative MGMT—regulation between miR-181d-5p and miR-409-3p in glioblastoma. Oncotarget 7(19), 28195 (2016) 26. Song, H., et al.: Cortical neuron-derived exosomal MicroRNA-181c-3p inhibits neuroinflammation by downregulating CXCL1 in astrocytes of a rat model with ischemic brain injury. NeuroImmunoModulation 26(5), 217–233 (2019) 27. Luce, A., et al.: A proteomic approach reveals that miR-423-5p modulates glucidic and amino acid metabolism in prostate cancer cells. Int. J. Mol. Sci. 24(1), 617 (2022) 28. Ferri, C., et al.: MiR-423-5p prevents MALAT1-mediated proliferation and metastasis in prostate cancer. J. Exp. Clin. Cancer Res. 41(1), 1–6 (2022) 29. Tang, X., et al.: MiR-423-5p serves as a diagnostic indicator and inhibits the proliferation and invasion of ovarian cancer. Exp. Ther. Med. 15(6), 4723–4730 (2018) 30. Shang, Y., et al.: Downregulation of miR-423–5p contributes to the radioresistance in colorectal cancer cells. Front. Oncol. 11(10), 582239 (2021) 31. Stiuso, P., et al.: MicroRNA-423-5p promotes autophagy in cancer cells and is increased in serum from hepatocarcinoma patients treated with sorafenib. Mol. Ther. Nucl. Acids. 1(4), e233 (2015) 32. Li, S., Zeng, A., Hu, Q., Yan, W., Liu, Y., You, Y.: MiR-423-5p contributes to a malignant phenotype and temozolomide chemoresistance in glioblastomas. Neuro Oncol. 19(1), 55–65 (2017) 33. Wang, Y., et al.: Profiles of differential expression of circulating microRNAs in hepatitis B virus-positive small hepatocellular carcinoma. Cancer Biomark. 15(2), 171–180 (2015)

244

L. Kadri´c et al.

34. Wu, W.R., et al.: Methylation—associated silencing of miR-200b facilitates human hepatocellular carcinoma progression by directly targeting BMI1. Oncotarget 7(14), 18684 (2016) 35. Liu, M., et al.: Association of serum microRNA expression in hepatocellular carcinomas treated with transarterial chemoembolization and patient survival. PLoS ONE 9(10), e109347 (2014) 36. Xu, C., et al.: LncRNA small nucleolar RNA host gene 12 promotes renal cell carcinoma progression by modulating the miR-200c-5p/collagen type XI α1 chain pathway. Mol. Med. Rep. 22(5), 3677–3686 (2020) 37. Chen, H., Zhang, J., Yang, L., Li, Y., Wang, Z., Ye, C.: Circ-ZEB1 regulates epithelialmesenchymal transition and chemotherapy resistance of colorectal cancer through acting on miR-200c-5p. Transl. Oncol. 1(28), 101604 (2023) 38. Wang, D.J., Legesse-Miller, A., Johnson, E.L., Coller, H.A.: Regulation of the let-7a-3 promoter by NF-κB. PLoS ONE 7(2), e31240 (2012) 39. Mizuno, R., Kawada, K., Sakai, Y.: The molecular basis and therapeutic potential of Let-7 microRNAs against colorectal cancer. Can. J. Gastroenterol. Hepatol. (2018) 40. Süslüer, S.Y, Biray Avci, Ç., Si˘ ¸ gva Zö., Balci, T., Kayaba¸si, Ç., akalin, T., Dalbasti, T., Gündüz, C.: Detection of miRNA expression alteration in diffuse and high grade glial tumors. J. Neurol. Sci. 32(1) (2015) 41. Brueckner, B., et al.: The human let-7a-3 locus contains an epigenetically regulated microRNA gene with oncogenic function. Can. Res. 67(4), 1419–1423 (2007) 42. Li, Y., et al.: Overexpressed let-7a-3 is associated with poor outcome in acute myeloid leukemia. Leuk. Res. 37(12), 1642–1647 (2013) 43. Lu, L., Katsaros, D., Rigault de la Longrais, I.A., Sochirca, O., Yu, H.: Hypermethylation of let- 7a-3 in epithelial ovarian cancer is associated with low insulin-like growth factor-II expression and favorable prognosis. Cancer Res. 67(21), 10117–22 (2007) 44. Liang, H., et al.: MicroRNA-744–5p suppresses tumorigenesis and metastasis of osteosarcoma through the p38 mitogen-activated protein kinases pathway by targeting transforming growth factor-beta 1. Bioengineered 13(5), 12309–12325 (2022) 45. Zhao, L.G., Wang, J., Li, J., Li, Q.F.: MiR-744-5p inhibits cellular proliferation and invasion via targeting ARF1 in epithelial ovarian cancer. Kaohsiung J. Med. Sci. 36(10), 799–807 (2020) 46. Deng, Y., Li, Y., Fang, Q., Luo, H., Zhu, G.: MicroRNA-744 is downregulated in glioblastoma and inhibits the aggressive behaviors by directly targeting NOB1. Am. J. Cancer Res. 8(11), 2238 (2018) 47. Fan, F., Yao, D., Yan, P., Jiang, X., Hu, J.: MicroRNA-744-5p inhibits glioblastoma malignancy by suppressing replication factor C subunit 2. Oncol. Lett. 22(2), 1–2 (2021) 48. Pourhanifeh, M.H., et al.: Autophagy in cancers including brain tumors: role of MicroRNAs. Cell Commun. Signal. 18, 1–22 (2020) 49. Zhao, J., et al.: MicroRNA-7: a promising new target in cancer therapy. Cancer Cell Int. 15(1), 1–8 (2015) 50. Han, W., Cui, H., Liang, J., Su, X.: Role of MicroRNA-30c in cancer progression. J. Cancer 11(9), 2593 (2020) 51. Zhang, H., Lu, Y., Wang, S., Sheng, X., Zhang, S.: MicroRNA-152 acts as a tumor suppressor microRNA by inhibiting Krüppel-like factor 5 in human cervical cancer. Oncol. Res. Featur. Preclin. Clin. Cancer Ther. 27(3), 335–340 (2019) 52. Ma, J., et al.: MiR-152 functions as a tumor suppressor in glioblastoma stem cells by targeting Krüppel-like factor 4. Cancer Lett. 355(1), 85–95 (2014) 53. Zhu, X., Shen, Z., Man, D., Ruan, H., Huang, S.: MiR-152-3p Affects the progression of colon cancer via the KLF4/IFITM3 axis. Comput. Math. Methods Med. 1, 2020 (2020)

Analysis of MicroRNAs in Correlation to Astrocytoma

245

54. Ramalho-Carvalho, J., et al.: A multiplatform approach identifies miR- 152–3p as a common epigenetically regulated onco-suppressor in prostate cancer targeting TMEM97. Clin. Epigenetics 10, 1–5 (2018) 55. Sun, J., et al.: Regulation of human glioma cell apoptosis and invasion by miR-152-3p through targeting DNMT1 and regulating NF2. J. Exp. Clin. Cancer Res. 36(1), 1–3 (2017) 56. Wang, F., Ren, X., Zhang, X.: Role of microRNA-150 in solid tumors. Oncol. Lett. 10(1), 11–16 (2015) 57. Jiang, X., et al.: Blockade of miR-150 maturation by MLL-fusion/MYC/LIN-28 is required for MLL—associated leukemia. Cancer Cell 22(4), 524–535 (2012) 58. Sakr, M., Takino, T., Sabit, H., Nakada, M., Li, Z., Sato, H.: MiR-150-5p and miR-133a suppress glioma cell proliferation and migration through targeting membrane-type-1 matrix metalloproteinase. Gene 587(2), 155–162 (2016) 59. Tian, W., Zhu, W., Jiang, J.: MiR-150-5p suppresses the stem cell-like characteristics of glioma cells by targeting the Wnt/β-catenin signaling pathway. Cell Biol. Int. 44(5), 1156–1167 (2020) 60. Deborde, S., et al.: Schwann cells induce cancer cell dispersion and invasion. J. Clin. Investig. 126(4), 1538–1554 (2016)

How Inherited Thrombophilia Affects Success Rate of IVF Treatment in Women Damilola M. Ajayi(B)

and Emmanuel Ajayi

International Burch University, Francuske Revolucije Bb, 71210 Ilidza, Bosnia and Herzegovina {damilola.mildred.ajayi, Emmanuel.oluwatosin.ajayi}@stu.ibu.edu.ba

Abstract. The inclination towards conception, is one that has proven to be innate to human beings through generations. Unfortunately, as natural and earnest a desire it might be, so many willing couples and individuals find it difficult and sometimes impossible to conceive naturally. With the advancement of medical science, significant headway has been made, enough to provide a solution for individuals or couples who are unable to conceive naturally, using a series of technique called Assisted Reproductive Technique (ART) which refers to any treatment or procedure for assisting reproduction that includes the handling of human eggs, sperm, or embryos. In-vitro fertilization (IVF) is an example of an assisted reproductive method. In recent years, IVF has steadily become the most popular assisted reproductive procedure and, in most cases, the last resort for infertility therapy. Several factors have been identified as impacting IVF–embryo transfer success or failure rates. Factors to consider include age, parity, previous successful pregnancy, basal hormonal levels, number of antral follicles before stimulation, endometrial thickness, embryo grade, position and length of uterus, and embryo transfer technique. For decades, inherited thrombophilia has been suspected in spontaneous miscarriages during in vitro fertilization procedures. This work investigates the impact of inherited thrombophilia on IVF success rates. The IVF process and what causes couples to need IVF are discussed. The molecular characteristics of thrombophilia are thoroughly studied, and existing articles are evaluated and contrasted to provide additional clarity. Finally, a conclusion is reached regarding the effect of inherited. Keywords: Inherited thrombophilia · IVF · Spontaneous abortion

1 Introduction 1.1 IVF and Inherited Thrombophilia Medical science has made significant advancement, enough to provide a solution for individuals or couples who are unable to conceive naturally, using a series of technique called assisted reproductive technique (ART) which refers to any treatment or procedure for assisting reproduction that includes the handling of human eggs, sperm, or embryos [1]. In vitro fertilization (IVF) is an example of assisted reproductive tech- nique that has © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 246–252, 2024. https://doi.org/10.1007/978-3-031-49068-2_27

How Inherited Thrombophilia Affects Success

247

gradually become a common technique in assisted reproductive technol- ogy. IVF at the beginning is routinely carried out in an artificial environment outside the biological body, in contrast to the conventional method where the egg and sperm are fertilized inside the female’s body [2]. Inherited thrombophilia is due to complete genetic thrombophilias causes like Protein C deficiency, Protein S deficiency, An- tithrombin III deficiency, mutation in factor V gene, Prothrombin gene mutation. There are also partial genetic thrombophilia causes, such as high levels factor VIIIc, and mild hyperhomocysteinemia [3]. 1.2 Objective of Research The objective of this paper was to study the correlation between hereditary thrombophilia and spontaneous abortion rates in women with inherited thrombophilia who have undergone IVF procedures, and to determine if indeed hereditary thrombophilia hinders the successful fertilization of the embryo. When determining a successful pregnancy term, many factors are considered, including age, weight, genetics, race, hormonal levels, and so on. Until a few decades ago, the role of genetics was overlooked, most likely due to a lack of in-depth understanding of how genes work at the molecular level, but due to scientific advancement, the importance of genetics has been widely established. A comprehensive search of previously published literature was done, followed by the literature review on in vitro fertilization, inherited thrombophilias and their interactions with different genes and cell receptors. Different studies carried out on various groups of women with inherited thrombophilia were reviewed to determine if the theory was correct, the results were expatiated and discussed, and conclusion was drawn as regards the role of inherited thrombophilia in the success rate of an IVF. In case studies where the results were inconclusive, other factors were taken into consideration in their working with inherited thrombophilia and their effects on IVF success rate or spontaneous abortion rates.

2 Methods and Materials 2.1 Materials Forty potentially relevant studies were identified with the use of search tools like NCBI PubMed, ResearchGate, and Google Scholar. Papers published from 1987 to 2022, a wide time frame was used to get more insight into the relationship between IVF and inherited thrombophilia. A broad search strategy including combinations of search terms such as: thrombophilia, IVF, assisted reproductive technique, miscarriage, etc. The inclusion criteria used were as follows: Scientific articles written in the English language, Fulltext, free scientific articles, and Scientific articles that focused on IVF and inherited thrombophilia. The exclusion criteria used were as follows: Scientific articles written in any other language, articles without free access, and Scientific articles that were not focused on IVF and inherited thrombophilia (Table 1).

248

D. M. Ajayi and E. Ajayi

2.2 Methods These articles were analyzed critically to understand the process of IVF, and the relationship between IVF and spontaneous abortion rate in women who had inherited thrombophilia. The diagnostic tests such as ovarian reserve testing, Prolactin and thyroid hormone test, an ultrasound test, and a hystero-salpingography test, carried out to determine the eligibility of the female before the process of IVF began were stated and explained in detail. The necessary procedures followed in an IVF technique were shown to aid understanding of the process and prevent complexity. Studies carried out to realize the correlation between both were researched on to test the hypothesis and determine if there was any significance. The genetic components of inherited thrombophilia such as Protein C deficiency, Protein S deficiency, Antithrombin III deficiency, mutation in factor V gene, Prothrombin gene mutation, were studied Table 1 An overview of Inherited thrombophilia impact on IVF success rate in women. Inherited thrombophilia effect on IVF

Major findings

References

Thrombophilia and pregnancy complications

Women with both inherited and acquired thrombophilia are at increased risk of developing both early and late complications in pregnancy. However, the absolute risk of these adverse outcomes remains low

Simcox et al. [4]

Effects of multiple inherited and acquired thrombophilia on outcomes of in-vitro fertilization

The study concluded that in women Di Nisio et al. [5] undergoing IVF, the effects of multiple thrombophilic defects on outcomes of in-vitro fertilization remain unclear. The presence of two or more thrombophilic de fects was rare and showed no statistically significant associations with IVF out-comes

Association between in vitro fertilization outcomes and inherited thrombophilias

There was no significant difference in patients with inherited thrombophilia mutations in comparison with non-mutation carrier cases

Factor v Leiden and prothrombin gene g20210a mutation and in vitro fertilization: prospective cohort study. Human reproduction

IR were not significantly different Ricci et al. [7] between carrier and non-mutation carrier women before and after adjustment for female age, infertility diagnosis and the number of IVF cycles

Tan et al. [6]

(continued)

How Inherited Thrombophilia Affects Success

249

Table 1 (continued) Inherited thrombophilia effect on IVF

Major findings

References

Effects of obesity and thrombophilia on the risk of abortion in women undergoing in vitro fertilization. Frontiers in endocrinology

The results showed that compared to Candeloro et al. [9] patients without thrombophilia, women with hereditary thrombophilia had a five-fold higher risk of abortion. The risk seemed increased also in women with acquired thrombophilia, but the difference did not reach statistical significance

Acquired And Inherited Thrombophilia: Implication in Recurrent IVF And Embryo Transfer Failure

Women with repeated IVF–embryo Qublan et al. [8] transfer failure should be screened for thrombophilia. Prospective randomized controlled interventional studies with large numbers are needed to determine the effect of thromboprophylaxis in such cases

at molecular level, to understand the molecular characteristics and its relation to their effect on successful implantation rate or spontaneous abortion.

3 Results According to a review by Tan and colleagues, the success of given IVF cycle is confirmed by implantation. Meaning, if the implantation is successful, the chances of successfully carrying the pregnancy to full term is not affected by thrombophilia. In a study carried out by Di Nisio and colleagues [5], on 687 women with the mean age of 34, the study concluded that in women undergoing IVF, the effects of multiple thrombophilic defects on outcomes of in-vitro fertilization remain unclear. The presence of two or more thrombophilic defects was rare and showed no statistically significant associations with IVF outcomes [5]. In another meta-analysis carried out by Tan and others [6], with the aim to determine whether IVF outcomes are associated with inherited thrombophilias, there was no significant difference in patients with inherited thrombophilia mutations in comparison with non-mutation carrier cases suggesting no associations between IVF and inherited thrombophilias [6]. Ricci and colleagues [7], found that IR were not significantly different between carrier and non-mutation carrier women before and after adjustment for female age, infertility diagnosis and the number of IVF cycles. Thus, the conclusion was that inherited thrombophilias did not affect embryo implantation [7].

250

D. M. Ajayi and E. Ajayi

Fig. 1. Risk of abortion in non-obese and obese patients according to the presence of thrombophilia [9].

According to a study carried out by Candeloro et al. [9], on 682 women including 633 (92.8%) non-obese and 49 (7.2%) obese women. to determine the effects of obesity and thrombophilia on the risk of abortion in women undergoing in vitro fertilization, the results showed that compared to patients without thrombophilia, women with hereditary thrombophilia had a fivefold higher risk of abortion. The risk seemed increased also in women with acquired thrombophilia, but the difference did not reach statistical significance [9]. The findings support previous findings that the risk of abortion is increased in obese women who also have thrombophilia. Obesity and thrombophilia may act synergistically to activate blood coagulation beyond the physiological state of hypercoagulability associated with pregnancy, potentially increasing the risk of placental vessel thrombosis, pregnancy loss, and pregnancy complications. In another research by Simcox et al. [4], women with both inherited and acquired thrombophilia are at increased risk of developing both early and late complications in pregnancy. However, the absolute risk of these adverse outcomes remains low [4]. Finally, the results of another study carried out by Qublan et al. [8] on acquired and inherited thrombophilia and its’ implication in recurrent IVF and embryo transfer fail- ure, indicates that thrombophilia may have a significant role in IVF–embryo transfer implantation failure. Women with repeated IVF–embryo transfer failure should be screened for thrombophilia. Prospective randomized controlled interventional studies with large numbers are needed to determine the effect of thromboprophylaxis in such cases [8].

How Inherited Thrombophilia Affects Success

251

4 Discussion From the review carried out on the different research done in the research papers used in this project, it is not farfetched to say that the risk of inherited thrombophilia on IVF success, is mostly inconclusive. Although numerous studies have been conducted to support the theory that thrombophilia has a negative impact on the success rate of IVF procedures in women, it remains, broadly speaking, a theory and speculation lacking sufficient evidence to determine how definite this theory is Other than correlations formed from coincidences and theorized assumptions, there is no specific factor in play that links thrombophilia to unsuccessful IVF procedures. Asides from this, it can also be deduced from this research that other factors come into play along with inherited thrombophilia, for example, obesity. In Fig. 1, it is seen that the rate of pregnancy loss in thrombophilia patients is higher in patients who were medically obese. This is another fact that shows that thrombophilia might not be the link to complications during the procedure of IVF in women. From the review carried out, it is not farfetched to say that the risk of inherited thrombophilia on IVF success, is mostly inconclusive.

References 1. National Academy of Sciences (US), National Academy of Engineering (US), Institute of Medicine (US) and National Research Council (US) Committee on Science, Engineering, and Public Policy. Scientific and Medical Aspects of Human Reproductive Cloning. Washington (DC): National Academies Press (US). Assisted Reproductive Technology, vol 4 (2002) 2. Nisal, A., Diwekar, U., Bhalerao, V.: Personalized medicine for in vitro fertilization procedure using modeling and optimal control. J. Theor. Biol. 487(110105) (2020) 3. Tsikouras, P., Deftereou, T., Anthoulaki, X., Bothou, A., Chalkidou, A., Christoforidou, A., Chatzimichael, E., Gaitatzi, F., Tsirkas, I., Bourazan, A.C., Bampageorgaka, E., Iatrakis, G., Zervoudis, S., Rath, W., Galazios, G.; Thrombophilia and pregnancy: diagnosis and management. In: Stawicki, S.P., Firstenberg, M.S., Swaroop, M. (eds.) Embolic Diseases Evolving Diagnostic and Management Approaches. Intechopen (2019) 4. Simcox, L.E., Ormesher, L., Tower, C., Greer, I.A., Thrombophilia and Pregnancy Complications. Int. J. Mol. Sci., J. 16(12), 28418–28428 (2015) 5. Di Nisio, M., Ponzano, A., Tiboni, G.M., Guglielmi, M.D., Rutjes, A., Porreca, E.: Effects of multiple inherited and acquired thrombophilia on outcomes of in-vitro fertilization. Thromb. Res. 167, 26–31 (2018) 6. Tan, X., et al.: Association between in vitro fertilization outcomes and inherited thrombophilias: a meta-analysis. J. Assist. Reprod. Genet. J. 33(8), 1093–1098 (2016) 7. Ricci, G., Bogatti, P., Fischer-Tamaro, L., Giolo, E., Luppi, S., Montico, M., Ronfani, L., Morgutti, M.: Factor V Leiden and prothrombin gene G20210A mutation and in vitro fertilization: prospective cohort study. Hum. Reprod. J. 26(11), 3068–3077 (2011). (Oxford, England) 8. Qublan, H.S., Eid, S.S., Ababneh, H.A., Amarin, Z.O., Smadi, A.Z., Al-Khafaji, F.F., Khader, Y.S.: Acquired and inherited thrombophilia: implication in recurrent IVF and embryo transfer failure. Human reproduction. Journal 21(10), 2694–2698 (2006). Oxford, England

252

D. M. Ajayi and E. Ajayi

9. Candeloro, M., Di Nisio, M., Ponzano, A., Tiboni, G.M., Potere, N., Tana, M.., Rutjes, A., Porreca, E.: Effects of obesity and thrombophilia on the risk of abortion in women undergoing in vitro fertilization. Front. Endocrinol. 594867, 11 (2020)

Examining the Therapeutic Potential of Stem Cells in Treatment of Infertility Lejla Hadži´c1(B) , Sara Sejdi´c1 , and Faruk Guti´c2 1 International Burch University, Sarajevo, Bosnia and Herzegovina

[email protected] 2 Medical Faculty, University of Sarajevo, Sarajevo, Bosnia and Herzegovina

Abstract. Millions of couples worldwide are affected by infertility, which is becoming a bigger health concern. Both men and women can experience infertility, which has a variety of causes including hormonal imbalances, genetic disorders, and anatomical anomalies. Although there are many therapeutic options, including assisted reproductive technologies, these procedures are frequently expensive and may only be partially successful. This paper presents an overview of possibilities for using stem cell therapy for infertility. The review focuses on identification of the sources of the stem cells, the delivery techniques, and the potential advantages and disadvantages of this strategy. The paper’s overall goal is to give a summary of recent findings in this area and to highlight stem cell therapy’s potential as a promising method of treating infertility. Keywords: Stem cell therapy · Infertility therapy · Assisted reproductive technologies · Hormonal imbalances

1 Introduction According to research, 8–12% of couples worldwide experience infertility issues. After a year of routine, unprotected sexual activity, a couple is said to be infertile if they are unable to conceive. There are many potential causes, including a mix of both male and female factors. Hormonal imbalances, structural abnormalities, genetic disorders, and lifestyle choices are just a few of the causes of infertility [1]. Different methods can be used to treat infertility. One popular technique is assisted reproductive technology (ART), which includes processes like intracytoplasmic sperm injection (ICSI) and in vitro fertilization (IVF), which combine eggs and sperm in a lab before implanting the resulting embryo in the uterus. Additionally, to address the root causes of infertility and increase the likelihood of a successful conception, hormonal therapies, surgical procedures, and lifestyle changes may be used [2]. Infertility research is now also combining to explore the potential of stem cells as a promising route for creating new treatments to deal with reproductive issues. The body’s building blocks are stem cells, which are cells that give rise to all other cells with specific roles. No other cell in the body has the capacity to naturally produce different cell types [3]. Stem-cell therapy has given the field of managing reproductive disabilities new hope in recent years. Embryonic stem © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 253–260, 2024. https://doi.org/10.1007/978-3-031-49068-2_28

254

L. Hadži´c et al.

cells (ESCs), mesenchymal stem cells (MSCs), spermatogonial stem cells (SSCs), and induced-pluripotent stem cells are the four main types of stem cells (iPSCs). By regenerating ovarian tissue and enhancing ovarian function, MSCs can treat PCOS (polycystic ovary syndrome) and POF (premature ovarian failure). They can also transdifferentiate into Leydig cells and male germ cells to treat oligospermia and azoospermia. According to studies, MSC transplantation can improve hormonal profiles and restore menstruation in POF patients, as well as reduce inflammation and improve ovarian function in PCOS patients. Research into MSC therapy’s potential to treat infertility in both sexes is still in progress [4]. The goal of this review paper is to thoroughly examine stem cells’ potential as a game-changing treatment for infertility. This paper aims to provide a comprehensive overview of the current understanding of how stem cells can be used to restore or enhance fertility by synthesizing existing research and developments in the field. This review focuses on four infertility conditions as an example which are polycystic ovary syndrome, premature ovarian failure, azoospermia and oligospermia. In the end, this paper seeks to add to the growing body of reproductive medicine knowledge and encourage additional investigation and research in this rapidly developing field.

2 Methods A search was conducted to identify relevant studies on the therapeutic potential of stem cells in the treatment of infertility. Electronic databases, including PubMed, Scopus, and Google Scholar, were systematically searched using relevant keywords and their combinations. The search terms used included “stem cells,” “infertility,” “therapeutic potential,” “regenerative medicine,” and “reproductive medicine.” The search was limited to studies published in the English language from 2001 to present. The inclusion criteria for study selection were as follows: (1) studies investigating the therapeutic use of stem cells in the treatment of infertility, (2) studies involving human subjects, (3) peer-reviewed articles published in scientific journals, and (4) studies reporting outcomes related to reproductive health and fertility restoration. Studies that did not meet these criteria or were duplicates, review articles, opinion pieces, or conference abstracts were excluded. Data were extracted from the selected studies using a standardized form. The following information was extracted: study characteristics (e.g., authors, publication year, study design), participant characteristics (e.g., sample size, age, infertility etiology), type of stem cells used, therapeutic approaches employed, and key outcomes related to infertility treatment. A narrative synthesis approach was used to summarize and analyze the extracted data. The findings from the selected studies were grouped according to stem cell types, therapeutic approaches, and outcomes reported. No quantitative meta- analysis was performed due to the heterogeneity of the included studies. This review study did not involve direct data collection from human subjects; thus, ethical approval was not required. The review adhered to ethical guidelines and principles, ensuring confidentiality, privacy, and proper citation of the included studies.

Examining the Therapeutic Potential of Stem Cells

255

3 Results The findings of this review’s analysis of pertinent studies on the therapeutic potential of stem cells in the management of infertility serve as the foundation for its conclusions. The conclusions came from a thorough search and systematic analysis of peer-reviewed articles that were released in academic journals. At Al-Azhar University and the Al Azhar Regenerative Medicine International Center (ARMIC), researchers carried out a prospective study. The purpose of the study was to evaluate the efficacy of autologous mesenchymal stem cell (MSC) transplantation in premature ovarian failure (POF) patients. Laparoscopic ovarian autologous MSC transplantation was performed on ten POF patients using MSCs taken from the bone marrow of the iliac crest. The patients’ clinical outcomes, immunohistochemical results, and histopathological findings were assessed over a three to eleven month period [4]. The effects of human umbilical cord mesenchymal stem cells (hUC MSCs) in a PCOS model induced by DHEA injections were examined in the second study using female C57/BL6 mice. The control group received regular saline, while the PCOS model group underwent hUC-MSC transplantation. After seven days of hUC-MSC therapy, estrous cycle evaluation and histological analysis were carried out [5]. The third investigation focused on male rat azoospermia. Mesenchymal stem cells derived from adipose tissue (rAT-MSCs) were isolated, cultured, and their properties were assessed using a variety of methods, such as flow cytometry, immunostaining, and in vitro differentiation. MSCs were injected into the testicles, and the results regarding infertility were evaluated [6]. In the fourth study, busulfan injections were used to create a rat model of oligospermia. The rats were divided into several groups, including a control group, a group for the spermatogenic arrest model, a group that wasn’t treated, and a group that received treatment with BM-MSCs that were GFP-labeled. To assess the effects of BM-MSC treatment, testicular parameters, histological staining, and biochemical tests were carried out [7]. The findings from the articles under review show how effective stem cell treatments can be for treating conditions linked to infertility. Autologous mesenchymal stem cell (MSC) transplantation demonstrated promising results in the study focusing on premature ovarian failure (POF) patients. Positive clinical outcomes and favorable histopathological and immunohistochemical results followed the transplantation of MSCs into the ovaries. Similar to this, the injection of human umbilical cord MSCs (hUC-MSCs) improved the estrous cycle and showed positive effects on ovarian and uterine tissues in the polycystic ovary syndrome (PCOS) model. Additionally, azoospermia and oligospermia have been addressed with the help of rat adipose tissue-derived MSCs (rAT-MSCs). Testicular weight, sperm count, motility, and markers of apoptosis and inflammation all improved after adipogenic differentiation and bone marrow MSC transplantation (BMMSCs). The potential of stem cell-based therapies as a promising strategy for the treatment of infertility is highlighted by these findings. To determine the security, effectiveness, and long-term effects of stem cell-based interventions in infertility management, additional study and clinical trials are required.

256

L. Hadži´c et al.

4 Infertility in Women and Treatment with Stem Cells 4.1 Premature Ovarian Failure Many causes, including as hormone imbalances, anatomical abnormalities, and genetic illnesses, might contribute to infertility in women. Premature ovarian failure (POF), which happens when a woman’s ovaries stop operating normally before the age of 40, is one probable factor in infertility. Genetic abnormalities, autoimmune diseases, chemotherapy, and radiation therapy are only a few of the causes of POF. Ten cases of POF were planned for laparoscopic ovarian autologous MSC transplantation as part of the trial at the Al-Azhar University hospitals and Al Azhar Regenerative Medicine International Center (ARMIC).To confirm proper karyotyping, women were assessed via medical history, physical examinations, laboratory tests, and chromosomal analysis.Candidates for mesenchymal stem cell transplantation, post menarche females under the age of 40, and FSH levels more than or equivalent to 20 IU/L with normal karyotyping were required for inclusion. Breast cancer and ovarian cancer were disqualifying conditions. For women with POF, stem cell therapy has become a possible treatment option. Unspecialized cells called stem cells have the capacity to differentiate into a variety of cell types, including those in the ovaries [8]. Stem cells are implanted into the ovaries as part of stem cell therapy for POF in order to rebuild ovarian tissue and return ovarian function. MSCs were made from a 10 ml sample taken from the iliac crest bone marrow and injected into the ovaries by laparoscopy. Following transplantation, two instances who had atrophic endometrium and distorted proliferative endometrial both produced focused secretory endometrium that showed positive stem cell expression and started menstruation. One of them conceived a child. After transplantation, 100% of ESS-5 cases started menstruation, and one of them became pregnant. The POF patient’s hormonal profile was improved by stem cell transplantation, which also brought about the return of menstruation, a pregnancy, and the delivery of a fully developed, healthy child [9]. 4.2 Polycistic Ovarian Syndrome Insulin resistance, type 2 diabetes, obesity, hyperlipidemia, and elevated cardiovascular risk are all linked to PCOS. Although the pathogenesis and etiology of PCOS are yet unknown, it is thought to be an autoimmune and chronic inflammatory illness. Clinical recommendations recommend making lifestyle modifications or taking medications as the only viable treatments for PCOS at the moment. Mesenchymal stem cells (MSCs), which have immunomodulatory capabilities, have been proven in trials to be a possible new treatment for severe autoimmune disorders that are resistant to current therapies. The ovarian and uterine pathological changes in PCOS mice are significantly improved by hUC- MSC treatment, according to this study’s assessment of the therapeutic effects of human umbilical cord-derived MSCs (hUC-MSCs). This is accomplished by inhibiting both local and systemic inflammatory responses. It appears that the experiment was successful. Proinflammatory cytokines, activated T cells, and white blood cells were

Examining the Therapeutic Potential of Stem Cells

257

present in greater quantities in PCOS-affected women. According to the study, hUCMSC therapy effectively reduced the pathological changes in PCOS mice, including polycystic ovaries and an irregular estrous cycle, and enhanced ovarian function. MSCs have the ability to change the polarization of macrophages, decreasing peripheral M1 macrophages while increasing M2 macrophages. They can also reduce the production of IFN- and inhibit T cell proliferation, which both suppress the proinflammatory state [10].

5 Infertility in Man and Treatment with Stem Cells 5.1 Azoospermia A noteworthy advancement in stem cell-based therapeutics is the research of mesenchymal stem cells (MSCs) generated from adipose tissue to cure azoospermia in rats. Azoospermia is a disorder that prevents males from having sperm in their semen and can cause infertility [11]. According to the study, injecting GFP+ MSCs into the left rete testes of rats caused spermatogenesis in a few tubules, pointing to the possibility of using MSCs to treat azoospermia in people. Mesenchymal stromal cells (AT-MSCs) generated from adipose tissue have demonstrated potential in a variety of tissue cell transdifferentiation processes. Recent research has shown that MSCs from bone marrow and adipose tissue can develop into Leydig cells and male germ cells, which are steroidogenic cells. The treatment of male infertility and low testosterone may benefit from this [12]. In the seminiferous tubules, the spermatogenesis process converts spermatogonial stem cells (SSCs) into spermatozoa. The basal compartment, where SSCs are found, promotes controlled proliferation and fulfills the function of stem cells in the body. The research discovered that rAT-MSCs injected into the testes were located both inside and outside of the seminiferous tubules, and that when they came into contact with the Sertoli cells’ cytokines and the niche within the tubules, they were able to transdifferentiate into SSCs and begin the spermatogenesis process. The testis of male rats treated with busulfan for infertility were successfully transplanted with AT-MSCs, restoring the males’ fertility. MSCs may directly contribute to spermatogenesis (via transdifferentiation) or indirectly through interactions with the testis tissue niche. According to the study, MSCs made from adipose tissue may one day be effectively used to treat male infertility [13, 14]. 5.2 Oligospermia Low sperm counts in semen are the hallmark of the medical disorder known as oligospermia, which can lead to male infertility. It is one of the most frequent root causes of male infertility and can be brought on by a number of conditions, including endocrine problems, systemic diseases, testicular trauma, environmental contamination, and other conditions. Oligospermic infertile men’s testicular biopsies typically have unique histological characteristics. The ability of bone marrow mesenchymal stem cells (BM-MSCs) to restore spermatogenic arrest in oligospermic rats was examined by researchers. They separated the rats into four groups after inducing an oligospermia model in them. The

258

L. Hadži´c et al.

testicular oxidative stress and germ cell death in the oligospermic rats were significantly higher than those in the control group. The induced spermatogenic arrest in the rat model was significantly improved by intra testicular injection of BM-MSCs, which did so primarily through anti-apoptotic and antioxidant paracrine actions. Around 1 in 20 guys experience male infertility, which can be brought on by a number of conditions, including systemic illnesses, environmental toxins, endocrine problems, and testicular injuries. One of the often occurring causes of male infertility is oligospermia. Most pharmacological and hormonal treatments for male infertility have disappointing results. BM-MSCs offer hope for novel and successful treatments [15, 16].

6 Prediction for Future of Stem Cell Therapy Scientists anticipate to: 1. Improve knowledge about disease etiology. Researchers may gain a better understanding of the progression of diseases and ailments by observing stem cells evolve into cells found in bones, heart muscle, neurons, and other organs and tissue. 2. Create disease-free cells to replace unhealthy ones (regenerative medicine). It is possible to direct stem cells to differentiate into particular cells that can be employed in individuals to regenerate and restore tissues that have been harmed or impacted by illness. 3. Check the effectiveness and safety of new medications. Certain kinds of stem cells can be used by researchers to test new medications for quality and safety prior to testing them on humans. Some potential problems; To be useful, stem cells must differentiate into the appropriate cell types, according to researchers. Regulating their proliferation and differentiation, as well as avoiding potential problems like immunological reactions or malfunction, remain difficult tasks. There are still many unexplored aspects of stem cell therapy, which means there is still a great deal of unrealized potential for its use in treating conditions like infertility [17]. 6.1 Stem Cells as a New Opportunity for Infertility Therapy One-third of cases of infertility are due to male factors, which include precocious puberty, hereditary illnesses, structural issues like testicular blockage, genital damage or injury that results in sperm dysfunction, and environmental and psychological factors. Ovulation dysfunction, abnormalities in the uterus or fallopian tubes, endometritis, primary ovarian insufficiency, and pelvic adhesions are all female specific risk factors [18, 19]. MSCs can treat POF and PCOS in women by regenerating ovarian tissue and restoring ovarian function. Studies show that MSC transplantation can improve hormonal profiles, restore menstruation, and enable pregnancy. MSCs from adipose tissue also show promise in treating azoospermia and oligospermia by aiding in spermatogenesis through transdifferentiation into Leydig cells and male germ cells [20].

Examining the Therapeutic Potential of Stem Cells

259

7 Conclusion Stem cell therapy is becoming a major game changer for medicine after decades of research. The potential of stem cells is expanding with every experiment, but there are still numerous challenges to be solved. Nonetheless, stem cells have a significant impact on transplantology and regenerative medicine. Both male and female infertility can be treated with stem cell therapy, which has the potential to regenerate damaged reproductive tissues. Hope for more development and better infertility treatments can be found in the field’s ongoing research.

8 Limitations It is important to acknowledge that this review is subject to certain limitations. The inclusion of only English language publications and the exclusion of unpublished studies and gray literature may introduce a potential source of publication bias. Additionally, the heterogeneity of the included studies may limit the ability to draw definitive conclusions.

References 1. Drukker, M.: Recent advancements towards the derivation of immune-compatible patient specific human embryonic stem cell lines. Semin. Immunol. 20(2), 123–129 (2008). [PubMed] [Google Scholar] [Ref list] 2. World Health Organization: Infertility. (2019) https://www.who.int/news-room/fact-sheets/ detail/infertility 3. National Institutes of Health: Stem Cell Basics. (2020). https://stemcells.nih.gov/info/basics 4. Edessy, M., Hosni, H.N., Shady, Y., Waf, Y., Bakr, S., Kamel, M.: Autologous stem cells therapy, The first baby of idiopathic premature ovarian failure. Acta Med. Int. 3, 19–23 (2016) https://www.actamedicainternational.com/article.asp?issn=23490578;year=2016;vol ume=3;issue=1;spage=19;epage=23;aulast=Edessy 5. March, W.A., Moore, V.M., Willson, K.J., Phillips, D.I.W., Norman, R.J., Davies, M.J.: The prevalence of polycystic ovary syndrome in a community sample assessed under contrasting diagnostic criteria. Hum. Reprod. 25(2), 544–551 (2010) 6. Lin, C.Y., Lee, B.S., Liao, C.C., Cheng, W.J., Chang, F.M., Chen, M.H.: Transdiffer entiation of bone marrow stem cells into acinar cells using a double chamber system. J. Formos. Med. Assoc. 106(1), 1–7 (2007) 7. Microcystin-leucine arginine mediates apoptosis and engulfment of Leydig cell by testicular macrophages resulting in reduced serum testosterone levels Aquat. Toxicol. (2018). https:// www.sciencedirect.com/science/article/abs/pii/S0940960221000765 8. Jia, R., Gao, M., Yu, Z., Yao, D., Qian, Y., Wang, Q.: Mesenchymal stem cells therapy: a promising method for the treatment of uterine scars and premature ovarian failure. Tissue Cell. (2021). https://www.sciencedirect.com/science/article/pii/S0040816621001920?via% 3Dihub [Google Scholar] 9. Ding, L., Yan, G., Wang, B., et al.: Transplantation of UC-MSCs on collagen scaffold activates follicles in dormant ovaries of POF patients with long history of infertility. Sci. China Life Sci. 61(12), 1554–1565 (2018). https://doi.org/10.1007/s11427-017-9272-2.[PubMed][Cro ssRef][GoogleScholar]

260

L. Hadži´c et al.

10. Balen, A.H., Morley, L.C., Misso, M., et al.: The management of anovulatory infertility in women with polycystic ovary syndrome: an analysis of the evidence to support the development of global WHO guidance. Hum. Reprod. Update 22(6), 687–708 (2016). https://doi. org/10.1093/humupd/dmw025.[PubMed][CrossRef][GoogleScholar] 11. Li, H., Fang, F., Li, Z., Zhao, Q., Xiong, C.: Academic.oup.com (2018). https://academic. oup.com/humrep/article/33/2/188/4791141 12. Pruksananonda, R., Pramuan, R., Kamthorn, V.: Mesenchymal stem cells for restoring endometrial. Wiley online library (2020). https://doi.org/10.1002/rmb2.12339 [Google Scholar] 13. Katz, D.J., Teloken, P., Shoshany, O.: Male infertility—the other side of the equation. Aust. Fam. Phys. 46(9), 641–646 (2017). [PubMed] [Google Scholar] [Ref list] 14. Hajiesmailpoor, A., Emami, P., Jalali, B., Ghorbani, M.: Stem cell therapy as a recent advanced approach in male infertility. Tissue Cell (2021). https://www.sciencedirect.com/science/art icle/abs/pii/S0040816621001506 15. Smith, R.P., Lipshultz, L.I., Kovac, J.R.: Stem cells, gene therapy, and advanced medical management hold promise in the treatment of male infertility. Asian J. Androl. 18(3), 364 (2016). https://doi.org/10.4103/1008-682X.179249.[PMCfreearticle][PubMed][CrossR ef][GoogleScholar] 16. Schönfeldt, V., Meachem, S., Schlatt, S.: Spermatogonia: Stem Cells with a great perspective–researchgate (2001). https://www.researchgate.net/profile/Viktoria-Schoenfeldt/public ation/11965502_Spermatogonia_Stem_cells_with_a_great_perspective/links/0deec51f22d8 7baeeb000000/Spermatogonia-Stem-cells-with-a-great-perspective.pdforigin=publication_ detail 17. Zakrzewski, W., Dobrzy´nski, M., Szymonowicz, M., et al.: Stem cells: past, present, and future. Stem Cell Res. Ther. 10, 68 (2019). https://doi.org/10.1186/s13287-019-1165-5 18. Gunn, D.D., Bates, G.W.: Evidence-based approach to unexplained infertility: a systematic review. Fertil. Steril. 105(6), 1566–1574 (2016). https://doi.org/10.1016/j.fertnstert.2016.02. 001.[PubMed][CrossRef][GoogleScholar][Reflist] 19. 1Wang, J., Liu, C., Fujino, M., Tong, G., Zhang, Q., Li, X.-K., Yan, H.: Stem Cells as a resource for treatment of infertility-related diseases. Latest TOC RSS. (2019). https://www.ingentaco nnect.com/content/ben/cmm/2019/00000019/00000008/art00001 [Google Scholar] 20. Mouka, A., Tachdjian, G., Dupont, J., Drévillon, L., Tosca, L.: In vitro gamete differentiation from pluripotent stem cells as a promising therapy for infertility. Stem Cells Dev. 25(7), 509–521 (2016). https://doi.org/10.1089/scd.2015.0230.[PubMed][CrossRef][GoogleSchola r][Reflist]

An Overview of Personalized Medicine Development Through Recent Advances in Genome-Wide Association Studies Dženita Omerki´c(B)

and Adna Aši´c

International Burch University, Francuske Revolucije Bb, Ilidža 71210, Bosnia and Herzegovina [email protected]

Abstract. Genome-wide association studies, a term often in use following completion of The Human Genome Project, refers to the analysis and comprehensive understanding of human genetic landscape for practical purposes. An example of such application would be the development of personalized medicine approach and subsequent abandonment of the “one-size-fits all” model. In personalized medicine, a patient’s therapy is tailored according to their indication, clinical parameters, genotype and environmental effects in order to enable the best possible therapeutic outcome with as few adverse side effects, as possible. This review is about a series of exciting events that were important for the development of personalized medicine and to introduce the most important genome-wide association studies analysis approaches that made this possible, such as next-generation sequencing platforms and biomarker identification. In addition, we are presenting several examples of how personalized medicine improved our understanding of adverse drug reactions and ways to optimize patient’s therapy to their best interest. Keywords: Adverse drug reaction · Genome-wide association studies · Next-generation sequencing · Personalized medicine · Pharmacogenetics · Pharmacogenomics

1 Introduction The principle of personalized medicine (PM) is driven by the rule of prescribing the right dose of the right treatment at the right time for the patient, according to a definition from The United States Food and Drug Administration (FDA); it also encompasses referring to documented patient’s medical records, history, and genetic analysis under scrutiny [1]. The Human Genome Project (HGP) opened new challenges and options for scientific and health advancements after 2003, involving personalized medicine. In 2006, PM is represented by the US Genomics PM Act in a more detailed way, to level up patients’ recovery in a more precise way by following new methods [2]. Broad knowledge from interconnected scientific fields, such as pharmacogenetics and pharmacodynamics, are indispensable for effective treatment. Moreover, the focus is further related to the adverse drug reaction (ADR), e.g., in slow metabolizer patients [3, 4]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 261–274, 2024. https://doi.org/10.1007/978-3-031-49068-2_29

262

D. Omerki´c and A. Aši´c

The term ‘pharmacogenetics’ was invented in 1959 by Friedrich Vogel [5], delivering a reliable and successful interpretation of tests performed on the human genome [6], for the improvement of drug efficacy [7], whereby heredity presents the base for pharmacogenetics examination [8]. Primary care physicians have shown interest in pharmacogenomic testing (PGx testing), suggesting its importance for healthcare systems [9]. A few years ago, France invested 700 million dollars in mass sequencing centers, which is described as the first step in the PM strategy because it is impossible to work without such large-scale data. In contrast, China invested 10 trillion [10]. According to the National Institute of Health (NIH), six projects are realized for research in the area of the human genome and genetic medicine in the U.S. Significant progress and prosperity of personalized medicine are recorded in the U.S. when compared to other countries, which includes the presentation of a new biobank program called the Cohort Program of Precision Medicine Initiative (PMI). The main idea of this effort is to collect the basic information about genetic material, lifestyle, and samples of biological origin from one million participants involved. Demoraphic differences, for example ethnicity, were also reported [2]. In addition, U.S. approach involves the engagement of professionals from different scientific branches that work on the same goal and level up their education to be deeply involved in the concept of PM [3, 11]. BMBF (German: Bundesministerium für Bildung und Forschung; English: Federal Ministry of Education and Research) in Germany predicts that a successful outcome from PM can be only achieved if the disease is detected at an early stage. PGx testing enables an option for evaluation of the therapy effect on a specific gene or a group of genes [12]. Clinical Pharmacogenetics Implementation Consortium (CPIC) labeling of PGx-tested products defines dosage requirements regarding a patient’s genotype. For the full impact of PGx testing and implementation, it is important for physicians and everyone involved to be continually educated [3]. In the USA, due to incomplete diagnosis, communication, outdated information, and lack of detail-orientated approach towards the patients, 180,000–251,000 individuals die; since medical reports are not read, completed, or sorted, there was a breakdown in communication and inaccurate diagnosis [13]. Accuracy, economic profitability, clinical verifiability, and bioethical correctness are the key characteristics that PM strategy must contain to be enforced on a global level [14]. Challenges in the implementation of PM in healthcare are related to the lack of information technology (IT) infrastructure, interoperability, and data-related criteria, but also insufficient economic support and standard program for the global level [10]. See the Fig. 1. For an overview of the future of the modern medicine.

2 Objective and Methodology The purpose of this article was to write an overview of the development of personalized medicine through the development of technologies that are key and supporting factors of this remarkable scientific era and its strategy. Also, our goal was to clarify the action of certain drugs and the non-exclusive role of genetics in personalized medicine, so we described examples through pharmacogenomic testing and highlighted their importance. Our criteria included the following:

An Overview of Personalized Medicine Development

263

Fig. 1. The future of modern treatment represents improved personalized treatment that requires interoperability of medical and data science and modern technologies. A revolutionary change in healthcare is necessary in progressive treatment, diagnosis and monitoring of patients.

– Full available book/paper in English, – Research the following databases: Google Scholar, PubMed, Research Gate, and Science Direct, – Basic search terms we used: Personalized medicine, GWAS, Pharmacogenomics, Pharmacogenetics, and NGS.

3 Genome-Wide Association Study Genome-wide association study (GWAS) represents detailed human genome analysis concerning structures, combinations, properties, and phenotype monitoring. More than 15 years ago, when the human genome was sequenced, it was essential to steer toward a personalized medicine approach [10]. The era of GWAS started in 2005 [15] and the clinical relevance of GWAS is reflected in the identification of genetic variants associated with diseases (Ginsburg and Phillips [10]. GWAS studies contributed to the detection of risk loci in ethnic groups [16]. Epigenetics and epigenomics have their stake in engagement in precision medicine, especially because the environment affects the genetics of the population [17]. GWAS enables the detection of genomic loci bearing hypersensitivity and/or resistance to some drugs by analyzing candidate genes [18]. Genetic determinants, defined as quantitative traits, were also better understood by GWAS. An example is 34 identified loci associated with bone mineral density, resulting in better procedures for osteoporosis analysis. Another example is the relation of genes to signaling pathways in organisms or diabetes. Among mentioned complex traits, height is also controlled by many loci, approximately 180. All detected loci or genetic variants are useful and eligible for further research [19]. GWAS material is available on

264

D. Omerki´c and A. Aši´c

PheGenI (https://www.ncbi.nlm.nih.gov/gap/phegeni), Ensembl (https://www.ensembl. org/index.html), or UCSC Genome Browser (https://genome.ucsc.edu/). GWAS sample size or data is the biggest challenge upon other limitations [20], in addition to proper collection and interpretation of GWAS data [21]. Genome alignments provide comparison and detection of unknown locations of the human genome. Copy number variants (CNVs) are particularly interesting research material for different events during evolution [22]. The GWAS presents an indispensable tool in clinical analysis for disease detection caused by the expression of one or more genes. These are often different types of gene loci. Due to the various combinations among genes, GWAS depict necessary applications in medical research and healthcare [23]. 3.1 Next-Generation Sequencing (NGS) Next-generation sequencing (NGS) leads to significant progress in therapy development or molecular analysis in personalized medicine [24]. The NGS platforms, utilizing an approach of massively parallel sequencing [25], represent a high-throughput and highresolution technology that has revolutionized the field of genomics [26]. NGS platforms are the major achievement in scientific research and their advancement lately is due to the constant updates in bioinformatics and technology, contributing to the development of PM. NGS enables single apparatus to amplify and clone the enormous DNA amount, followed by reading the nucleotide sequence of the analyzed fragment. This concept, however, relies on the Sanger’s original idea [25]. Sanger’s method is the most popular sequencing technique due to its high accuracy and automatization procedures available, especially when it comes to the analysis of shorter DNA sequences. Many advancements beyond Sanger’s original idea have contributed to the progression of prognosis for patients at risk for certain diseases [27], including improvements in PCR amplification and enzymology [28], as well as the use of the capillary electrophoresis in DNA analysis and especially for STR analysis and Sanger sequencing [29]. Shankar Balasubramanian and David Klenerman are among the pioneers of the NGS platforms as the founders of the Illumina Solex idea. Ten years after 1997 and their first venture, the first appliances were available for purchase [25]. The significant advancement of Illumina is due to the fast technology that has improved camera performance during signal capture, polymerase, and other features of this method [30], with HiSeq and MiSeq being currently two most popular Illumina platforms [31]. Other technologies are equally popular and efficient in the last few years, including PacBio, used for real-time nucleotide sequencing [32], and Ion Torrent, which enables each incorporation of DNA nucleotide detection by pH measurement [30]. Laboratories should be fully equipped for NGS preparation work and data storage (meta-data) [26]. Based on the need for implementation and gathering of data, and information regarding the prescriptions and recordings, there is a need for a computational solution [33]. The aggregation of data to create an algorithm in decision-making related to the diagnosis or prescription of drugs is the first step in designing a model based on artificial intelligence (AI) [10]. There are algorithms for individual variants, or wholegenome comparisons, in addition to available exome analyses. However, this step in data

An Overview of Personalized Medicine Development

265

processing is still challenging due to the large amount of data and the need to improve precision in determining similarities and differences between samples [34]. 3.2 Biomarkers Biomarkers are used in the systematic and precise diagnosis of the patient’s health status, whether it is a disease, pathological condition, or general analysis [35]. Biomarkers are prognostic factors and can be of diverse biological types, for example, transcriptomic, genomic, proteomic, pathological, and imaging. One of the examples is MYCN gene associated with neuroblastoma in children, leading to a chance of earlier death. More than one thousand studies related to prognostic factors are published every year, leading to an enormous accumulation of prognostic factors in diagnosis [36]. Biomarkers can be described by the quality of measurement, as well as invasiveness, accuracy, and price. Hlatky and colleagues [37] suggested that for making incremental value, the best option for the optimal result is using multivariable properties. In many cases, there were incorrect predictions, especially with positive tests indicating the importance of a precise approach in associating a marker with the corresponding condition. The phenomenon in which two randomly selected samples have the same order is explained by a measure defined as c-index (Harrell’s concordance index). Good c-indexes have a value above 0.7 [38]. The P value determines the possibility under the assumption that the previously observed result is the product of chance, and this measure or value enables the observation of the minor changes and if there are any, the null hypothesis is rejected and significant association is reported [39]. From the statistical point of view, the c-index is described as not sensitive enough, while P values are a good tool for predictions of useful biomarkers but by use of related-odds. In designing novel biomarkers, if any doubt is recognized due to their accuracy, the costs for further analysis are not recommended. Therefore, upgrading the information on the existing biomarkers is recommended [40]. The question always arises whether the research of a new biomarker leads to changing the “gold standard” or whether it will only be a type of additional testing with an already existing standard biomarker [37]. Finally, it is important to note that the advances in the development of high-throughput technology have made it possible to reduce the costs of genome sequencing. Different statistical approaches depend on the interpretation of data processing after the experimental results are obtained [41]. In high-throughput data manipulation, legal regulations and safety are required, since it was obtained from the patients. The European Union adopted the law in 2016 related to data safety usage and data cleaning after its manipulation was done and implemented [42].

266

D. Omerki´c and A. Aši´c

4 Pharmacogenomics as an Indispensable Link in the Development of Personalized Medicine The classification of adverse drug reactions (ADRs) can be done into two broad groups, A and B, where A group consists of ADRs caused by the structure of the drug itself, while in group B there are unforeseen ADRs as a result of a genetic factor. Therefore, pharmacogenomics is of great importance in the approach to treatment and drug prescription [43]. In 1990, Gonzalez et al. reported and explained that during the evolutionary process, drug-metabolizing enzymes (DME) genes evolved through the process of defense of animals in their interaction with plants. The effect of these genes was recognized in plants, whereby, in turn, they evolved new metabolites that triggered the production of toxins and, in the meantime, new genes in animals arose. Often, this process is described by the term co-evolution [8]. Hereditary variations and their influence on the metabolism of drugs in humans are identified and reported, since it is important to interpret and analyze the genetic factors as well as to document the patient’s genome to tailor a personalized drug prescription and treatment [43]. Based on this, pharmacology serves for searching therapies that will lead to less frequent ADRs. Polymorphisms are used for testing the affinity scale, or rapidity of drug influence on the human genome. For public health, this is of great importance [8]. A reliable threshold value for investment and a meaningful price for implementing personalized medicine are quite diverse. Therefore, there is no available, precise data to determine an acceptable, cost-effective price for this strategy, because the research in many countries is limited due to the financial possibilities and schedule of investments in the health care system [4]. In 2014, was reported that more than 140 U.S. companies contain labels on pharmacogenetics-related information for use in personalized medicine and are confirmed by the FDA [12]. 4.1 Warfarin Warfarin, also known as Coumadin® , is prescribed for the prevention of anticoagulation [44], used by patients with venous thromboembolism and as a method of prophylaxis [43]. The warfarin dose must be accurately determined since it is the only way to prevent thrombosis or clotting. Otherwise, ADRs may occur, and one of these may be hemorrhage in the event of a person overdosing or being prescribed the wrong therapy. The adverse effect in patients using warfarin in combination with aspirin has been reported. Therefore, it is necessary to combine these types of medications with caution, especially in patients with vascular disease, as the warfarin dose may be reduced due to aspirin [45]. Genetic variant analysis is crucial step prior to drug prescription. Typically, alleles are analyzed at genes VKORC1 and CYP2C9. In testing, specific regions are in focus to better understand and predict feedback. For warfarin testing, VKORC1 promoters are predominantly used [46]. Polymorphisms found on cytochrome P450 (CYP450) protein-coding genes significantly affect drug metabolism. In the group of enzymes with the potential risk alleles, there are also CYP2C9 and CYP2C19, so the adequate dose of the drug is adjusted based on this [47]. In addition to CYP2C9 metabolizing steroid hormones and fatty acids, this enzyme also metabolizes warfarin [43]. When testing CYP2C9, particular attention is paid to alleles *2 (cysteine substitutes for arginine at

An Overview of Personalized Medicine Development

267

position 144) and *3 (leucine substitutes for isoleucine at residue 359) as they are risk alleles for the slow breakdown of warfarin [43, 48]. In this case, the drug effect is reduced compared to the usual dose. While these risk alleles have been found in Caucasians, they have not been reported in Asians [47]. In vitro and in vivo tests have shown that if an individual has risk alleles, the clearance of warfarin is reduced, which results in the risk of increased bleeding and the difficult establishment of warfarin treatment [43]. Cytochrome P450 (CYP) 4F2 (CYP4F2) is associated with catalysis and removal of vitamin K from the reduced form of vitamin K to oxidized hydroxyvitamin K [44, 49]. This enzyme also impacts the stable warfarin activation and dosage, and patients with the risk allele need a higher daily dose of warfarin than patients with the wildtype genotype [50], since this variant reduces the concentration of enzyme involved in vitamin K metabolism [51]. The risk allele is CYP4F2*3 and it requires a higher warfarin dose prescription [44]. In 2007, the FDA introduced information on the pharmacogenomics of warfarin, focusing on the CYP2C9 gene and the possible risk of bleeding in patients of European origin [51]. CYP2C9*2 and CYP2C9*3 alleles are detected in African Americans but in a smaller percentage. However, other CYP2C9 mutant alleles (CYP2C9*5, CYP2C9*6, CYP2C9*8, and CYP2C9*11) are common in African Americans [52], leading to reduced clearance of warfarin [53]. Given this, warfarin prescription for this population requires a reduced drug dose by one-third to one-sixth for the desired pharmacological response, and the use of different prescription algorithms and parameters is imperative when compared to Caucasians. Such an algorithm is as useful as many other algorithms designed by genetic data, but the genetic variation characteristic for only one population plays an important role [44]. Nash [1] reported on a specific study conducted by Stergiopoulos and his colleague Brown, related to routine genotyping before prescribing warfarin whose results did not indicate a reduction in major bleeding. Considering that meta-analysis, more studies are needed to recommend this approach in dosing and prescribing warfarin. The uncertain efficacy of medication, in this case warfarin, is the outcome of a different approach to prescribing therapy to a patient. Some people have high doses at the beginning while some are prescribed lower doses [54]. 4.2 5-Fluorouracil 5-Fluorouracil (5-FU) is in the group of chemotherapy drugs containing active metabolites that get incorporated in RNA and DNA and that disrupt the nucleotide synthetic enzyme thymidylate synthase (TS). This drug has promising results in a prescription for pancreatic, gastric [55], breast, and neck cancer with the best results reported in colorectal cancer therapy [56]. Capecitabine is a prodrug of 5-FU and the pharmacokinetics, as well as pharmacodynamics of this drug, is related to genetic factors [57]. In the drug activation phase, Capecitabine, a cytostatic type of drug, in its active form converts to 5-FU [48].

268

D. Omerki´c and A. Aši´c

Dihydropyrimidine dehydrogenase (DPD), an enzyme that breaks down 5-FU, is encoded by the DPYD gene. DPD defects, caused by genetic variation, lead to irregular enzyme activity resulting in cytotoxicity of a drug, confirming that no dose of 5- FU is approved and recommended, according to FDA [58]. Geno-typing prior to beginning with the treatment is recommended for this enzyme to predict the effect after treatment with 5-FU [59]. Non-functional copies of the DPYD gene are associated with alleles DPYD*2A (c.1905 + 1G > A) or DPYD*13 (c.1679T > G); and a single one-copy presence can lead to 5-FU drug toxicity [58]. DPYD heterozygotes were thus defined as the risk factors. Genotype testing is performed for alleles *2, *13, and *9B to avoid ADR, since non-functional copies reduce DPD activity and increased concentration of nonmetabolized 5-FU [48, 58]. In 2% of Europe Caucasians, the DPYD*2A allele is detected [55]. ADRs caused by 5-FU and Capecitabine are dizziness, angina pain, cardiovascular disease, toxicity [60], life-threatening conditions [58], and even mortality [48]. The presence of two other variants, DPYD c.2846A > T, and DPYD haplotype-B3, is also of clinical consideration, because in patients with DPYD c.2846A > T, a partial dysfunction of the enzyme was observed, while DPYD haplotype-B3 is associated with increased toxicity due to incomplete enzyme function [55]. 4.3 Evidence on Pharmacogenomics Testing and Opioids Opioids are widely used drugs in pain management [61]. PM is one of the solutions to problems such as the emergence of drug addiction [62]. Methadone, buprenorphine, naloxone, and naltrexone are the most prescribed opioids in the USA [61]. Cytochrome P450 2D6 (CYP2D6) metabolizes drugs in a different way depending on the patient’s genetic variations and there are over 130 variant alleles of CYP2D6 [63]. Four phenotype categories are defined based on the genetic makeup of individuals: poor metabolizers (PMs), intermediate metabolizers (IMs), normal (extensive) metabolizers (NMs), and ultrarapid metabolizers (UMs) [64]. CYP2D6 is the first cloned gene in the cluster of metabolic process genes. It is involved in the metabolism of around 25% of medications considered for daily treatments [65]. The study conducted in Italy on opioids and generally CYP2D6 alleles was designed in two groups. Patients who complained about lower-body pain had CYP2D6*2 alleles and were characterized by ultrarapidmetabolism (UM), having an increased risk for side effects [62]. UM is related to an increased risk of side effects, with patients having less benefit from codeine since this phenotype is distinguished for increased formation of morphine and risk for toxicity [62]. CYP2D6*2 is one of the most studied alleles in the Chinese population, as well as CYP2D6*10 [66]. The dyplotype CYP2D6*41/*2N is related to a failure of drug activity. *1/*1xN, *1/*2xN, and *2/*2xNc are CYP2D6 dyplotypes referred to as UM, since their activity score is above 2.25, and activity score is used for genotype-phenotype translation. CYP2D6 alleles with normal function are *1, *2, *35. Poor metabolizers are defined by *9, *17, *29, *41 alleles (Pratt et al. n.d.; CPIC®, n.d.). Ramsey et al. [67] suggest that drugs metabolized by CYP2D6 should be prescribed genotypically due to the widespread prescription in pediatrics (in the USA), referring to patients with poor and ultrarapid metabolisms. Since ultra-rapid metabolizers may lead to the risk of toxic events, e.g., in children with duplicate CYP2D6 gene, resulting in the excessive conversion of the drug into morphine, and slow or poor metabolizers may be

An Overview of Personalized Medicine Development

269

insufficient for therapeutic benefit, being the reason why some pediatric hospitals have banned the use of codeine [68]. However, the Clinical Pharmacogenetics Implementation Consortium (CPIC) believes that there is not enough effective evidence about the mentioned side effects [67]. The FDA thus issued several warnings in 2012 and 2013 with an emphasis on codeine-containing products that caused respiratory depression and death in patients with CYP2D6 polymorphism [68]. The evidence shows that less than 10% of the Caucasian population was identified by ADRs due to opioids because CYP2D6 is poorly effective in these individuals. Due to some studies that were clear indications that codeine could cause respiratory distress, the FDA has made strict and clear recommendations regarding prescription specifically for combination with tramadol. Also, codeine and active CYP2D6 is the only combination that is most permissible for clinical purposes, confirmed by CPIC [65]. Support in decision-making for the prescription of codeine should be followed by the electronic medical record, presenting the next step of reducing the risk of adverse reactions to the drug [68]. Understanding the so-called “opioid crisis” is crucial for a population group that contains genomic risk [61]. Tests for opioid pharmacogenomics are not widely used, but it is necessary to introduce a reduction in test costs to encourage such use in clinical practice [64]. An algorithm for personalized opioid dosing is one of the current solutions [69].

5 Future Perspectives in Personalized Medicine Improvement Unexplored and enigmatic properties are the main characteristics of the human genome, and therefore, every accomplishment opens new questions leading to important scientific results. In the PubMed search engine in March 2023, 133,954 results (citations) are found after searching for “personalized medicine”, and in the last few years, this number is growing very quickly. Currently, tests related to opioid pharmacogenomics are paid for individually without a specific strategy implemented by the health system [64]. Bugada et al. [69] reported that family testing for a specific gene in case of opioid prescription is recommended, referring that this approach applies to other drugs with a test target for different genes. Nowadays, genetic screening is successfully implemented in the diagnosis and monitoring of offspring’s condition, primarily during pregnancy and beyond. Furthermore, it is used in the detection of early stages of the disease or risk detection [10]. It was reported that the younger population in Europe is far more dissatisfied with the medical service. In contrast, the elderly population is unmistakably the possible potential group for an integrated personalized approach in testing and treatment, given that the common occurrences and states of the disease are very well known and reported for this group and that individual differences (of genetic type) can be used for the purpose of personalized medicine [1]. At the moment, any new research in this field is useful for mastering the knowledge about unwanted phenomena caused by the action of drugs, as related to the human genome. It is necessary to work on the implementation of a strategy in PM to bring the treatment and efficiency to a level that corresponds to each patient. Hence, scientists are still asking relevant questions and further cooperation of experts from various fields is needed to improve the practice of prescribing medicines.

270

D. Omerki´c and A. Aši´c

Data sharing is another important segment of the PM strategy [10]. By designing new algorithms, data gathering, and sharing, a complete picture would be reached faster and more efficiently, leading to an advanced solution in the healthcare system. Its importance is reflecting on diagnosis without time-consuming procedures, achieving the great interest of any stakeholder in PM development. Digital health, data science, and precision medicine are inevitable platforms in the definition and realization of successful healthcare treatment in the future.

References 1. Nash, D.B.: Personalized medicine: are we there yet? Am. Health Drug Benefits 7(7), 371–372 (2014) 2. Kichko, K., Marschall, P., Flessa, S.: Personalized medicine in the US and Germany: awareness, acceptance, use and preconditions for the wide implementation into the medical standard. J. Pers. Med. 6(2), 15 (2016) 3. Hess, G.P., Fonseca, E., Scott, R., Fagerness, J.: Pharmacogenomic and pharmacogeneticguided therapy as a tool in precision medicine: current state and factors impacting acceptance by stakeholders. Genet. Res. 97 (2015) 4. Vellekoop, H., et al.: The net benefit of personalized medicine: a systematic literature review and regression analysis. Value Health: J. Int. Soc. Pharmacoeconomics Outcomes Res. 25(8), 1428–1438 (2022). https://doi.org/10.1016/j.jval.2022.01.006 5. Carr, D., Alfirevic, A., Pirmohamed, M.: Pharmacogenomics: current state-of-the-art. Genes 5(2), 430–443 (2014) 6. Chang, K.L., Weitzel, K., Schmidt, S.: Pharmacogenetics: using genetic information to guide drug therapy. Am. Fam. Physician 92(7), 588 (2015) 7. Lee, S.J.: Clinical application of CYP2C19 pharmacogenetics toward more personalized medicine. Front. Genet. 3, 318 (2013) 8. Nebert, D.W.: Polymorphisms in drug-metabolizing enzymes: what is their clinical relevance and why do they exist? Am. J. Hum. Genet. 60(2), 265 (1997) 9. Haga, S.B., Tindall, G., O’Daniel, J.M.: Public perspectives about pharmacogenetic testing and managing ancillary findings. Genet. Test. Mol. Biomarkers 16(3), 193–197 (2012) 10. Ginsburg, G.S., Phillips, K.A.: Precision medicine: from science to value. Health affairs (Project Hope) 37(5), 694–701 (2018). https://doi.org/10.1377/hlthaff.2017.1624 11. OwusuObeng, A., Fei, K., Levy, K., Elsey, A., Pollin, T., Ramirez, A., Horowitz, C.: Physicianreported benefits and barriers to clinical implementation of genomic medicine: a multi-site IGNITE-network survey. J. Pers. Med. 8(3), 24 (2018) 12. Hresko, A., Haga, S.: Insurance coverage policies for personalized medicine. J. Pers. Med. 2(4), 201–216 (2012) 13. Quazi, S.: Artificial intelligence and machine learning in precision and genomic medicine. Med. Oncol. (Northwood, London, England) 39(8), 120 (2022). https://doi.org/10.1007/s12 032-022-01711-1 14. Phillips, K.A., Douglas, M.P., Trosman, J.R., Marshall, D.A.: What goes around comes around”: lessons learned from economic evaluations of personalized medicine applied to digital medicine. Value Health: J. Int. Soc. Pharmacoeconomics Outcomes Res. 20(1), 47–53 (2017). https://doi.org/10.1016/j.jval.2016.08.736 15. Tam, V., Patel, N., Turcotte, M., Bossé, Y., Paré, G., Meyre, D.: Benefits and limitations of genome-wide association studies. Nat. Rev. Genet. 1 (2019)

An Overview of Personalized Medicine Development

271

16. Verma, M.: Genome-wide association studies and epigenome-wide association studies go together in cancer control. Future Oncol. 12(13), 1645–1664 (2016) 17. Carlsten, C., Brauer, M., Brinkman, F., Brook, J., Daley, D., McNagny, K., Pui, M., Royce, D., Takaro, T., Denburg, J.: Genes, the environment and personalized medicine: we need to harness both environmental and genetic data to maximize personal and population health. EMBOreports 15(7), 736–739 (2014). https://doi.org/10.15252/embr.201438480 18. Daly, A.K.: Pharmacogenomics of adverse drug reactions. Genome Med. 5(1), 5 (2013) 19. Visscher, P.M., Brown, M.A., McCarthy, M.I., Yang, J.: Five years of GWAS discovery. Am. J. Hum. Genet. 90(1), 7–24 (2012) 20. Motsinger-Reif, A.A., et al.: Genome-wide association studies in pharmacogenomics: successes and lessons. Pharmacogenet. Genomics 23(8), 383 (2013) 21. Welter, D., et al.: The NHGRI GWAS catalog, a curated resource of SNP-trait associations. Nucleic Acids Res. 42(D1), D1001–D1006 (2013) 22. Levy-Sakin, M., Pastor, S., Mostovoy, Y., Li, L., Leung, A. K., McCaffrey, J., Chung, C.Y., et al.: Genome maps across 26 human populations reveal population-specific patterns of structural variation. Nat. Commun. 10(1), 1025 (2019) 23. Cannon, M.E., Mohlke, K.L.: Deciphering the emerging complexities of molecular mechanisms at GWAS loci. Am. J. Hum. Genet. 103(5), 637–653 (2018) 24. Alekseyev, Y.O., et al.: A next-generation sequencing primer—how does it work and what can it do? Acad. Pathol. 5, 2374289518766521 (2018) 25. Voelkerding, K.V., Dames, S.A., Durtschi, J.D.: Next-generation sequencing: from basic research to diagnostics. Clin. Chem. 55(4), 641–658 (2009) 26. Yin, Y., Butler, C., Zhang, Q.: Challenges in the application of NGS in the clinical laboratory. Hum. Immunol. 82(11), 812–819 (2021). https://doi.org/10.1016/j.humimm.2021.03.011 27. Mardis, E.R., Wilson, R.K.: Cancer genome sequencing: a review. Hum. Mol. Genet. 18(R2), R163–R168 (2009) 28. Mardis, E.R.: Next-generation sequencing platforms. Annu. Rev. Anal. Chem. 6, 287–303 (2013) 29. Butler, J.M.: Advanced topics in forensic DNA typing: methodology. Academic Press (2011) 30. Bruijns, B., Tiggelaar, R., Gardeniers, H.: Massively parallel sequencing techniques for forensics: a review. Electrophoresis 39(21), 2642–2654 (2018) 31. Kozich, J.J., Westcott, S.L., Baxter, N.T., Highlander, S.K., Schloss, P.D.: Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl. Environ. Microbiol. 79(17), 5112–5120 (2013) 32. Weirather, J. L., de Cesare, M., Wang, Y., Piazza, P., Sebastiano, V., Wang, X.J., Buck, D., Au, K.F.: Comprehensive comparison of Pacific biosciences and oxford nanopore technologies and their applications to transcriptome analysis. F1000Research 6, 100 (2017). https://doi. org/10.12688/f1000research.10571.2 33. Bell, G.C., Crews, K.R., Wilkinson, M.R., Haidar, C.E., Hicks, J.K.,Baker, D.K., Freimuth, R.R. et al.: Development and use of active clinical decision support for preemptive pharmacogenomics. J. Am. Med. Inform. Assoc. 21(e1), e93–e99 (2013) 34. Nayarisseri, A., et al.: Artificial intelligence, big data and machine learning approaches in precision medicine and drug discovery. Currentdrugtargets 22(6), 631–655 (2021). https:// doi.org/10.2174/1389450122999210104205732 35. Méndez Hernández, R., Ramasco Rueda, F.: Biomarkers as prognostic predictors and therapeutic guide in critically ill patients: clinical evidence. J. Pers. Med. 13(2), 333 (2023). https:// doi.org/10.3390/jpm13020333

272

D. Omerki´c and A. Aši´c

36. Riley, R.D., Hayden, J.A., Steyerberg, E.W., Moons, K.G., Abrams, K.,Kyzas, P.A., Hemingway, H.: Prognosis research strategy (PROGRESS) 2: prognostic factor research. PLoS Med. 10(2), e1001380 (2013) 37. Hlatky, M.A., Greenland, P., Arnett, D.K., Ballantyne, C.M., Criqui, M.H., Elkind, M.S., Go, A.S., Harrell, F.E., Jr, Hong, Y., Howard, B.V., Howard, V.J., Hsue, P.Y., Kramer, C.M., McConnell, J.P., Normand, S.L., O’Donnell, C.J., Smith, S.C., Jr., Wilson, P.W.: American Heart Association Expert Panel on Subclinical Atherosclerotic Diseases and Emerging Risk Factors and the Stroke Council. Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association. Circulation 119(17), 2408–2416 (2009). https://doi.org/10.1161/CIRCULATIONAHA.109.192278 38. Liu, X., Xiao, C., Yue, K., Chen, M., Zhou, H., Yan, X.: Identification of multiomics biomarkers and construction of the novel prognostic model for hepatocellular carcinoma. Sci. Rep. 12(1), 12084 (2022). https://doi.org/10.1038/s41598-022-16341-w 39. Dahiru, T.: P—value, a true test of statistical significance? A cautionary note. Ann. Ib. Postgrad. Med. 6(1), 21–26 (2008). https://doi.org/10.4314/aipm.v6i1.64038 40. Moons, K.G.: Criteria for scientific evaluation of novel markers: a perspective. Clin. Chem. 56(4), 537–541 (2010) 41. Bragg, L.M., Stone, G., Butler, M.K., Hugenholtz, P., Tyson, G.W.: Shining a light on dark sequencing: characterising errors in Ion Torrent PGM data. PLoS Comput. Biol. 9(4), e1003031 (2013) 42. Ching, T., Himmelstein, D.S., Beaulieu-Jones, B.K., Kalinin, A.A., Do, B.T., Way, G.P., Ferrero, E., Agapow, P.M., Zietz, M., Hoffman, M.M., Xie, W., Rosen, G.L., Lengerich, B.J., Israeli, J., Lanchantin, J., Woloszynek, S., Carpenter, A.E., Shrikumar, A., Xu, J., Cofer, E.M., Greene, C.S. et al.: Opportunities and obstacles for deep learning in biology and medicine. J. R. Soc. Interface 15(141), 20170387 (2018). https://doi.org/10.1098/rsif.2017.0387 43. Pirmohamed, M., Park, B.K.: Genetic susceptibility to adverse drug reactions. Trends Pharmacol. Sci. 22(6), 298–305 (2001). https://doi.org/10.1016/s0165-6147(00)01717-x 44. Johnson, J.A., Caudle, K.E., Gong, L., Whirl-Carrillo, M., Stein, C.M., Scott, S.A., Anderson, J.L.: Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for pharmacogenetics-guided warfarin dosing: 2017 update. Clin. Pharmacol. Ther. 102(3), 397–404 (2017) 45. Lee, C.H., Jung, K.H., Cho, D.J., Jeong, S.K.: Effect of warfarin versus aspirin on blood viscosity in cardioembolic stroke with atrial fibrillation: a prospective clinical trial. BMC Neurol. 19(1), 82 (2019) 46. Allyn-Feuer, A., Ade, A., Luzum, J.A., Higgins, G.A., Athey, B.D.: The pharmacoepigenomics informatics pipeline defines a pathway of novel and known warfarin pharmacogenomics variants. Pharmacogenomics 19(5), 413–434 (2018). https://doi.org/10.2217/pgs2017-0186 47. Westervelt, P., Cho, K., Bright, D.R., Kisor, D.F.: Drug-gene interactions: inherent variability in drug maintenance dose requirements. P & T: Peer-Rev. J. Formul Manag. 39(9), 630–637 (2014) 48. Höppner, W., Primorac, D.: Pharmacogenetics in clinical practice: experience with 16 commonly used drugs. Zagreb: St. Catherine Specialty Hospital (2016) 49. Turner, R.M., Bula, M., Pirmohamed, M.: Personalized medicinein cardiovascular disease. Diagn. Mol. Pathol. 457–471 (2017). https://doi.org/10.1016/b978-0-12-800886-7.00036-4 50. Lee, K.E., et al.: Effects of CYP4F2 gene polymorphisms on warfarin clearance and sensitivity in Korean patients with mechanical cardiac valves. Ther. Drug Monit. 34(3), 275–282 (2012). https://doi.org/10.1097/FTD.0b013e318256a77c 51. Shendre, A., Dillon, C., Limdi, N.A.: Pharmacogenetics of warfarin dosing in patients of African and European ancestry. Pharmacogenomics 19(17), 1357–1371 (2018). https://doi. org/10.2217/pgs-2018-0146

An Overview of Personalized Medicine Development

273

52. Allabi, A.C., et al.: Functional impact of CYP2C95, CYP2C96, CYP2C98, and CYP2C911 in vivo among black Africans. Clin. Pharmacol. Ther. 76(2), 113–118 (2004). https://doi.org/ 10.1016/j.clpt.2004.04.001 53. Lindley, K.J., et al.: Warfarin dosing in patients with CYP2C9*5 variant alleles. Clin. Pharmacol. Ther. 111(4), 950–955 (2022). https://doi.org/10.1002/cpt.2549 54. Takeuchi,F., McGinnis, R., Bourgeois, S., Barnes, C., Eriksson, N., Soranzo, N., Holm, L.: A genome-wide association study confirms VKORC1, CYP2C9, and CYP4F2 as principal genetic determinants of warfarin dose. PLoS Genet. 5(3), e1000433 (2009) 55. Wigle, T.J., Tsvetkova, E.V., Welch, S.A., Kim, R.B.: DPYD and fluorouracil-based chemotherapy: mini review and case report. Pharmaceutics 11(5), 199 (2019). https://doi. org/10.3390/pharmaceutics11050199 56. Longley, D.B., Harkin, D.P., Johnston, P.G.: 5-fluorouracil: mechanisms of action and clinical strategies. Nat. Rev. Cancer 3(5), 330–338 (2003). https://doi.org/10.1038/nrc1074 57. Lam, S.W., Guchelaar, H.J., Boven, E.: The role of pharmaco- genetics in capecitabine efficacy and toxicity. Cancer Treat. Rev. 50, 9–22 (2016). https://doi.org/10.1016/j.ctrv.2016.08.001 58. Dean, L., Kane, M.: Capecitabine Therapy and DPYD Genotype. In: Pratt, V.M. et al. (eds.), Medical Genetics Summaries. National Center for Biotechnology Information (US) (2016) 59. deMan, F.M., Veerman, G.M., Oomen-de Hoop, E., Deenen, M. J., Meulendijks, D., Mandigers, C.M., Mathijssen, R.H.: Comparison of toxicity and effectiveness between fixed-dose and body surface area-based dose capecitabine. Ther. Adv. Med. Oncol. 11, 1758835919838964 (2019) 60. Peng, T., Ouyang, Y., Tong, K.: Rechallenge capecitabine after fluoropyrimidine-induced cardiotoxicity in rectal cancer: a case report. Medicine 98(2) (2019) 61. Balyan, R., Hahn, D., Huang, H., Chidambaran, V.: Pharmacokinetic and pharmacodynamic considerations in developing a response to the opioid epidemic. Expert Opin. Drug Metab. Toxicol. 16(2), 125–141 (2020). https://doi.org/10.1080/17425255.2020.1721458 62. Dagostino, C., et al.: CYP2D6 genotype can help to predict effectiveness and safety during opioid treatment for chronic low back pain: results from a retrospective study in an Italian cohort. Pharmacogenomics Pers. 11, 179–191 (2018). https://doi.org/10.2147/PGPM.S181334 63. Carranza-Leon, D., Dickson, A.L., Gaedigk, A., Stein, C.M., Chung, C.P.: CYP2D6 genotype and reduced codeine analgesic effect in real- world clinical practice. Pharmacogenomics J. 21(4), 484–490 (2021). https://doi.org/10.1038/s41397-021-00226-8 64. Wong, A.K., Somogyi, A.A., Rubio, J., Philip, J.: The role of pharmacogenomics in opioid prescribing. Curr. Treat. Options Oncol. 23(10), 1353–1369 (2022). https://doi.org/10.1007/ s11864-022-01010-x 65. Kaye, A.D., Garcia, A.J., Hall, O.M., Jeha, G.M., Cramer, K.D., Granier, L., Urman, R.D.: Update on the pharmacogenomics of pain management. Pharmacogenomics Pers. Med., 12, 125 (2019) 66. Dai, D.P., et al.: In vitro functional assessment of 22 newly identified CYP2D6 allelic variants in the Chinese population. Basic Clin. Toxicol. 117(1), 39–43 (2015). https://doi.org/10.1111/ bcpt.12363 67. Ramsey, L.B., Ong, H.H., Schildcrout, J.S., Shi, Y., Tang, L.A., Hicks, J.K.,El Rouby, N., Cavallari, L.H., Tuteja, S., Aquilante, C. L., Beitelshees, L., Lemkin, D.L., Blake, K.V., Williams, H., Cimino, J.J., Davis, B.H., Limdi, N.A., Empey, P.E., Horvat, C.M., Kao, D.P.: IGNITE Pharmacogenetics Working Group. Prescribing Prevalence of Medications With Potential Genotype-Guided Dosing in Pediatric Patients. JAMAnetworkopen 3(12), e2029411 (2020). https://doi.org/10.1001/jamanetworkopen.2020.29411

274

D. Omerki´c and A. Aši´c

68. Gammal, R.S., et al.: Pharmacogenetics for safe codeine use in sickle cell disease. Pediatrics 138(1), e20153479 (2016). https://doi.org/10.1542/peds.2015-3479 69. Bugada, D., Lorini, L.F., Fumagalli, R., Allegri, M.: Genetics and opioids: towards more appropriate prescription in cancer pain. Cancers 12(7), 1951 (2020). https://doi.org/10.3390/ cancers12071951

Neural and Rehabilitation Engineering

Association Between Variations in Kinematic Indexes of Manual Dexterity and Mu Rhythm Desynchronization Changes After Action Observation and Motor Imagery Federico Temporiti1,2(B) , Alessandra Calcagno2 , Stefania Coelli2 , Giorgia Marino1 , Roberto Gatti1,3 , Anna Maria Bianchi2 , and Manuela Galli2 1 Physiotherapy Unit, Humanitas Clinical and Research Center—IRCCS, Via Manzoni 56,

Rozzano, Milan, Italy [email protected] 2 Department of Electronic, Information and Bioengineering, Politecnico Di Milano, Via Ponzio 34, Milan, Italy 3 Department of Biomedical Sciences, Humanitas University, Via Rita Levi Montalcini 4, Pieve Emanuele, Milan, Italy

Abstract. The Mirror Neuron System functioning have enabled the development of Action Observation and Motor Imagery (AOMI) as effective approaches to enhance manual dexterity. Improvements in manual dexterity can be quantified using kinematic indexes during the execution of clinical and functional tests, and neurodynamical correlates of such changes may also be investigated through EEG recording. The current study aimed at investigating (1) kinematic and EEG mu rhythm desynchronization changes after an AOMI-training in healthy subjects and (2) the association between variations in kinematic indexes of manual dexterity and mu rhythm desynchronization changes induced by AOMI. Thirty healthy subjects performed a 3-week AOMI intervention consisting of the observation and motor imagery of transitive manual dexterity tasks. Manual dexterity assessed through the analysis of kinematic indexes and EEG mu rhythm desynchronizations were analyzed before and after the training. Results showed that AOMI improved kinematic indexes of manual dexterity and induced an increase in murhythm desynchronizations during motor performance. Furthermore, improvements in kinematic indexes of manual dexterity were related to mu rhythm desynchronization changes in central and parietal cortical areas during the execution of NHPT with the left hand. These findings are explanatory of parameters of human movements subtended to sensorimotor system activity changes induced by AOMI and may provide useful information for planning AOMI interventions aimed at enhancing manual dexterity. Keywords: Action observation · Motor imagery · Manual dexterity

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 277–285, 2024. https://doi.org/10.1007/978-3-031-49068-2_30

278

F. Temporiti et al.

1 Introduction The Mirror Neuron System (MNS) represents a frontoparietal neural network enabling the central nervous system to render visual inputs into a motor representation of observed movements [1]. This system has been reported to show an activity both during the performance and observation of a motor task, showing a mirror mechanism [2]. In addition, the imagination of a motor task in the absence of its execution also elicits brain activations at the level of neural substrates belonging to the MNS [1, 2]. In this context, the MNS ability to generate motor resonance during the observation of actions performed by others and the mental rehearsal of a motor tasks promoted the development of the Action Observation (AO) and Motor Imagery (MI) approaches for enhancing motor learning. During AO, subjects are usually asked to carefully observe video-clips showing motor contents, while during MI subjects are required to imagine the observed actions in the absence of imitation [3]. In the last decades, AO and MI have been applied to enhance the acquisition of motor skills through the development of a motor memory of observed and imagined tasks and, recently, higher benefits have been reported when AO and MI are merged into a single intervention (AOMI) [1, 4]. When considering motor performance improvements after AOMI, positive effects in terms of manual dexterity have been described in healthy subjects and patients with motor impairments using clinical and functional tests [5]. Such benefits were also supported by functional brain changes in MNS areas detected through fMRI or EEG techniques [1, 6, 7] However, the scores of functional and clinical tests usually provide a few information about the quality of motor performance and movement parameters responsible for dexterity changes, which can be instead quantified using kinematic indexes during tests execution [8]. For example, the duration of the sub-phases of a manual dexterity test (Nine Hole Peg Test—NHPT), hand velocity profile during the sub-phases and smoothness of hand trajectories have been reported to provide further information on the characteristic of hand performance [8]. These parameters might differently contribute to manual dexterity improvements observed after an AOMI intervention and some of these indexes might also reveal a relationship with sensorimotor system functional changes (e.g., mu rhythm desynchronization) induced by AOMI during motor performance. Against this background, the association between kinematic parameters and neurophysiological changes subtended to manual dexterity changes after AOMI-training have never been explored. Such investigation may represent an additional step for better understanding the parameters of human movements subtended to sensorimotor system activity changes after an AOMI intervention. Therefore, the study aimed at (1) assessing kinematic and mu rhythm desynchronization changes after an AOMI-training in healthy subjects, (2) investigating the association between kinematic and mu rhythm desynchronization changes after AOMI-training for manual dexterity improvement.

2 Materials and Methods 2.1 Subjects Thirty healthy volunteers between 18 and 30 years old and reported the right hand as dominant were enrolled. Subjects with history of musculoskeletal or neurological disorders, visual or auditory impairments or practicing of sports, jobs or activities requiring

AOMI effects on manual dexterity

279

advanced manual skills were excluded. All participants signed a written informed consent, and the Internal Ethical Committee of our Institution approved the study protocol (n. CLF20/08) [9]. 2.2 Experimental Design Participants were asked to watch 20-min video-clips 4 days a week per 3 weeks. Videoclips showed transitive daily tasks performed with the right hand and requiring advanced manual dexterity. Specifically, video-clips were composed of four tasks (5 min per task) delivered in third (2 min per task) and first-person (3 min per task) perspectives. Actors’ gender was congruent with those of the observers and first-person observation was concurrently associated with motor imagery. Over the 3 weeks of training, the complexity of the tasks progressively increased. In addition, participants were asked to focus on how the tasks were performed and practice no movements during observation and imagination. A daily message on the smartphone and the request to fill in a diary reporting the time of training sessions execution were adopted to ensure the participants’ adherence to the treatment. 2.3 Assessment Procedures All participants were assessed at baseline (T0) and after 3 weeks of training (T1) through kinematic and neurophysiological outcome measures. The kinematic assessment consisted of the detection of kinematic indexes during the Nine Hole Peg Test (NHPT) performed with the right and left hands. During the test, participants grasped nine pegs, inserted them into a nine-hole grid, and replaced the pegs back as quickly as possible. An optoelectronic system (SMART-DX, BTS, Italy) was used to detect hand kinematics (sampling rate at 100 Hz). In particular, three markers were placed on the left distal and left and right proximal corners of the table to define the global reference system (XYZ), while two markers were placed on the left and right middle phalanx of the index finger to estimate the following kinematic indexes using the procedures and algorithm described by Temporiti and co-workers: total and single phases times (peg-grasp, peg-transfer, peg-in-hole, hand-return), normalized jerk, mean and peak velocity values during peg-grasp and hand-return phases. These indexes resulted in valid and reliable parameters for manual dexterity assessment in healthy subjects [8]. EEG signals during NHPT execution were also recorded using a 64-channel continuous system (SD LTM 64 Express, Micromed, Italy) equipped with Ag/AgCl surface electrodes mounted on a cap according to the Standard International 10/20 system (sampling rate at 1024 Hz, impedance lower than 20kOhm and ground electrode between Pz and CPz locations). A 1-min eyes-open resting was recorded prior to NHPT execution. After data collection, data were band-pass filtered in the 1–45 frequency range using a FIR zero-phase filter and, subsequently, down-sampled at 256 Hz. Independent Component Analysis was implemented for artifact removal, and Common Average Rereferencing was applied. The individual mu rhythm spectral profile was extracted, defined as the 2-Hz frequency range centered around the individual alpha frequency of the motor cortex (frequency in the range 7–13 Hz revealing the highest activity during movement

280

F. Temporiti et al.

performance compared to the resting state at the level of C3 and C4 channels) [10, 11]. Mu rhythm desynchronizations during NHPT were computed as power variation (P_var) with respect to the resting condition for each electrode, using the formula: P_var = (P_mov − P_rest)/P_rest where, P_mov is the power in the mu band during NHPT performance, and P_rest represents the mu band power during resting state. Three regions of interests (ROIs) were identified by averaging P_var of the following clusters of electrodes: frontal (Fp1, Fp2, Af7, Af3, Af4, Af8, F7, F5, F3, F1, F2, F4, F6, F8) central (Fc2, Fc4, C2, C4, C6, Cp4, Cp6, Fc1, Fc3, C1, C3, C5, Cp1, Cp3) and parietal (P3, Pz, P4, P5, P1, P2, P6). 2.4 Statistical Analysis Sample size was calculated a-priori based on the NHPT total time. Specifically, assuming a two-tailed alpha error of 5%, a minimum of 30 participants were required to provide 80% power to detect a Cohen’s d = 0.5 (medium effect size) between T0 and T1 [12]. Data were checked for normality using Shapiro-Wilk tests and reported as mean and standard deviation. First, paired t-tests were used to assess significant changes in terms of kinematic parameters and mu rhythm power variations during the NHPT performed with the right and left hands after the 3-week AOMI-training. In the case of significant changes, effect sizes were calculated and reported as Cohen’s d, and deltas () between T0 and T1 (T1-T0) were computed. Negative  values for phases and sub-phases durations and normalized jerk indicate better performance, whereas enhanced performance for velocity parameters was indicated by positive  values. Moreover, negative  values of P_var at the level of frontal, central and parietal ROIs indicate higher cortical activity. Finally, correlations between kinematic indexes and mu rhythm desynchronization changes were assessed using the Pearson’s correlation coefficient. The strength of correlation was interpreted as small (r < 0.3), moderate (0.3 < r < 0.6) and strong (r > 0.6) [13]. The level of statistical significance was set at α = 0.05.

3 Results All participants (mean age 25.5 SD 2.6 years, mean, mean weight 64.6 SD 11.2 kg, mean height 173.4 SD 10.3 cm, 15 women and 15 men) completed the study and performed the evaluation sessions correctly. During NHPT with the right hand, participants showed a significant decrease in total (p < 0.001, d = 1.01), removing (p < 0.001, d = 1.02), peg-transfer (p = 0.017, d = 0.51) and peg-in-hole (p = 0.012, d = 0.55) times. Moreover, participants revealed a significant decrease in terms of mu rhythm P_var in the frontal ROI from T0 to T1 (p = 0.028, d = 0.29) (Table 1). No significant correlations were found between kinematic and mu rhythm P_var changes during NHPT with the right hand (Table 3). During NHPT with the left hand, participants showed a significant decrease for total (p < 0.001, d = 0.81), removing (p = 0.001, d = 0.75), peg-transfer (p = 0.017, d = 0.51), peg-in-hole (p = 0.048, d = 0.41) times and a significant increase for mean

AOMI effects on manual dexterity

281

Table 1. Changes in kinematic parameters and mu rhythm power variations during NHPT performed with the right hand after a 3-week AOMI-training. T0

T1

p-value

Total time (s)

21.8 ± 3.7

19.2 ± 2.3

65 mmHg (T5). The resuscitation protocol was based on current clinical guidelines [6]. ABP was continuously recorded during the experiment in the ascending aorta (5F pressure catheter, Transonic System Europe, The Netherlands) with a sampling frequency of 500 Hz (Notocord Hem, Notocord, France). Fifteen consecutive beats were selected for one pig at all time points for the analyses, The ABP was resampled at 125 Hz and each beat’s onset was identified by standard algorithms [7]. IP was first detected by using the 3rd order derivative method proposed in [5] on a single beat. Specifically, it is implemented into 3 steps: 1. First, the systolic peak is identified by looking for the global maximum within the ABP beat signal. 2. Second, the first order derivative of the signal is computed, and its maximum is detected; this point is typically called dP/dt max. 3. Finally, the third order derivative of the signal is computed, and the IP can be detected as the point where the third order derivative crosses zero line passing from positive to negative values after dP/dtmax and before systolic point occurrences.

340

M. Carrara et al.

The method has been tested both on a single beat and on a beat template obtained by averaging 15 consecutive beats. 2.2 Bandpass Filtering Algorithm for Inflection Point Detection A new method is proposed based on a bandpass filter-based approach, with the underlying idea to enhance the high frequencies related to signal variations without amplifying the eventual high frequency noise superimposed to the registration. A zero-phase elliptic infinite impulse response filter has been implemented with the following characteristics: (i) bandpass frequencies set equal to 5 and 15 Hz; (ii) stopband attenuation of 60 dB; (iii) passband ripples of 0.1 dB; (iv) transition width equal to 0.785 Hz corresponding to a steepness of the filter of 0.85. The order of the filter is minimized based on the signal length, and it typically assumes values around 19. The block diagram of the proposed algorithm is represented in Fig. 2. The ABP waveform is filtered with the passband filter previously described and two strategies: are adopted: (a) The IP occurrence is detected as the local maximum of the filtered signal during systolic upstroke, representing the point where the slow variation due to the inflection is filtered away by the high pass filter with cut-off frequency at 5 Hz. (b) The IP occurrence is detected as at the intersection point between the original ABP pulse and its filtered version. The method has been tested both on a single beat and on a beat template consisting of the average of 15 consecutive beats. 2.3 ECHOPAC Validation Cardiac echography images (ECHOPAC® software) were performed at each time point of the experiment, as part of the experimental protocol. Among several echo measures available, we were interested in the left ventricular outflow track (LVOT) image, which represents the portion of the left ventricle through which blood passes to enter the aorta through the aortic valve. The spectral Doppler technology allowed to record the blood flow velocity over time in this anatomical district, and this can yield indirect information on the cardiac cycle and the timing of blood flow ejection: during the systolic phase the blood flow velocity will increase due to the ejection of blood from the left ventricle to the aorta; at the time of aortic valve closure the flow will return to zero. From these images we could retrieve information about the time of occurrence of the peak of blood flow with respect to the onset of ejection, which should correspond to the time instant of IP on ABP pulse. Indeed, the peak of blood flow typically occurs at the time instant of IP, as shown in Fig. 1. Therefore, the time of the peak of the LV outflow (Tr) was obtained from the echo images and compared with the occurrence time of IP obtained from ABP waveform by the methods previously described. Figure 3 shows an example of an echography image and the extraction of the features of interest. In particular, the onset of the beat, i.e., the point where the left ventricular flow starts to increase, is represented by the sky-blue line; the peak of flow is marked by the pink line, and its occurrence time relative to the onset (i.e., Tr) is taken as an independent measure of IP on the arterial blood pressure waveform.

A New Algorithm for Arterial Inflection Point Identification

341

Fig. 2. Bandpass-filtering-based algorithm flow diagram for inflection point detection.

Fig. 3. Example of an echography image stored in ECHOPAC® and used for the analysis.

3 Results Figure 4 reports the performance of the algorithms on a single beat under two conditions: (i) at baseline with some high-frequency acquisition noise superimposed to ABP (left panel); (ii) during septic shock condition, when the overall cardiovascular system is

342

M. Carrara et al.

highly altered, and this is reflected by an acute change in ABP wave morphology (right panel). We can notice that the algorithm based on the 3rd order derivative (pink square) doesn’t identify the IP close to the instant retrieved from echo images (green cross), in both conditions, mainly due to the superimposed noise and to the smoother ABP waveform, respectively. On the contrary, the proposed algorithm (yellow and red squares) is able to give more reliable results.

Fig. 4. Performance of the different algorithms for inflection point (IP) identification on a single beat with superimposed noise during baseline (left panel), and during septic shock condition (right panel).

Fig. 5. Performance of the different algorithms for inflection point (IP) identification on beat template obtained by averaging 15 consecutive beats. The green cross represents the IP occurrence derived from echo images. The red and the yellow squares show the IP identified by the bandpassfiltering based technique proposed using the intersection point method and the local maximum method, respectively. The pink square shows the result of the 3rd order derivative approach (5). T1: baseline; T2: septic shock; T3: after fluids resuscitation; T4: after administration of drugs for heart rate control; T5: after noradrenaline administration.

Figure 5 shows the performance of the algorithms over the beat template, obtained by averaging 15 consecutive beats, at each timepoint of the experiment. This averaging

A New Algorithm for Arterial Inflection Point Identification

343

procedure is thought to eliminate some high frequency noise which can disturb the algorithms, as previously shown. Table 1 reports the error values computed as absolute difference between the output of each algorithm and the echo IP values, both in terms of time and pressure values. Table 1. Absolute error relative to the echo inflection point position of each algorithm.

Time error (ms)

Pressure error (mmHg)

T1

T2

T3

T4

T5

Average error

3rd derivative

16

16

24

16

32

20.8

BP local max

32

8

8

8

32

17.6

BP intersection

8

24

16

56

0

20.8

3rd derivative

1.2

5.3

7.6

3.6

9.3

5.4

BP local max

2.6

2.2

2.5

2.0

9.3

3.7

BP intersection

0.6

5.6

3.4

10.9

0

4.1

T1: baseline; T2: septic shock; T3: after fluids resuscitation; T4: after administration of drugs for heart rate control; T5: after noradrenaline administration; BP: bandpass

The proposed method consisting in a bandpass filter and a search for local maximum resulted superior to the 3rd order derivative method. Although at baseline all the approaches have similar results, the 3rd order derivative method tends to underestimate and to anticipate the IP from time point T2 (septic shock condition) until the end of experiment, where the ABP waveform becomes very smooth, and the inflection almost disappears on a visual inspection (Fig. 5); it can be noticed indeed that its performance worsens during the experiment, i.e. it shows an almost increasing bias from T2 to T5.

4 Discussion and Conclusion In this work a new method for IP identification has been explored and compared to the traditional 3rd order derivative method; the reference value of IP position for the validation of the methods was extracted from echo images of left ventricular outflow tract. The main interest was to test these methods on ABP waveform during acute critical illness, such as septic shock, where the typical morphological characteristics of the physiologic arterial beat can undergo important alterations. The results showed that the bandpass filtering method proposed may represent a good approach for IP identification in cases of very smooth waveform induced by the septic shock condition. Table 1 shows that the best performance was variably assigned to the local maximum search or the intersection point identification depending on the different waveform at the time point. However, the local maximum approach appeared as more promising, accounting for the best performance in the majority of cases (T2, T3, and T4). Finally, this work highlighted how variable can be the output of this kind of algorithms, depending on the morphological features of ABP and some additional factors,

344

M. Carrara et al.

such as the possible noise superimposed to the signal. For these reasons, we may suggest using a template beat to enhance the actual waveform features and avoiding noise or other erroneous information. 4.1 Limitations A major limitation of this study is that the echo images were performed over a window of 1 h and this prevented to have a perfect synchronization between the arterial waveforms analyzed and the echo images. Moreover, the two anatomical districts of acquisition are not the same, i.e. the portion of the left ventricle close to aortic valve where the echocardiography was acquired and the ascending aorta where the ABP was measured by the catheter; this may have generated small delays. 4.2 Conclusion This work represents a preliminary exploration of alternative methods for inflection point identification to be used in critical cardiovascular conditions such as septic shock, where the arterial waveform can appear very different from the typical morphology described under physiological conditions. The availability of a robust algorithm for reliable morphological features extraction from the arterial waveform of critically ill patients will allow to derive important insights on cardiovascular biomechanical properties, pressure wave dynamics and pulse wave propagation, which are all highly altered by the acute disease and the resuscitation maneuvers. Main weaknesses of the standard algorithms for inflection point detection typically used in chronic patients, such as hypertensive patients, have been pointed out when they are applied to data derived from critical care settings. The simultaneous measure of blood pressure and flow at the same point is the gold standard in order to validate these algorithms. Further studies are necessary to refine the IP detection algorithm so to translate into clinical practice markers and indices related to pulse wave propagation which are currently neglected.

References 1. Nichols, W.W., O’Rourke, M.F., Vlachopoulos, C.: McDonald’s Blood Flow in Arteries Theoretical, Experimental and Clinical Principles, 6th edn., 768 pp. CRC Press (2011) 2. Torjesen, A., Wang, N., Larson, M.G., Hamburg, N.M., Vita, J.A., Levy, D., et al.: Forward and backward wave morphology and central pressure augmentation in men and women in the Framingham Heart Study. Hypertension 64(2), 259–265 (2015). https://doi.org/10.1161/HYP ERTENSIONAHA.114.03371 3. Li, Y., Gu, H., Fok, H., Alastruey, J., Chowienczyk, P.: Forward and backward pressure waveform morphology in hypertension. Hypertension 69(2), 375–381 (2017). https://doi.org/10. 1161/HYPERTENSIONAHA.116.08089 4. Kaya, M., Balasubramanian, V., Li, J.K.: Augmentation index in the assessment of wave reflections and systolic loading. Comput. Biol. Med. 113(Oct), 103418 (2019). https://doi.org/10. 1016/j.compbiomed.2019.103418

A New Algorithm for Arterial Inflection Point Identification

345

5. Karamanoglu, M.: A system for analysis of arterial blood pressure waveforms in humans. Comput. Biomed. Res. 30(3), 244–255 (1997). https://doi.org/10.1006/cbmr.1997.1450 6. Evans, L., Rhodes, A., Alhazzani, W., Antonelli, M., Coopersmith, C.M., French, C., et al.: Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. 47(11), 1181–1247 (2021). https://doi.org/10.1007/s00134-021-065 06-y 7. Zong, W., Heldt, T., Moody, G.B., Mark, R.G.: An open-source algorithm to detect onset of arterial blood pressure pulses. Comput. Cardiol. 2003, 259–262 (2003). https://doi.org/10. 1109/CIC.2003.1291140

Hemodynamic Cardiovascular Indices to Predict the Response to Angiotensin-II in Septic Shock Marta Carrara1(B)

, Bruno Garcia2 , Antoine Herpain2,3

, and Manuela Ferrario1

1 Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milan,

Italy [email protected] 2 Experimental Laboratory of Intensive Care, Université Libre de Bruxelles, Brussels, Belgium 3 Intensive Care Department, St-Pierre University Hospital, Brussels, Belgium Abstract. Septic shock is known to severely impair cardiovascular function, leading to altered pulse wave transmission and blood pressure (BP) control. Current therapies for septic shock have still poor outcomes. Recently, the effectiveness of angiotensin-II (ang-II) in catecholamine-resistant vasodilatory shock has been established. However, excessive dosage of ang-II may have pro-inflammatory and pro-oxidative effects, worsening the cardiovascular condition of septic shock patients. Currently there are no clinical guidelines to optimally guide ang-II administration, and the therapy targets consist mainly in static indices, like mean arterial pressure, which often mask different cardiovascular responses. The main objective of this study is to characterize the cardiovascular response to ang-II in a polymicrobial septic shock animal population, in terms of arterial vascular properties, blood pressure propagation, and blood pressure regulation. Eight pigs underwent a protocol of polymicrobial septic shock and resuscitation with fluids and ang-II, and they were subdivided into low-angII (n = 5) and high-angII (n = 3) according to the total ang-II dosage. Continuous cardiac output (CO), BP in aorta and femoral artery were recorded. The 2-element Windkessel model was used to estimate arterial compliance (AC), total peripheral resistance (TPR), and arterial time constant τ; pulse pressure (PP) amplification was computed as the ratio between femoral and aortic PP; autonomic activity was studied by cardiac baroreflex sensitivity (BRS). The results highlighted a more compromised cardiovascular condition in high-angII pigs, as resulted from the reduced AC, TPR, τ, BRS and PP amplification compared to low-angII. If this behavior was related to shock severity has still to be investigated. Keywords: Therapy responsiveness · Cardiovascular system · Windkessel model

1 Introduction Sepsis has been called a “hidden” healthcare disaster [1] and it has been recognized as a global health priority by the World Health Organization (WHO) in 2017 [2], accounting for almost 20% of all global deaths [3]. Septic shock is a subset of sepsis in which particularly profound circulatory, cellular, and metabolic abnormalities are associated with a greater risk of mortality than with sepsis alone; hospital mortality rates of septic shock patients are still greater than 40% [4]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 346–355, 2024. https://doi.org/10.1007/978-3-031-49068-2_35

Hemodynamic Cardiovascular Indices to Predict the Response

347

The typical phenotype of septic shock is severe vasodilation and capillary leakage triggered by systemic inflammation, which can lead to hypotension and hypovolemia and, finally, to tissues hypoperfusion and cellular death. Currently, the resuscitation protocols aim at preventing the tissue hypoperfusion and the multi organ failure by maneuvers of hemodynamic stabilization, which include the administration of fluids and vasopressors until specific targets are reached, such as mean arterial pressure > 65 mmHg, central venous pressure > 8 mmHg, urine output > 0.5 mL kg h, and mixed venous oxygen saturation > 70%. These thresholds are thought to guarantee a sufficient tissue oxygenation [5]. The first line vasopressor recommended by the international guidelines is noradrenaline (norepinephrine), which is a physiological neurotransmitter of the catecholamine family, with a strong adrenergic agonist effect. Thus, it mediates an increase in sympathetic outflow, leading to vasoconstriction and increased heart rate. However, some patients may develop a catecholamine-resistant vasodilatory shock, and this condition is typically associated to a poor prognosis, with 30-day all-cause mortality exceeding 50% [6]. In the recent years different strategies have been investigated as alternative to catecholamine administration, angiotensin-II was shown to be beneficial in patients with vasodilatory septic shock not responding to high doses of conventional vasopressors, increasing mean arterial pressure and allowing reductions in the total amount of administered catecholamines [7]. Angiotensin-II is a naturally occurring peptide produced by the renin-angiotensin system (RAS) with strong vasopressor activity. RAS is typically activated during sepsis and provides an important physiologic mechanism to counteract hypotension, together with the autonomic sympathetic system. As angiotensin-II binds to its receptor AT1, it mediates several biological processes, including increased vascular permeability, oxidative stress and proinflammatory mediators expression, and promoting endothelial dysfunction [8, 9]. However, the dosage of angiotensin-II plays a key role, and in a recent work [10], a low dosage administration was found to be beneficial; indeed, patients receiving ≤5 ng/kg/min angiotensin-II at 30 min from start of treatment were more likely to safely recover from shock. In particular, these patients had higher survival rate at 28 days, and a higher proportion of these patients showed a mean arterial pressure response after 3 h of angiotensin-II administration. Currently there are no clinical guidelines in order to guide angiotensin-II administration and to decide the optimal dosage; moreover, the therapy targets consist mainly in static indices, like mean arterial pressure, which often mask different cardiovascular responses. The main objective of this study is to characterize the cardiovascular response to angiotensin-II in a polymicrobial septic shock animal population, in terms of arterial vascular properties, blood pressure propagation along the arterial tree, and blood pressure regulation. In particular, if there is a dose-dependent effect, the dynamical changes of cardiovascular and autonomic system following angiotensin-II administration, could be used to predict the responsiveness to the treatment in clinical settings.

348

M. Carrara et al.

2 Methods 2.1 Experimental Protocol The experiments were performed in the Experimental Intensive Care Laboratory (LA1230336) at the Université Libre de Bruxelles with the approval of the local ethics committee (Comité Ethique du Bien-être Animal – CEBEA ULB), and it was conducted in accordance with the EU Directive 2010/63/EU for animal experiments and the ARRIVE guidelines for animal research. Eight pigs were instrumented and successively allowed to rest for ~2 h, after which the first baseline measurements and blood samples were taken (baseline, T1). Sepsis was induced by intraperitoneal instillation of autologous feces, filtered and diluted in 300 mL of 10% glucose. During septic shock onset, fluid maintenance was limited until the animal reached a severe condition of hypotension (mean arterial pressure (MAP) goal of 45–50 mmHg for one hour). The end of this period was used as reference for septic shock condition (T2). Immediately after, a full fluid resuscitation was initiated by administration of a crystalloid and colloid perfusions in order to maintain pulse pressure variation (PPV) ≤ 12% throughout the entire experiment. After two hours of hemodynamic stabilization, defined by a stable MAP and in the absence of further increases in cardiac output (CO), the animal was considered stable; therefore, blood samples were taken at this time point (T3). Finally, a continuous infusion of angiotensin II (ang-II) was administered for 8 h in order to maintain MAP between 65 and 75 mmHg. Two time points were considered during this period: T4 after the first 3 h of ang-II infusion, i.e. the short term response to the drug, and T5 at the end of the experiment after 8 h of ang-II administration. Animals were then euthanized with a potassium chloride injection and an overdose of thiopental. Arterial blood pressure (ABP) was continuously recorded in the ascending aorta (5F pressure catheter, Transonic System Europe, The Netherlands) and in the common femoral artery (SPR-350S Mikro-TipR, Millar, United States); continuous CO (CCO) was acquired through pulmonary artery catheter (Edwards LifeSciences, California, USA). All acquired signals were converted using an A/D converter (Notocord Hem, Notocord, France) with high temporal resolution (250 or 500 Hz). For each time point, stationary segments (average duration of 10 min) of aortic and femoral pressure signals were extracted from the Notocord software together with CCO and imported into MATLAB® framework for the analyses. 2.2 Signals Processing Beat-to-beat time series of systolic (SAP), diastolic (DAP), and mean (MAP) arterial pressure were obtained from both aortic and femoral ABP waveforms using standard algorithms [11, 12]. The time series of pulse pressure (PP) was computed as the difference between SAP and DAP within the same beat. The time series of heart period (HP) was obtained by computing the time difference between consecutive onsets of ABP, and it is considered as a surrogate for the RR-intervals time series; heart rate (HR) was derived as 60/HP (bpm). Beat-to-beat mean CO was computed by averaging CCO values over each cardiac cycle; beat-to-beat stroke volume (SV) time series was obtained as CO/HR

Hemodynamic Cardiovascular Indices to Predict the Response

349

(mL). All the beat-to-beat indices were then averaged and considered for successive analyses. 2.3 Arterial Vascular Properties The two-element Windkessel model (Fig. 1) was adopted to study the arterial vascular system in terms of arterial compliance (AC), total peripheral resistance (TPR) and characteristic arterial time constant τ [13].

Fig. 1. The two-element Windkessel model of arterial tree. CO, cardiac output; ABP, arterial blood pressure; TPR, total peripheral resistance; AC, arterial compliance.

According to the model, TPR can be estimated as TPR =

Pao,mean Pao,mean − Pven,mean ≈ CO CO

(1)

where Pao,mean and Pven,mean are the mean value of aortic and venous ABP, respectively. The compliance is defined as the ratio of a volume change, V, and the resulting pressure change P; if ABP is measured in aorta the AC can be estimated as AC =

SV V = P PP

(2)

The time constant τ of the arterial tree, representing the time constant of the exponential diastolic decay, is defined by the following equation: τ = AC ∗ TPR

(3)

This relationship is true on a beat-to-beat basis if we consider the ABP waveform measured centrally where the cumulative effects of wave reflections are attenuated [14]; in this case, ABP should decay like a pure exponential during each diastolic interval with a time constant τ. However, this relationship is also consistent if we consider a time scale sufficiently long such that the wavelengths of the propagating waves are much larger than the dimension of the arterial tree. At such time scales, the arterial tree acts as a single blood reservoir, and the Windkessel model is therefore valid. So, for example, if pulsatile activity abruptly ceased, then peripheral ABP may eventually decay like a pure exponential as soon as the faster wave reflections died out [15]. Based on this concept, we computed the time constant τ on long time intervals (6-min windows) of the measured ABP waveforms by adopting the method proposed by Mukkamala et al. [15].

350

M. Carrara et al.

Finally, the PP amplification (PPamp) phenomenon was studied by computing the ratio between femoral and aortic PP (PPfem and PPao, respectively): PPamp =

PPfem PPao

(4)

In physiological conditions PP increases from central to peripheral sites due to waves reflection. Indeed, when blood pressure is measured in the periphery close to the reflection sites, the backward wave has to travel back a shorter distance before summing up to the forward wave and the lower compliance of peripheral arteries contributes to increase PP value as well. 2.4 Arterial Blood Pressure Regulation Cardiac baroreflex sensitivity (BRS) was estimated in order to investigate the mechanism of short-term blood pressure control, which is mediated by the ANS. In particular, cardiac BRS evaluates the changes in RR-intervals or HP induced by SAP oscillations and mediated by the ANS. BRS was estimated via the bivariate model method on aortic ABP [16]. The parameters of interest are the feedback gain (FB), which quantifies the autonomic-mediated cardiac baroreflex, and the feedforward gain (FF) or runoff effect, which explains the oscillations in blood pressure generated by oscillations in HR, due to the mechanical coupling between the two systems. Both gains were assessed in the lowfrequency range ([0.04–0.15] Hz). Granger causality from SAP to HP and vice versa was verified before computation of the gains. The order of the model was optimized based on the Akaike information criterion, ranging from 5 to 15.

3 Results The pigs were divided into two groups (Fig. 2) based on total amount of ang-II they received: high-angII group (3 pigs) with total ang-II dose at the end of resuscitation > 100 μg/kg and low-angII group (5 pigs) with total cumulative ang-II dosage 0.75 were identified as stable, whereas features with ICC < 0.5 after translation were

Assessing Left Ventricle Radiomic Features Robustness

359

considered discriminative. Only first order and texture features were evaluated using this method. After obtaining a set of stable and discriminant features, the selection of nonredundant features was performed by mean of a correlation-based step. The Spearman correlation coefficient ρ was computed for each pair of features. In case |ρ| > 0.95 only the feature with the lower mean Spearman coefficient with all the others (n − 2) features was selected, ‘n’ being the total number of features.

Fig. 1. Example of perturbations applied to the ROI. The filled structure represents the original ROI, while the dashed red line represents the modified ROI. (a) ROI zoom-in with τ = 0.5. (b) ROI zoom-out with τ = −0.5. (c) ROI translation of 30% in the x positive direction. (d) ROI contour randomization with superpixels area equal to 1.44 mm2 .

Finally, the selection of more relevant features was completed by performing the Wilcoxon signed rank test. Features showing significant difference (p < 0.05) between CA and HCM group were included in the further analysis.

360

F. Lo Iacono et al.

2.4 Machine Learning Classification The dataset was divided into training and test set (70% and 30% of the data, respectively). The features were z-scored, and feature selection was performed on the training set. Then, the most relevant features were used to train a logistic regression model aimed to classify CA versus HCM. The machine learning classification model was implemented in Scikit-Learn Python library. The entire workflow was repeated 15 times using different stratified train-test set splits. For each of them accuracy, sensitivity and specificity were computed and averaged when presenting the results.

3 Results Robustness analysis step identified as stable 89 features after erosion (18 first-order statistics and 71 texture features), 74 features after dilation (18 first-order statistics and 56 texture features) and 89 features after contour randomization (18 first-order statistics and 71 texture features). 84 features (18 first-order statistics and 66 texture features) were indeed identified as discriminant performing translation. Such features, divided by perturbation method, are shown in the Euler-Venn diagrams in Fig. 2.

Fig. 2. Euler-Venn diagrams representing features identified as stable and/or discriminative in each perturbation method. (a) Features identified as stable after erosion, dilation and contour randomization. (b) Features selected as robust, i.e., stable, and discriminative.

Thus, according to the pre-defined criteria, 65 radiomic features were identified as robust: 18 first-order statistics and 47 texture features. To these, 14 shape and size features – not submitted to robust assessment – were added. ICC values, computed for each feature, are reported in Fig. 3 distinguishing the four perturbation techniques. At each train-test split 40 ± 3 features were selected within the correlation step, and 14 ± 1 features with the p-value based method. Classification results showed a mean classification accuracy of 0.81 ± 0.1, a mean sensitivity of 0.75 ± 0.1 and a mean specificity of 0.84 ± 0.1.

4 Discussion Radiomics features are vulnerable to variability caused by several factor, such as differences in image segmentation due to inconsistencies in delineating the ROI among different observers and within the same observer. Thus, within radiomic workflow, robustness analysis is a crucial preliminary step to determine the extent to which such variations can be tolerated for each specific application, i.e., without affecting predictive performance.

Assessing Left Ventricle Radiomic Features Robustness

361

Fig. 3. ICCs as a function of the perturbation technique. In correspondence of erosion, dilation and random contour, black circles represent ICCs higher than the stability threshold value 0.75, and red stars represents the features under-threshold. In correspondence of translation, black circles represent ICCs lower than the threshold value 0.5, identifying discriminative features, while red stars represent over-threshold features.

In this work erosion and dilation were coupled with ROI contour randomization and translation. Erosion, dilation, and contour randomization allow to evaluate radiomic features stability against human error and variability. Translation, indeed, provides a selection based on features discrimination capacity. Findings have shown that most of the features show higher stability to ROI erosion and contour randomization with respect to dilation. Also, results obtained from erosion and dilation suggest that a narrower segmentation might perform better than a wider one. To the best of our knowledge, this is the first study adopting this approach to evaluate CCT radiomic features extracted from left ventricle. The main limitation of the study is represented by the limited sample size which has to be enlarged in order to validate the results achieved. However, this approach could provide a valuable alternative to accurate assess radiomic features robustness.

References 1. Aerts, H.J.W.L., et al.: Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat. Commun. 5, 4006 (2014). https://doi.org/10.1038/ncomms 5006 2. Gevaert, O., et al.: Glioblastoma multiforme: exploratory radiogenomic analysis by using quantitative image features. Radiology 273, 168–174 (2014). https://doi.org/10.1148/radiol. 14131731 3. Balagurunathan, Y., et al.: Reproducibility and prognosis of quantitative features extracted from CT images. Transl. Oncol. 7, 72–87 (2014). https://doi.org/10.1593/tlo.13844

362

F. Lo Iacono et al.

4. Le, E.P.V., et al.: Assessing robustness of carotid artery CT angiography radiomics in the identification of culprit lesions in cerebrovascular events. Sci. Rep. 11, 3499 (2021). https:// doi.org/10.1038/s41598-021-82760-w 5. Hu, W., et al.: Novel radiomics features from CCTA images for the functional evaluation of significant ischaemic lesions based on the coronary fractional flow reserve score. Int. J. Cardiovasc. Imaging 36, 2039–2050 (2020). https://doi.org/10.1007/s10554-020-01896-4 6. Oikonomou, E.K., et al.: A novel machine learning-derived radiotranscriptomic signature of perivascular fat improves cardiac risk prediction using coronary CT angiography. Eur. Heart J. 40, 3529–3543 (2019). https://doi.org/10.1093/eurheartj/ehz592 7. Shang, J., et al.: Prediction of acute coronary syndrome within 3 years using radiomics signature of pericoronary adipose tissue based on coronary computed tomography angiography. Eur. Radiol. 32, 1256–1266 (2022). https://doi.org/10.1007/s00330-021-08109-z 8. Zhou, X.Y., et al.: Diagnosis of cardiac amyloidosis using a radiomics approach applied to late gadolinium-enhanced cardiac magnetic resonance images: a retrospective, multicohort, diagnostic study. Front. Cardiovasc. Med. 9, 818957 (2022). https://doi.org/10.3389/fcvm. 2022.818957 9. Hinzpeter, R., Wagner, M.W., Wurnig, M.C., Seifert, B., Manka, R., Alkadhi, H.: Texture analysis of acute myocardial infarction with CT: first experience study. PLoS ONE 12, e0186876 (2017). https://doi.org/10.1371/journal.pone.0186876 10. Raisi-Estabragh, Z., et al.: Repeatability of cardiac magnetic resonance radiomics: a multicentre multi-vendor test-retest study. Front. Cardiovasc. Med. 7 (2020) 11. Reproducibility of segmentation-based myocardial radiomic features with cardiac MRI. https://pubs.rsna.org/doi/epdf/10.1148/ryct.2020190216. Last accessed 23 Feb 2023 12. Yang, M., et al.: Development and validation of a machine learning-based radiomics model on cardiac computed tomography of epicardial adipose tissue in predicting characteristics and recurrence of atrial fibrillation. Front. Cardiovasc. Med. 9 (2022) 13. van Griethuysen, J.J.M., et al.: Computational radiomics system to decode the radiographic phenotype. Cancer Res. 77, e104–e107 (2017). https://doi.org/10.1158/0008-5472.CAN-170339 14. Zwanenburg, A., et al.: The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping. Radiology 295, 328–338 (2020). https://doi.org/10.1148/radiol.2020191145 15. Zwanenburg, A., et al.: Assessing robustness of radiomic features by image perturbation. Sci. Rep. 9, 614 (2019). https://doi.org/10.1038/s41598-018-36938-4 16. Bologna, M., et al.: Assessment of stability and discrimination capacity of radiomic features on apparent diffusion coefficient images. J. Digit. Imaging 31, 879–894 (2018). https://doi. org/10.1007/s10278-018-0092-9 17. Koo, T.K., Li, M.Y.: A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J. Chiropr. Med. 15, 155–163 (2016). https://doi.org/10.1016/j.jcm. 2016.02.012

Assessing the Performance of MRI-Radiomic Prognostic Signatures in Head and Neck Cancer Patients: A Comparative Analysis Anna Corti1(B)

, Luca Mainardi1

, and Valentina D. A. Corino1,2

1 Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milan,

Italy [email protected] 2 Cardiotech Lab, Centro Cardiologico Monzino IRCCS, Milan, Italy

Abstract. The number of radiomic studies has dramatically increased in the last decade. However, the reproducibility of radiomic analyses remains challenging, although being fundamental for their clinical transferability. Herein we proposed an analysis of published radiomic signatures based on magnetic resonance imaging (MRI) for prognosis of overall survival (OS) in head and neck cancer (HNC) patients. Specifically, 5 reproducible MRI-based radiomic signatures were identified and their performance was tested on an external dataset of n = 137 HNC patients. Although the lack of complete methodological details was encountered in the analyzed radiomic studies, the analysis performed herein allowed identifying 3 over 5 significantly prognostic signatures for OS of HNC patients. In particular, the radiomic signature with the highest stratification capability between high/low risk groups, provided a C-index 0.60, HR 2.43 and log-rank p = 0.0022. Overall, the study demonstrated the feasibility of comparing published radiomic signatures on an external dataset (provided that sufficient methodological details are reported). In future, major efforts should be put in reporting radiomic analyses in order to enable their full reproduction in view of their potential translation in clinics. Keywords: Radiomics · Overall survival · Head and neck cancer

1 Introduction Head and neck squamous cell carcinoma (HNC) are a group of highly heterogeneous malignancies representing the seventh most common and the sixth most deadly tumor worldwide (accounting for more than 350,000 annual deaths) [1]. Nowadays, the tumornode-metastasis (TNM) staging system is the main factor guiding risk assessment, treatment choice and prognosis [2]. However, the stratification performance of staging-based system is quite low. The high heterogeneity of HNC and the emergence of personalized medicine fostered the development of additional biomarkers. Radiomics, namely the quantitative extraction of high throughput data from medical images, demonstrated to hold the potential for providing HNC biomarkers [3]. The number of radiomic studies © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 363–368, 2024. https://doi.org/10.1007/978-3-031-49068-2_37

364

A. Corti et al.

has dramatically increased in the last decade. In this context, the repeatability and reproducibility of radiomic analyses remains an open issue, which needs to be addressed in order to enable the clinical translation of radiomics [4]. The aim of the present study was to test the prognostic performance of reproducible radiomic signatures for prognosis of overall survival (OS) based on magnetic resonance imaging (MRI) on a HNC patient dataset.

2 Methods 2.1 Patient Dataset, MRI Image Preprocessing and Features Extraction A subset of locally advanced HNC (clinical TNM III/IV according to the 8th edition of AJCC/UICC) of the BD2Decide project (NCT02832102) presenting with pre-contrast T1-weighted (T1w) and T2-weighted (T2w) and post-contrast T1w (T1wCont) MR image sequence (acquired with 1.5 T scanner) was used in the present study (n = 137 patients) [5]. The region of interest (ROI) corresponding to the primary tumor volume was segmented by an expert radiologist. MRI images underwent the following preprocessing steps: (i) denoising, through a 3D Gaussian filter with a 3 × 3 × 3 voxel kernel and σ = 0.5; (ii) intensity non-uniformities correction, through the N4ITK algorithm [6]; (iii) intensity standardization, through Z-score; (iv) voxel size resampling, through B-spline interpolation. Radiomic features were extracted from the original image and transformed images, including the Laplacian of Gaussian (σ = 0.5, 1.0, 1.5, 2.0 and 5.0 mm) the wavelet, the square, the square root and the logarithm filters. For each original and transformed image, features belonging to first order statistics, shape and size (only for original images), grey level cooccurrence matrix, grey level size zone matrix, neighboring gray tone difference matrix and grey level dependence matrix were extracted, for a total of n = 5064 features. Pyradiomics 2.2.0 software was used to extract the features [7]. 2.2 Radiomic Signature Testing A literature survey was performed to retrieve reproducible MRI-radiomic signatures for prognosis of OS in HNC patients and compute the radiomic scores on our HNC dataset. To reproduce the signatures, the methods applied in the studies were followed. However, if information regarding features normalization and/or signature dichotomization was missing, the following criteria were applied. As regards features normalization, (i) if the information was missing, the original features were considered, (ii) if features normalization was mentioned, but without providing additional details (e.g., method and features median/standard deviation), Z-score normalization was applied on our data. Similarly, to dichotomize the radiomic signatures (between high/low risk groups), if the threshold was not provided, the median value of the radiomic signature in the present data was used. The identified reproducible radiomic signatures were thus computed on our dataset. The prognostic performance of the radiomic signatures was compared in terms of Harrel’s concordance index (C-index), hazard ratio (HR) and high/low risk group patient stratification with Kaplan-Meier curves.

Assessing the Performance of MRI-Radiomic Prognostic Signatures

365

3 Results 3.1 Survey of Literature Five reproducible MRI-based radiomic studies [8–12], providing information on the selected features along with regression coefficients, were identified and reported in Table 1. Four monomodal and one multimodal radiomic signatures were reported. Most of the radiomic signatures were based on T1wCont sequence (Rad-1, Rad-2, and Rad4), one radiomic signature was based on T2w sequence (Rad-5) and one multimodal radiomic signature was based on the T1w, T1wCont and T2w sequences (Rad-3). The number of the selected features span from 2 to 10. Moreover, the selected features were all different among the studies, leading to a total of 25 potentially prognostic features identified (i.e., 21 from T1wCont, 3 from T2w and 1 from T1w). Specifically, (i) among the 21 T1wCont features, 13 were extracted from the wavelet transformation (textural features), 5 from the Laplacian of Gaussian transformation (3 first order and 2 textural features) and 3 from the original image (2 shape and 1 textural features); (ii) among the 3 T2w features, one was extracted from the wavelet transformation (textural feature), one from the Laplacian of Gaussian transformation (first order feature) and one from the original image (shape feature) and (iii) the T1w feature was extracted from the original image (textural feature). As regards features normalization, (i) to compute Rad-1, Z-score standardization was applied on the features, as reported in the study [8], (ii) to compute Rad-2 and Rad-4, the original features were considered because standardization processes were not mentioned in the corresponding studies [9, 11], (iii) to compute Rad-3, Z-score standardization was applied on the features, by considering the mean and standard deviation reported in the study [10], and (iv) to compute Rad-5, Z-score standardization was applied on the features, because feature normalization was mentioned in the study, but details on the normalization method were not provided [12]. Finally, as regards the cutoff value to dichotomize the signature, except for Rad-4 for which the cutoff value was available, the median value was considered. See Table 1. Table 1. MRI-based radiomic signatures included in the study. Signature

Study

Features

Feature normalization

Cutoff

Rad-1

Bos 2021 [8]

10 from T1wCont

Z-score – no details

NA

Rad-2

Chen 2022 [9]

6 from T1wCont

NA

NA

Rad-3

Alfieri 2022 [10]

3 from T1w, T1wCont, T2w

Z-score – details

NA

Rad-4

Siow 2022 [11]

4 from T1wCont

NA

0.5

Rad-5

Mossinelli 2023 [12]

2 from T2w

Standardized – no details

NA

366

A. Corti et al.

3.2 Radiomic Features Testing Table 2 lists the results of the radiomic signatures and the combined radiomic score and Fig. 1 shows the corresponding Kaplan-Maier curves. Three over five radiomic signatures (i.e., Rad-2, Rad-3 and Rad-5) were significantly prognostic for the OS of the considered dataset, with Rad-5 providing the best performance, with C-index 0.60 (IQR 0.57–0.63), HR 2.43 (95% confidence interval 1.41–4.20) and log-rank p = 0.0022. Table 2. Performance of radiomic signatures. Signature

C-index

Log-rank HR

Log-rank p

Rad-1

0.56 [IQR 0.53 0.59]

1.40 [95% CI 0.82 2.41]

0.28

Rad-2

0.55 [IQR 0.52 0.57]

1.78 [95% CI 1.04 3.06]

0.05

Rad-3

0.60 [IQR 0.58 0.62]

2.06 [95% CI 1.20 3.55]

0.013

Rad-4

0.48 [IQR 0.45 0.51]

0.92 [95% CI 0.24 3.61]

0.82

Rad-5

0.60 [IQR 0.57 0.63]

2.43 [95% CI 1.41 4.20]

0.0022

C-index: Harrel’s concordance index; HR: hazard ratio

Fig. 1. Kaplan-Meier curves for (a) Rad-1, (b) Rad-2, (c) Rad-3, (d) Rad-4 and (e) Rad-5.

4 Discussion The number of radiomic studies has dramatically increased in the last decade. However, the lack of common consensus in the applied methodologies and the paucity of details reported in the studies, make the reproducibility of published results a challenge. This

Assessing the Performance of MRI-Radiomic Prognostic Signatures

367

in turn hampers the potential translation of radiomics to clinics [4]. The present study proposed an analysis of published MRI-based radiomic signatures for prognosis of OS in HNC patients, with the goal of testing their replicability and performance on an external dataset of n = 137 HNC patients. Overall, a faithful reproduction of the signatures was not possible because of the lack of complete methodological details (e.g., either about feature normalization or cutoff value for dichotomization). This sheds light on the need for defining a common consensus about the transparency of the delivered information, which is required to correctly replicate the radiomic analyses and subsequently to lay the foundations for a potential clinical translation. However, to overcome the lack of details and compute the radiomic signatures, some assumptions were introduced. With the exception of Rad-4, all the radiomic signatures separated the high - from the low-risk groups, with 3 of them presenting with significant p-value of the log-rank test. The present study is not exempt from limitations, mainly associated to the necessary assumptions that were considered in order to compute the radiomic signatures. In future, performing a meta-analysis of HNC radiomic prognostic signatures is thought to provide a remarkable impact in the field.

5 Conclusion This study demonstrated the feasibility of comparing published radiomic signatures on an external dataset, provided that sufficient methodological details are described. Future efforts should be put in reporting radiomic analyses in order to enable their full reproduction in view of their potential translation in clinics.

References 1. Bray, F., Ferlay, J., Soerjomataram, I., Siegel, R.L., Torre, L.A., Jemal, A.: Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 68, 394–424 (2018). https://doi.org/10.3322/caac.21492 2. Machiels, J.-P., René Leemans, C., Golusinski, W., Grau, C., Licitra, L., Gregoire, V.: Squamous cell carcinoma of the oral cavity, larynx, oropharynx and hypopharynx: EHNS-ESMOESTRO Clinical Practice Guidelines for diagnosis, treatment and follow-up (2020). https:// doi.org/10.1016/j.annonc.2020.07.011 3. Bruixola, G., et al.: Radiomics and radiogenomics in head and neck squamous cell carcinoma: potential contribution to patient management and challenges. Cancer Treat. Rev. 99, 102263 (2021). https://doi.org/10.1016/j.ctrv.2021.102263 4. Pfaehler, E., et al.: A systematic review and quality of reporting checklist for repeatability and reproducibility of radiomic features. Phys. imaging Radiat. Oncol. 20, 69–75 (2021). https:// doi.org/10.1016/j.phro.2021.10.007 5. Cavalieri, S., et al.: Development of a multiomics database for personalized prognostic forecasting in head and neck cancer: The Big Data to Decide EU Project. Head Neck 1–12 (2020). https://doi.org/10.1002/hed.26515 6. Tustison, N.J., Cook, P.A., Gee, J.C.: N4ITK: improved N3 bias correction. IEEE Trans. Med. Imaging 29, 1310–1320 (2010). https://doi.org/10.1109/TMI.2010.2046908.N4ITK 7. van Griethuysen, J.J.M., et al.: Computational radiomics system to decode the radiographic phenotype. Cancer Res. 77, e104–e107 (2017)

368

A. Corti et al.

8. Bos, P., et al.: Improved outcome prediction of oropharyngeal cancer by combining clinical and MRI features in machine learning models. Eur. J. Radiol. 139, 109701 (2021). https:// doi.org/10.1016/j.ejrad.2021.109701 9. Chen, J., et al.: An MRI-based radiomics-clinical nomogram for the overall survival prediction in patients with hypopharyngeal squamous cell carcinoma: a multi-cohort study. Eur. Radiol. 32, 1548–1557 (2022). https://doi.org/10.1007/s00330-021-08292-z 10. Alfieri, S., et al.: Prognostic role of pre-treatment magnetic resonance imaging (MRI)-based radiomic analysis in effectively cured head and neck squamous cell carcinoma (HNSCC) patients. Acta Oncol. 60, 1192–1200 (2021). https://doi.org/10.1080/0284186X.2021.192 4401 11. Siow, T.Y., et al.: MRI radiomics for predicting survival in patients with locally advanced hypopharyngeal cancer treated with concurrent chemoradiotherapy. Cancers (Basel) 14 (2022). https://doi.org/10.3390/cancers14246119 12. Mossinelli, C., et al.: The role of radiomics in tongue cancer: a new tool for prognosis prediction. Head Neck (2023). https://doi.org/10.1002/hed.27299

In Vitro Electrochemotherapy Experiments to Quantify the Number of Cisplatin Molecules Needed for a Cytotoxic Effect When Different Types of Pulses Are Delivered Maria Scuderi1(B)

1 , Janez Šˇ ˇ , Janja Dermol-Cerne canˇcar2 , Stefan Markovi´c2 1 Lea Rems , and Damijan Miklavˇciˇc1

,

1 Faculty of Electrical Engineering, University of Ljubljana, Tržaška 25, 1000 Ljubljana,

Slovenia [email protected] 2 Department of Environmental Sciences, Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia

Abstract. Electrochemotherapy treats various types of tumors by combining chemotherapeutic drugs, such as bleomycin or cisplatin, with the delivery of high-voltage pulses resulting in cell membrane electroporation. Conventionally 8 square monopolar pulses of 100 µs duration at a repetition frequency of 1 or 5000 Hz are delivered. Nevertheless, other types of high voltage pulses are being explored for electrochemotherapy and other electroporation-based applications, including a few µs long (high-frequency) bipolar pulses (shown to reduce muscle contractions and pain), and several ms long monopolar pulses (used for gene transfection). It remains unclear whether different types of pulses have any significant effect on drug uptake and cytotoxicity in the context of electrochemotherapy. Thus, the aim of this work was to experimentally quantify the number of cisplatin molecules needed to achieve a cytotoxic effect when different types of pulses are used. We tested three different types of pulses: (1) bursts of 2 µs bipolar pulses, (2) 100 µs monopolar pulses, and (3) 5 ms long monopolar pulses. Our results of in vitro electrochemotherapy experiments show that different number of cisplatin molecules is needed to achieve a comparable cytotoxic effect for the tested pulse types. Keywords: Electrochemotherapy · Electroporation · Cisplatin uptake · Cell survival · Inductively coupled plasma mass spectrometry

1 Introduction Electrochemotherapy is a local treatment used in clinics to treat cutaneous and subcutaneous tumors, with ongoing clinical trials for the treatment of deep-seated tumors [1]. The standard operating procedures for electrochemotherapy require first the administration of a chemotherapeutic drug (cisplatin or bleomycin) intravenously or intratumorally, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 369–375, 2024. https://doi.org/10.1007/978-3-031-49068-2_38

370

M. Scuderi et al.

and then the application of high-voltage pulses to the tumor [2]. During pulse application, an electric field is established in the tissue that, if sufficiently high, renders cell membranes transiently more permeable (i.e., results in membrane electroporation) and consequently enhances the drug uptake into cells. Both cisplatin and bleomycin kill proliferating (cancerous) cells by acting on DNA, however, both drugs poorly permeate the cell membrane under physiological conditions. By increasing the drug uptake with electroporation, the cytotoxic effect of the drug can be potentiated several folds, minimizing the amount of the needed drug and its side effects. In electrochemotherapy, conventionally 8 square monopolar pulses of 100 µs duration at a repetition frequency of 1 or 5000 Hz are used [2]. Drawbacks of the use of 100 µs long pulses are pain and muscle contractions, as well as possible interference with the heart rhythm. Thus, there is a need to use local or general anesthesia and muscle relaxants and synchronize the pulses with the heart rhythm. To overcome these drawbacks, recent studies suggest the use of bursts of short (around 1–10 µs) high-frequency bipolar pulses, which minimize pain and muscle contractions [3]. Such pulses are already used for the ablation of tumors and cardiac tissue by irreversible electroporation, and they can potentially be used in electrochemotherapy [4, 5]. Electrochemotherapy is also used in combination with gene therapy to induce immune stimulation [6]. When using electroporation for gene therapy, several ms long monopolar pulses are commonly used to deliver DNA inside the cells [1, 7]. Currently, the effects that different types of pulses have on cisplatin uptake and cytotoxicity in electrochemotherapy are unclear and not well understood. Thus, in this paper, we investigate the effects that different types of pulses have on cell survival by quantifying the number of cisplatin molecules needed to achieve a comparable cytotoxic effect.

2 Materials and Methods 2.1 Cell Preparation We used the Chinese Hamster Ovary cell line (CHO-K1; cat. no. 85051005). After trypsinizing the cells, they were centrifuged and resuspended in Dulbecco’s Modified Eagle’s Medium (DMEM, cat. no. D5671) at desired cell concentration (~4 × 106 cells/mL). 2.2 Electroporation Setup We used three different types of pulses: (1) High-frequency bipolar pulses (50 bursts at repetition frequency 1 Hz). Each burst contained 50 short bipolar pulses of 2 µs positive and 2 µs negative pulse. The interpulse delay was 2 µs, Fig. 1A. The high-frequency bipolar pulses of different voltages (80–320 V with a step of 40 V) were delivered by L-POR V0.1 (mPOR, Slovenia). (2) 100 µs monopolar pulses (8 pulses at repetition frequency 1 Hz) of different voltages (80–320 V with a step of 40 V) were delivered by a laboratory prototype pulse generator (University of Ljubljana) based on H-bridge digital amplifier with 1 kV MOSFETs (DE275-102N06A, IXYS, USA), Fig. 1B [8]. (3)

In Vitro Electrochemotherapy Experiments to Quantify the Number

371

5 ms long monopolar pulses (8 pulses at repetition frequency 1 Hz) of different voltages (80–160 V with a step of 20 V) were delivered by BTX Gemini X2 pulse generator (Harvard Apparatus, USA), Fig. 1B. For brevity, we refer to the three types of pulses used as 50 × 50 HF pulses, 8 × 100 µs pulses, and 8 × 5 ms pulses. The electric pulses were applied to 2 mm aluminium cuvettes. The voltage and the current were monitored in all experiments with an oscilloscope Wavesurfer 422, 200 MHz, a differential voltage probe ADP305, and a current probe CP030 (from LeCroy, USA), to ensure the quality of the delivered pulses, according to the recommendations [9].

Fig. 1. (A) Short high-frequency bipolar pulses. From left to right: 50 bursts were applied with a repetition frequency of 1 Hz; one burst was 400 µs long and consisted of 50 bipolar pulses; one bipolar pulse of amplitude U consisted of a 2 µs long positive pulse, and a 2 µs long negative pulse (both of voltage U) with a 2 µs long interpulse delay. (B) One monopolar pulse of amplitude U and pulse duration of 100 µs or 5 ms was applied with a repetition frequency of 1 Hz.

2.3 Determination of the Optimal Electric Field Strength Before performing in vitro electrochemotherapy experiments, we determined the optimal electric field for each type of pulse used, by first determining the permeability and survival curves without any chemotherapeutic drugs. To determine the permeability curve, the cells were suspended in DMEM with 1 µM YO-PRO-1, and then pulses were delivered. Three minutes after pulse delivery the samples were diluted 10× in DMEM. The uptake of YO-PRO-1 was measured on the flow cytometer (Attune NxT; USA). Cells were excited with a blue laser at 488 nm, and the emitted fluorescence was detected through a 530/30 nm band-pass filter. The percentage of permeabilized cells was determined from the histogram of YO-PRO-1 fluorescence. To determine the survival curve, the cell suspension was electroporated, and 25 min after pulse delivery, the samples were diluted in HAM-F12 and 2 × 104 cells were transferred in each well of a 96-well plate in triplicates. After 24 h of incubation in a humidified atmosphere at 37 °C and 5% CO2 , the MTS assay was performed. The MTS assay was used to quantify the number of viable cells by evaluating their metabolic activity by measuring the formazan absorbance at 490 nm on a microplate reader (Tecan Infinite 200; Austria). Cell survival was determined by first subtracting the background from all measurements and then normalizing the absorbance of the treated samples to the absorbance of the control sample. After determining the permeability and survival curves, the optimal electrical field was chosen where the highest cell membrane permeability and highest cell survival were achieved.

372

M. Scuderi et al.

2.4 Cytotoxicity and Intracellular Concentration of Cisplatin We performed in vitro electrochemotherapy experiments by exposing cells to a fixed optimal electric field (see Subsect. 2.3) in the presence of different concentrations of cisplatin. First, the cell suspension with added cisplatin (10, 30, 50 µM) was transferred to 2 mm aluminium cuvettes and then pulses were delivered. Control samples received no pulses (only cisplatin). The samples were then processed either for determining cytotoxicity or intracellular concentration of cisplatin. For the cytotoxicity experiments: 25 min after pulse delivery each sample was diluted in HAM-F12 and 5.5 × 103 cells were transferred in each well of a 96-well plate in triplicates. After 72 h the MTS assay was performed, as described in Subsect. 2.3. For the intracellular concentration experiments: 25 min after pulse delivery, each sample was diluted in HAM-F12, centrifuged, and washed twice. The cell pellet was separated from the supernatant and digested, and the platinum content was measured with inductively coupled plasma mass spectrometry (7900 ICP-MS, USA) with 193 Ir (Merck, Germany) used as an internal standard during the measurement. 2.5 Statistical Analysis Statistical analysis was performed using Excel, SigmaPlot 11.0 (Systat Software, USA), and Prism 9.4.1 (GraphPad Software, USA). For survival and permeability experiments significant differences from the control were determined by Kruskal-Wallis One Way ANOVA on Ranks test. For cytotoxicity and intracellular concentration experiments, significant differences from the control were determined by two-way ANOVA test.

3 Results and Discussion We aimed to investigate the effects of different types of pulses on cell survival by quantifying the number of cisplatin molecules needed to achieve a comparable cytotoxic effect. 3.1 The Optimal Electric Field Strength We determined the permeability and survival curves in the absence of cisplatin to choose the optimal electric field strength for each of the three tested types of pulses. Permeability curves show how the percentage of permeabilized cells increases with increasing electric field strength, whereas the survival curves show how the percentage of viable cells decreases with increasing electric field strength. The obtained permeability (dashed) and survival (solid) curves as a function of the applied electric field are shown in Fig. 2. The chosen optimal electric fields (i.e. highest permeability and highest survival) are 1.4 kV/cm for 50 × 50 HF pulses, 1.2 kV/cm for 8 × 100 µs pulses, and 0.6 kV/cm for 8 × 5 ms pulses. Consistent with Pucihar et al. [10], longer pulses (8 × 5 ms) require lower electric fields to obtain a similar fraction of electroporated cells than shorter pulses (8 × 100 µs). Moreover, 50 × 50 HF pulses require a higher electric field to achieve a similar fraction of permeabilized cells than classical 8 × 100 µs pulses, which agrees with our previous studies [5, 8].

In Vitro Electrochemotherapy Experiments to Quantify the Number

373

Fig. 2. Cell survival (solid) and cell membrane permeability (dashed) as a function of the electric field when (A) 50 × 50 HF pulses; (B) 8 × 100 µs pulses; (C) 8 × 5 ms pulses are used. The chosen optimal electric fields are circled in red. Each data point presents the mean ± standard deviation from 3–4 experiments. * - statistically significant differences from control (p < 0.05).

3.2 Cytotoxicity and Intracellular Concentration of Cisplatin We next determined the cytotoxicity and intracellular concentration of cisplatin when exposing cells to the three tested types of pulses of the optimal electric field strength. Figure 3A shows how cell survival decreases as the extracellular concentration of cisplatin increases. In the absence of applied pulses, the tested range of cisplatin concentrations (10–50 µM) does not affect cell viability (black curve). However, cytotoxicity is strongly potentiated with all three types of pulses, decreasing the cell survival to ~40% at the highest cisplatin concentration (50 µM). Results for all three types of pulses are similar and are not statistically significantly different. We also measured the amount of intracellular Platinium (Pt) and quantified the number of cisplatin molecules per cell assuming that 1 mol of Pt is equivalent to 1 mol of cisplatin, Fig. 3B. When no electric pulses are applied (black line), the number of intracellular cisplatin molecules increases slightly with increasing cisplatin concentration due to passive (i.e. diffusion) and active (i.e. membrane transporters, endocytosis, pinocytosis, macrocytosis) transport of cisplatin [11, 12]. However, when electric pulses are applied, the number of intracellular cisplatin molecules increases significantly. The greatest increase is observed for 8 × 5 ms pulses (up to 6.7 × 107 cisplatin molecules at 50 µM). Roughly 2× lower increase is observed for both 50 × 50 HF pulses and 8 × 100 µs pulses. There is a statistically significant difference between 8 × 5 ms pulses and the other two types of tested pulses when extracellular cisplatin concentration is 50 µM. Combining the results of Fig. 3A, B we can determine the number of cisplatin molecules needed to have a cytotoxic effect when the three tested types of pulses are applied, Fig. 3C. We can observe that more than two times higher number of cisplatin molecules is needed with 8 × 5 ms pulses to achieve the same cytotoxic effect as when using 50 × 50 HF pulses or 8 × 100 µs pulses. It is also possible to notice a trend in which decreasing the pulse duration decreases also the number of molecules needed to achieve a cytotoxic effect. This trend has also been shown by Vizintin et al. [13], where nanosecond pulses required a lower number of cisplatin molecules to achieve a cytotoxic effect compared with 8 × 100 µs pulses. However, when comparing the effects of different pulse duration and pulse shapes, it is important to be aware that different cell death pathways might be triggered [14].

374

M. Scuderi et al.

Fig. 3. Cytotoxicity of cisplatin (A) and cisplatin molecules per cell (B) at different concentrations of cisplatin at a fixed electric field: 1.4 kV/cm for 50 × 50 HF pulses, 1.2 kV/cm for 8 × 100 µs pulses and 0.6 kV/cm for 8 × 5 ms pulses. (C) Cell survival as a function of cisplatin molecules per cell in combination with electroporation. Each data point presents the mean ± standard deviation from 3–4 experiments. *- statistically significant differences from control (p < 0.05). The color of the asterisk corresponds to the line color for a specific pulse type.

In conclusion, our results show that the number of cisplatin molecules needed to achieve a cytotoxic effect depends on the type of the delivered pulses. However further experimental and modeling work is needed to confirm and explain the obtained findings.

References 1. Geboers, B., et al.: High-voltage electrical pulses in oncology: irreversible electroporation, electrochemotherapy, gene electrotransfer, electrofusion, and electroimmunotherapy. Radiology 295(2), 254–272 (2020) 2. Gehl, J., et al.: Updated standard operating procedures for electrochemotherapy of cutaneous tumours and skin metastases. Acta Oncol. 57(7), 874–882 (2018) 3. Cvetkoska, A., Maˇcek-Lebar, A., Trdina, P., Miklavˇciˇc, D., Reberšek, M.: Muscle contractions and pain sensation accompanying high-frequency electroporation pulses. Sci. Rep. 12(1), 1–15 (2022) 4. Miklavˇciˇc, D., et al.: The effect of high frequency electric pulses on muscle contractions and antitumor efficiency in vivo for a potential use in clinical electrochemotherapy. Bioelectrochemistry 65(2), 121–128 (2005) 5. Scuderi, M., Rebersek, M., Miklavcic, D., Dermol-Cerne, J.: The use of high-frequency short bipolar pulses in cisplatin electrochemotherapy in vitro. Radiol. Oncol. 53(2), 194–205 (2019) 6. Cemazar, M., et al.: Efficacy and safety of electrochemotherapy combined with peritumoral IL-12 gene electrotransfer of canine mast cell tumours. Vet. Comp. Oncol. 15(2), 641–654 (2017) 7. Sachdev, S., Potoˇcnik, T., Rems, L., Miklavˇciˇc, D.: Revisiting the role of pulsed electric fields in overcoming the barriers to in vivo gene electrotransfer. Bioelectrochemistry 144, 107994 (2022) 8. Sweeney, D.C., Reberšek, M., Dermol, J., Rems, L., Miklavˇciˇc, D., Davalos, R.V.: Quantification of cell membrane permeability induced by monopolar and high-frequency bipolar bursts of electrical pulses. Biochim. Biophys. Acta BBA-Biomembr. 1858(11), 2689–2698 (2016) 9. Cemazar, M., Sersa, G., Frey, W., Miklavcic, D., Teissié, J.: Recommendations and requirements for reporting on applications of electric pulse delivery for electroporation of biological samples. Bioelectrochemistry 122, 69–76 (2018)

In Vitro Electrochemotherapy Experiments to Quantify the Number

375

10. Pucihar, G., Krmelj, J., Reberšek, M., Napotnik, T.B., Miklavˇciˇc, D.: Equivalent pulse parameters for electroporation. IEEE Trans. Biomed. Eng. 58(11), 3279–3288 (2011) 11. Shen, D.-W., Pouliot, L.M., Hall, M.D., Gottesman, M.M.: Cisplatin resistance: a cellular self-defense mechanism resulting from multiple epigenetic and genetic changes. Pharmacol. Rev. 64(3), 706–721 (2012) 12. Spreckelmeyer, S., Orvig, C., Casini, A.: Cellular transport mechanisms of cytotoxic metallodrugs: an overview beyond cisplatin. Molecules 19(10), 15584–15610 (2014) 13. Vizintin, A., Markovic, S., Scancar, J., Kladnik, J., Turel, I., Miklavcic, D.: Nanosecond electric pulses are equally effective in electrochemotherapy with cisplatin as microsecond pulses. Radiol. Oncol. 56(3), 326–335 (2022) 14. Napotnik, T.B., Polajžer, T., Miklavˇciˇc, D.: Cell death due to electroporation—a review. Bioelectrochemistry 141, 107871 (2021)

Spectrogram-Driven Convolutional Neural Network for Real-Time Non-invasive Hyperglycaemia Detection in Paediatric Type-1 Diabetes via Wearable Sensors Owain Cisuelo1(B) , Muhammad Salman Haleem1 , John Hattersley2 , and Leandro Pecchia1,3 1 School of Engineering, University of Warwick, Coventry CV4 7AL, UK

[email protected]

2 Human Metabolism Research Unit, University Hospitals Coventry and Warwickshire,

Coventry CV2 2DX, UK 3 Università Campus Bio-Medico di Roma, 00128 Roma, Italy

Abstract. Real-time detection of glycaemic events is crucial in the effective management of type 1 diabetes, particularly in paediatric patients. Recent advances in wearable sensors and machine learning have allowed for the inference of glycaemic events based on non-invasive physiological signals such as electrocardiogram (ECG). However, existing approaches have limitations due to the limited number of ECG features analysed and their applicability to real-life conditions. To overcome these limitations, we propose a spectrogram-driven deep learning methodology for real-time glycaemic event detection. We calculated beat-level spectrograms using Short Time Fourier Transform (STFT) on ECG beats extracted from continuous signals using our deep learning ECG segmentation tool. Subjectspecific multi-layer 2D convolutional neural networks were trained on the spectrograms. We evaluated our methodology on an original dataset comprising continuous ECG and interstitial glucose data collected from children with type-1 diabetes over several days in real-life conditions. Our approach achieved an average personalised hyperglycaemia detection accuracy of 96.9%. Keywords: Hyperglycaemia detection · Diabetes · Deep learning · Wearable sensors · Electrocardiogram

1 Introduction Type-1 diabetes (T1D) is a chronic disorder caused by autoimmune damage to the insulinproducing cells in the pancreas, leading to elevated blood glucose (BG) concentration [1]. The prevalence of T1D globally is 5.9 per 10,000 with an incidence of 15 per 100,000 people [2]. There is no cure for T1D, therefore, the development of tools that enable effective management can be crucial in reducing the risk of adverse events as well as delaying the onset of long-term complications [3]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 376–386, 2024. https://doi.org/10.1007/978-3-031-49068-2_39

Spectrogram-Driven Convolutional Neural Network for Real-Time

377

With the advent of modern technology, the use of wearable sensors is now becoming pivotal in effective diabetes management [4–6]. This is due to their capability of realtime physiological data acquisition which can be inferred to changes in blood glucose non-invasively due to modern data-driven methods [7]. There have been several attempts in the design and development of data-driven methods for detecting hyperglycaemic or hypoglycaemic events via ECG signals [8–10]. The main challenge associated with these methods is the limited feature representation of the ECG signal which results in the robustness and trustworthiness of these methods being questionable. Secondly, most of the existing methods are based on Heart Rate Variability (HRV) which limits the real-time detection capability. Thirdly, most of the ECG features are represented in the time domain, limiting the detection power of glycaemic status as sympathetic and parasympathetic autonomic activity are reflected in the frequency components of HRV [11]. Finally, most of the state-of-the-art methods are based on traditional machine learning approaches. Although several attempts have been made towards exploiting the ECG changes associated with changes in BG, leveraging the power of modern deep learning algorithms to infer glycaemic status from ECG [12–15], the studies were limited to a non-diabetic population. In order to address these shortcomings, we designed and developed a spectrogramdriven deep learning model for real-time glycaemic event detection for effective diabetes management. We transformed a one-dimensional time-series signal to two-dimensional time-frequency space via the short-time Fourier transform (STFT). The STFT of a signal represents how the frequency content of a nonstationary signal changes over time. The time-frequency approach can provide a suitable basis for the discovery of complex, high-dimensional properties of physiological signals [16]. Previously, STFT has been investigated for the detection of cardiovascular disease [17] and arrhythmia [18], and glycaemic event detection is a novel application that remains unexplored. A commonly used deep learning technique is the convolutional neural network (CNN) which excels in the classification of n-dimensional data due to its ability to transform the input data to extract local spatial features. CNNs process an input through layers of convolution operations (filters) which detect spatial patterns; and pooling; where the most prominent features from the region covered by the filter propagate onto the next layer [19]. We developed spectrogram-driven CNNs which convert individual beats extracted from continuous ECG to time-frequency representation which were trained for real-time glycaemic event detection. The aim of this paper is to represent a proof-of-concept CNN model for the classification of glycaemic status of ECG beats during low activity extracted from a continuous ECG obtained from wearable sensors and is structured as follows; The methodology, including a description of our original dataset, data processing steps, model architecture, and a method to interpret the model by highlighting the key features learned by the model is presented in Sect. 2. Section 3 details the results of personalised models trained for 8 subjects for our method and a 1D convolution of the ECG beat for contrast. Finally, in Sect. 4, the results are put into context and we discuss future works and propose that the methodology has the potential to overcome the constraints of existing efforts, including physiological heterogeneity, by examining the ECG at the level of individual heartbeats.

378

O. Cisuelo et al.

2 Methodology

Fig. 1. Work flow of the system. a) Data extraction, processing, and normalisation. b) Model architecture for the proposed convolutional neural network that takes the STFT coefficients of an ECG beat as input and outputs the glycaemic status (hyperglycaemia or normal). c) Visualisation of the learned features in the final model.

2.1 Dataset Data for this study was collected from an ongoing observational study at the Unit of Endocrinology and Diabetes at Bambino Gesu‘ Children’s Hospital, Rome, Italy between April 2021 and December 2021. Paediatrics with T1D who were under the care of the unit and already routinely used a continuous glucose monitoring (CGM) device were enrolled onto the protocol and physiological data was recorded for a period of up to 3 days [20]. We selected 8 subjects aged 11 or younger with a mean age of 9.3 ± 1.3 years. The study was registered with the clinical trials database (ref NCT05278143). Continuous physiological data was recorded using the Medtronic Zephyr Biopatch, a CE marked device that records ECG across a single lead and operates within an amplitude range between 0.25 and 15 mV, with a sampling frequency of 250 Hz. In addition to ECG, the device also contains a 3-axis accelerometer to detect activity level and posture. Activity is reported by the device as Vector Magnitude Units (VMU) in units of g, where VMU = (x2 + y2 + z 2 ) and x, y, and z are the averages of the three axial acceleration magnitudes over the previous 1 s, sampled at 100 Hz. Activity levels of 0.2 and 0.8 correspond to walking and jogging respectively [21]. The CGMs eligible for this study were Dexcom G6, Abbott FreeStyle Libre Flash, and Medtronic Guardian. All 3 CGM

Spectrogram-Driven Convolutional Neural Network for Real-Time

379

devices sample glucose in the interstitial fluid and are situated on the back of the upper arm, abdomen, or upper buttocks, depending on the specific device specifications and user preference. The Libre Flash reports glucose measurements at 15-min intervals, whereas the Dexcom and Medtronic devices report every 5 min. 2.2 Data Processing and Selection Glucose data was resampled to 1 Hz using linear interpolation and the timestamp was set back 10 min uniformly to account for the delay between glucose in the blood and interstitial fluid [22]. ECG and glucose data were time-matched and ECG samples without corresponding glucose values were discarded. The dataset was obtained under real-world conditions, therefore it is subject to noise. Extracting usable excerpts from the ECG signal was performed in several stages. First, our deep learning tool [23] was used to filter the signal and remove baseline wander before identifying the ECG wave sections (P, QRS, and T). This was followed by a series of physiological constraints, such as minimum refractory time between successive QRS complexes and a negative slope between the R and S points to eliminate any potential false positive QRS detections. ECG beats consisting of 165 samples, 50 samples before the peak of the QRS complex and 114 after, with low noise during low activity (200 mg/dl), therefore, a binary classifier was developed. Models were trained and tested on 5,000 samples for each glycaemic status, resulting in a final dataset of 80,000 ECG beats. The glucose range of samples and subject characteristics are shown in Table 1. Table 1. Subject characteristics and glucose range and mean of the ECG beats in mg/dl Subject

Age

Sex

Normal range (mean)

Hyperglycaemic range (mean)

1

10

M

70–129 (110)

250–294 (265)

2

10

F

91–97 (93)

186–288 (213)

3

11

M

70–77 (75)

184–243 (208)

4

7

F

101–135 (118)

246–302 (265)

5

8

F

116–168 (146)

205–295 (220)

6

10

M

70–117 (103)

206–286 (225)

7

9

M

78–98 (91)

268–352 (281)

8

10

F

70–87 (81)

249–316 (285)

380

O. Cisuelo et al.

2.3 Short Time Fourier Transform The STFT is a sequence of Fourier transforms of a signal across short overlapping windows which provides time-localised signal strengths where the frequency of the signal varies over time. The STFT was computed for each beat and is given by the equation XSTFT [m, n] =

L−1 

k

x[k]g[k − m]e−2jπ n L

(1)

k=0

where x[k] is signal, g[k − m] is a Hann window function at time m, n is frequency, and L is the number of samples in the signal. The absolute value of the coefficients were normalised {0, 1}. The size of the coefficient array is given by the length of the windowed signal and the number of frames. Therefore, an ECG beat defined to be 165 samples (660 ms), yields a STFT array with dimensions of 129 × 166. An example of the resulting image of the STFT coefficients is shown in Fig. 2.

Fig. 2. Example of a 165 sample ECG beat and the corresponding STFT coefficients

2.4 Convolutional Neural Networks CNNs consist of many layers which are locally connected usually used for the processing and classification of two-dimensional data. In the convolution stage, input layers are used to extract deep, representative, and discriminative features from the input data followed by the activation layer where non-linearity in the data is charted. In the pooling stage features and computational complexities are reduced then a fully connected layer that takes the output of convolution and pooling to sort the output into the corresponding class. 2.5 Model Architecture and Training For each subject, the dataset was divided into training and testing sets, with an equal distribution of normal and hyperglycaemic ECG beats in each set. Models were trained

Spectrogram-Driven Convolutional Neural Network for Real-Time

381

over 100 epochs using the Adam optimisation algorithm with a learning rate of 0.001. Model training was conducted on a system comprising an Intel i7-11800H processor, 16 GB system memory, and an Nvidia RTX 3060 laptop graphics processing unit with 6 GB video memory. STFT 2D Each ECG beat is represented by a 129 × 166 × 1 single-channel image along with a label corresponding to the glycaemic status. The model consists of 3 convolutional layers each featuring rectified linear unit (ReLU) activation function and max pooling to reduce the dimensionality output from the convolution process as depicted in Fig. 1. In the first convolutional layer, we have 8 filters with kernel size 4 × 4 and a stride of 1 × 1 resulting in an output size of 126 × 163 × 8 which is reduced to 63 × 81 × 8 after max pooling. The second and third layers have 13 filters with kernel size 2 × 2 resulting in an output of 15 × 19 × 13. We then add 50% dropout to avoid overfitting followed by the fully connected dense layers. The final output of the model is a binary classification of the glycaemic status as either normal or hyperglycaemic. ECG 1D A 1D CNN was created to train on the ECG beats using the same configuration of filters, activation function, and pooling. The input signal was normalised {0, 1}. 2.6 Model Interpretation To explain how the model arrives at the binary classification we used the SHAP (SHAPley Additive exPlanations) approach. SHAP is a visualisation tool that can be used for making a machine learning model more explainable by assigning each feature an importance value [26]. We employed SHAPs DeepExplainer, which can determine the magnitude of each features contribution to the output of the model.

3 Results The primary objective of this study was to demonstrate the proof-of-concept methodology using real-world data obtained from wearable sensors in free-living conditions. Subject-specific models for the eight included subjects were trained on 2,500 of both normal and hyperglycaemic ECG beats and evaluated with a different 2,500 of each glycaemic status. Figure 3 (left) shows the receiver operating characteristic curves for all eight models demonstrating the relationship between sensitivity and specificity of the hyperglycaemia detection model. Across all eight subjects 40,000 ECG beats were evaluated (5,000 for each subject) and the results are summarised in a confusion matrix in Fig. 3 (right). An overall accuracy of 96.9% was achieved across all test beats. The proposed model offers improved overall performance than a 1D CNN based on the ECG beat, in particular in recall. Recall measures the proportion of actual positive cases that are correctly identified by the model and is regarded as being important for biomedical applications since it is desirable to capture as many true-positive cases as possible [24]. The performance metrics of the proposed models, aggregated across all 8 models are shown in Table 2. Towards developing a robust interpretation of the model we calculated the SHAP values for the frequency components at each column of the STFT array as shown in

382

O. Cisuelo et al.

Fig. 3. Receiver operating characteristic curve for subject-specific models (left) and corresponding confusion matrix (right) for our spectrogram-driven multi-layer 2D CNN.

Table 2. Performance metrics aggregated over all subject-specific models Performance metric

STFT (2D)

ECG (1D)

Accuracy

0.969

0.940

Precision

0.967

0.955

Recall

0.972

0.923

F-score

0.969

0.939

Fig. 4. Image of STFT coefficients (left) and SHAP values (right) for a single beat correctly classified as hyperglycemia for subject 2.

Fig. 4. The model for subject 2 was able to correctly identify all 2,500 tested hyperglycaemic beats correctly. SHAP values in Fig. 4 with the most positive contributions to the hyperglycaemia classification are shown in dark, and occur at columns 17, 33, 42, and 67 of the STFT. This corresponds to the P-wave, P-Q transition, Q-point inflection, and the transition from the QRS complex into the T-wave as shown by the morphological

Spectrogram-Driven Convolutional Neural Network for Real-Time

383

Fig. 5. The average ECG beat for subject 2 from all beats selected for testing and training. The shaded area shows 1 standard deviation. Vertical lines show the sample number where the SHAP analysis shows the greatest positive contribution to the hyperglycaemia classification.

differences between the ECG beats for the glycaemic states as shown in Fig. 5. We observe higher-order harmonic frequencies from the STFT also have an influence on the model, the impact of which bares further investigation.

4 Discussion In this study we presented a novel spectrogram-driven deep learning model for hyperglycaemia detection in a paediatric population with type-1 diabetes using wearable sensors. The ECG waveform is subject to individual differences such as overall health and age, particularly the QRS complex and t-wave [25]. Our system can effectively learn the unique characteristics of personalised ECG patterns and categorise the glyaemic status of each ECG beat with a high degree of accuracy, distinguishing between normal and hyperglycaemia for low-activity beats. The results should be seen in light of their limitations, which are severalfold. Firstly, the protocol is still ongoing, thus we were only able to use a subset of the total data that will be available upon completion of the study, therefore the results presented here should be viewed as intermediate. Secondly, only data where the subject had low activity and was at rest was included to minimise the covariates introduced by activity-related ECG changes. We did not select for diurnality, therefore the samples included in the training and evaluation of the models may be from day or night time. Ultimately the proposed method relies on obtaining a high-quality ECG trace from the wearable sensor. We must acknowledge the limitation of a single-lead ECG, which offers a restricted view of the cardiac electrical axis [27]. The pragmatic approach of the protocol provides continuous ECG recording in real-life conditions where the subject is able to go about their normal daily activities unencumbered by the wearable sensors versus a more complete picture of the heart’s electrical activity. The dataset contained few hypoglycaemia samples, therefore the development of a multiclass glycaemic event detection model was not

384

O. Cisuelo et al.

possible. Finally, the results presented are subject-specific. To develop a general cohort model more data is required. We propose that this methodology has the potential to overcome the constraint of physiological heterogeneity by the approach of analysing individual heartbeats and is therefore independent of resting heart rate and HRV. The results indicate that spectrogram-driven 2D CNNs may be suitable for time series problems where spatial or frequency patterns are also important. 1D CNNs may not be able to capture spatial or frequency patterns in the time series data, which may have significance in biomedical applications. Deep learning has emerged as the state-of-the-art tool in artificial intelligence research due to advanced techniques that minimise the percentage of error compared to more traditional machine learning methods, and the ability to handle large datasets. However, due to complex architectures and the number of parameters, there are still challenges around the interpretability of models for use by medical experts in a clinical setting [28]. To overcome this we propose that the relevant features learned by the model can be interpreted in such a way that the outcome can be linked to underlying physiological measurement. 2D CNNs produce feature maps, the learned representations of the input data, that are not aligned with the time axis, which can make it difficult to interpret the results. The choice of parameters in the STFT calculation was set to preserve the temporal axis in the input image for convenient mapping back to the time-series ECG beat. 4.1 Future Work In addition to the ongoing data collection protocol involving paediatric patients with T1D, a concurrent study is being conducted to gather data from 30 adults with T1D [29]. The study was registered with the clinical trials database (ref NCT05461144). Upon completion of both studies, it is anticipated that we will possess a unique dataset encompassing several days of continuous physiological data acquired under real-life conditions for people with T1D. For a model to be clinically useful it should demonstrate generalisability. Therefore, we will test the model on unseen data using a subject-wise crossvalidation approach [30]. We plan to investigate the performance of using excerpts of varying time intervals to classify the glycaemic status of the excerpt with a majority voting scheme based on ECG beats. This approach was adopted in a study to develop a cardiovascular disease detection model which demonstrated that using excerpts of 4 min improved classification performance [17]. We also aim to examine data of varying activity levels and investigate ECG morphology change with respect to glucose concentration for nocturnal and daytime periods. We will also experiment with different spectrogram methods such as wavelets, which may offer improved time-localisation [31] in addition to there being many wavelet functions available, allowing the most appropriate to be chosen for the morphology of the signal [32]. The steps laid out will go towards addressing the limitations discussed previously.

Spectrogram-Driven Convolutional Neural Network for Real-Time

385

5 Conclusion This study presents a novel method for non-invasive glycaemic event detection using single ECG beats extracted from a wearable sensor, incorporating our deep learning ECG segmentation tool which is able to reliably detect QRS segments from single-lead ECG and extract beats [23]. Real-time non-invasive monitoring of glycaemic events is crucial in the management of progressive conditions such as diabetes. Our novel spectrogramdriven deep learning-based glycaemic detection tool has the potential to be integrated into wearable sensors for non-invasive real-time glycaemic event detection for effective diabetes management.

References 1. Lasker, S.P., McLachlan, C., Wang, L., Ali, S.M.K., Jelinek, H.: Discovery, treatment and management of diabetes. J. Diabetol. Feb. 2010(1), 1–8 (2010) 2. Mobasseri, M., Shirmohammadi, M., Amiri, T., Vahed, N., Fard, H.H., Ghojazadeh, M.: Prevalence and incidence of type 1 diabetes in the world: a systematic review and metaanalysis. Health Promot. Perspect. 10(2), 98 (2020) 3. Nathan, D.M., Group, D.E.R.: The diabetes control and complications trial/epidemiology of diabetes interventions and complications study at 30 years: overview. Diabetes Care 37(1), 9–16 (2014) 4. Rodriguez-Le´on, C., Villalonga, C., Munoz-Torres, M., Ruiz, J.R., Banos, O.: Mobile and wearable technology for the monitoring of diabetes-related parameters: systematic review. JMIR Mhealth Uhealth 9(6), e25138 (2021) 5. Peng, P., Zhang, N., Huang, J., Jiao, X., Shen, Y.: Effectiveness of wearable activity monitors on metabolic outcomes in patients with type 2 diabetes: a systematic review and meta-analysis. Endocr. Pract. (2023) 6. Ahmed, A., Aziz, S., Abd-alrazaq, A., Farooq, F., Sheikh, J.: Overview of artificial intelligence–driven wearable devices for diabetes: scoping review. J. Med. Internet Res. 24(8), e36010 (2022) 7. Georga, E.I., Fotiadis, D.I., Tigas, S.K.: 7 - Prediction models of hypoglycemia. In: Georga, E.I., Fotiadis, D.I., Tigas, S.K. (eds.) Personalized Predictive Modeling in Type 1 Diabetes, pp. 153–171. Academic Press (2018) 8. Chan, K.Y., Ling, S.-H., Dillon, T.S., Nguyen, H.T.: Diagnosis of hypoglycemic episodes using a neural network based rule discovery system. Expert Syst. Appl. 38(8), 9799–9808 (2011) 9. Nuryani, N., Ling, S.S., Nguyen, H.: Electrocardiographic signals and swarm-based support vector machine for hypoglycemia detection. Ann. Biomed. Eng. 40, 934–945 (2012) 10. Cichosz, S.L., Frystyk, J., Hejlesen, O.K., Tarnow, L., Fleischer, J.: A novel algorithm for prediction and detection of hypoglycemia based on continuous glucose monitoring and heart rate variability in patients with type 1 diabetes. J. Diabetes Sci. Technol. 8(4), 731–737 (2014) 11. Association between heart rate variability measures and blood glucose levels: implications for noninvasive glucose monitoring for diabetes. Diabetes Technol. Ther. 18(6), 366–376 (2016) 12. Li, J., Tobore, I., Liu, Y., Nie, Z.: Non-invasive monitoring of three glucose ranges based on ECG by using DBSCAN-CNN. IEEE J. Biomed. Health Inform. 25(9), 3340–3350 (2021) 13. Porumb, M., Griffen, C., Hattersley, J., Pecchia, L.: Nocturnal low glucose detection in healthy elderly from one-lead ECG using convolutional denoising autoencoders. Biomed. Signal Process. Control 62, 102054 (2020)

386

O. Cisuelo et al.

14. Porumb, M., Stranges, S., Pecchia, L.: Precision medicine and artificial intelligence: a pilot study on deep learning for hypoglycemic events detection based on ECG. Sci. Rep. 10(1), 170 (2020) 15. Cordeiro, R., Park, Y.: Hyperglycemia identification using ECG in deep learning era. Sensors 21(18), 6263 (2021) 16. Alaskar, H.: Deep learning-based model architecture for time-frequency images analysis. Int. J. Adv. Comput. Sci. Appl. 9(12) (2018) 17. Haleem, M.S., Castaldo, R., Pagliara, S.M., Franzese, M., Pecchia, L.: Time adaptive ECG driven cardiovascular disease detector. Biomed. Signal Process. Control. 70, 102968 (2021) 18. Alqudah, A.M., Alqudah, A.: Deep learning for single-lead ECG beat arrhythmia-type detection using novel iris spectrogram representation. Soft. Comput. 26(3), 1123–1139 (2021). https://doi.org/10.1007/s00500-021-06555-x 19. Michelucci, U.: Fundamentals of convolutional neural networks. In: Advanced Applied Deep Learning: Convolutional Neural Networks and Object Detection, pp. 79–123. Apress (2019) 20. Andellini, M., et al.: Artificial intelligence for non-invasive glycaemic-events detection via ECG in a paediatric population: study protocol. Health Technol. 1–10 (2023) 21. Zephyr Technology: Bioharness log data descriptions. https://www.zephyranywhere.com/ media/download/bioharness-log-data-descriptions-07-apr-2016.pdf. Accessed 01 Feb 2023 22. Basu, A., Dube, S., Slama, M., Errazuriz, I., Cobelli, C., Basu, R.: Time lag of glucose from intravascular to interstitial compartment in humans. Diabetes 62(12), 4083–4087 (2013) 23. Haleem, M.S., Pecchia, L.: A deep learning based ECG segmentation tool for detection of ECG beat parameters. In: 2022 IEEE Symposium on Computers and Communications (ISCC), Rhodes, Greece, pp. 1–4 (2022). https://doi.org/10.1109/ISCC55528.2022.9912906 24. Hicks, S.A., Stru¨mke, I., Thambawita, V., Hammou, M., Riegler, M.A., Halvorsen, P., Parasa, S.: On evaluation metrics for medical applications of artificial intelligence. Sci. Rep. 12(1), 5979 (2022) 25. Gertsch, M.: The normal electrocardiogram and its (normal) variants. In: The ECG Manual: An Evidence-Based Approach, pp. 17–36 (2009) 26. Lundberg, S.M., Lee, S.-I.: A unified approach to interpreting model predictions. Adv. Neural Inf. Process. Syst. 30 (2017) 27. Abdou, A., Krishnan, S.: Horizons in single-lead ECG analysis from devices to data. Front. Signal Process. 2 (2022) 28. Hong, S., Zhou, Y., Shang, J., Xiao, C., Sun, J.: Opportunities and challenges of deep learning methods for electrocardiogram data: a systematic review. Comput. Biol. Med. 122, 103801 (2020) 29. Cisuelo, O., et al.: Development of an artificial intelligence system to identify hypoglycaemia via ECG in adults with type 1 diabetes: protocol for data collection under controlled and free-living conditions. BMJ Open 13(4), e067899 (2023) 30. Tougui, I., Jilbab, A., Mhamdi, J.E.: Impact of the choice of cross-validation techniques on the results of machine learning-based diagnostic applications. Healthc. Inform. Res. 27(3), 189–199 (2021) 31. Dinh, H.A.N., Kumar, D.K., Pah, N.D., Burton, P.: Wavelets for QRS detection. Australas. Phys. Eng. Sci. Med. 24(4), 207 (2001) 32. Addison, P.S.: Wavelet transforms and the ECG: a review. Physiol. Meas. 26(5), R155 (2005)

Spectrogram-Based Approach with Convolutional Neural Network for Human Activity Classification Martina Sassi1(B)

, Muhammad Salman Haleem2

, and Leandro Pecchia1

1 Department of Engineering, Università Campus Bio-Medico di Roma, Via Álvaro del Portillo,

200, 00128 Rome, Italy {martina.sassi,leandro.pecchia}@unicampus.it 2 School of Engineering, University of Warwick, Coventry CV4 7AL, UK [email protected]

Abstract. Human activity recognition (HAR) is an expanding research field for analyzing holistic wellbeing trajectory, frailty detection and prevention of critical situations. With the increased availability of wearables and novel machine learning methods, the automatic recognition of human activities is exploited by real-time signals via Deep Learning techniques. This is due to their capability of learning contextual and localized patterns which give them a significant edge over traditional machine learning approaches. However, most of the state-of-the-art deep learning techniques have limitations due to limited number of features present in temporal dimension. In this regard, we propose Spectrogram-driven multilayer 2D-Convolutional Neural Network (2D-CNN) to classify among different types of human activities using triaxial accelerometer data obtained under MEDICON Scientific Challenge. The spectrogram has significant advantage over 1D time domain signals due to their capability to extract power spectrum in time as well as in frequency domain. The dataset consists of twelve activities of daily living and three types of simulated falls performed by subjects wearing a single accelerometer. In total, the dataset was composed by 468 instances. The spectrograms were determined by Short Time Fourier Transform (STFT) from the continuous signal obtained from X-, Y-, and Z-axis of the accelerometer signals. Experimental results show that our spectrogram driven 2D-CNN model reach an overall accuracy of 86.02% and an overall F1 -score of 81.09% in classifying all the activity classes; significantly outperforming the deep learning architecture based on 1D time domain signal. Keywords: Accelerometer data · Human activity recognition · Convolutional neural network · Deep learning

1 Introduction Human activity recognition (HAR) is an expanding research field that aims at defining techniques able to automatically recognize physical activities performed by people [1–3]. This analysis has attracted much interest in recent years because of the rapid © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 387–401, 2024. https://doi.org/10.1007/978-3-031-49068-2_40

388

M. Sassi et al.

growth of application demands in several domains such as irregularity detection and correction in various motions, surveillance, injury prevention, monitoring of progress in rehabilitation, healthcare monitoring, elderly care, frailty detection/prevention and holistic wellbeing etc. [4–6]. The automatic recognition of activities of daily living (ADL) and fall detection are particularly crucial to assess the functional ability of an elderly person to live independently in everyday living [7, 8]. Generally, there are two types of HAR: video-based HAR and sensor-based HAR [2, 9, 10]. The first one analyzes images or videos from cameras, while sensor-based HAR relies on the motion data gathered from sensors [9, 11]. Applications such as monitoring ADL require a sensor as an acquisition source [12]. Among the available wearable sensors, the accelerometers are the most widely used for HAR due to their several advantages such as low cost, small size, and portability [1]. Traditionally, HAR is performed by machine learning (ML) techniques that consists in two steps, i.e., feature extraction and pattern classification [13, 14]. The first step is extremely important since the quality of the features extracted from raw data mostly determines the overall system accuracy [15]. However, the HAR based on traditional ML approaches has been challenging due limited capability of modelling sequential data. In recent years, deep learning (DL) techniques have shown their outstanding performances in pattern recognition application and sequential learning [13]. DL approaches considerably reduce the effort on feature engineering (i.e., feature extraction and feature selection) due to their capability to extract hidden information for accurate detection and interpretation [16]. Previously, DL methods have been developed to predict different activity types based on time series signals obtained from accelerometer via 1D-CNN [17]. However, the HAR based on 1D-time signal has been challenging due to limited capturing of features unique to specific activity types. To address these, we introduced spectrogram driven deep learning model due to its capability to capture power spectrum both in time as well as in frequency domain. We transformed a one-dimensional time series signal to two-dimensional time-frequency space via the short-time Fourier transform (STFT) in order to represent the changes in frequency content of a non-stationary signal over time [18]. The time-frequency approach can provide a suitable basis for the discovery of complex, high-dimensional properties of physiological signals. Commonly used DL method is Convolutional Neural Network (CNN) which trains n-dimensional features to extract local spatial patterns [6, 14]. CNN process through layers of convolutional operations to detect spatial patterns followed by pooling and dropout layers to detect most prominent spatial patterns in STFT while avoiding overfitting with respect to specific activity type [19]. The 2D-CNN had been trained on spectrograms obtained from triaxial accelerometer data to recognize twelve activities of daily living and three types of simulated falls. The aim of this paper is to represent proof-of-concept spectrogram driven CNN model for real-time detection of human activity recognition and is structured as follows: Sect. 2 explains the methodology, the dataset description where the methodology has been trained and implemented, the training protocol and the experimental settings. Section 3 presents the results and 1D-convolutional model for the contrast. Section 4 concludes the study while suggesting future works to improve the model.

Spectrogram-Based Approach with Convolutional Neural Network

389

2 Methodology 2.1 Materials Sixteen young healthy subjects were recruited (mean ± standard deviation: age 21.9 ± 2.2 years old; height 178.1 ± 7.8 cm; body mass 70.0 ± 12.1 kg). Each subject performed twelve activities of daily living (ADLs) and three types of simulated falls while wearing a Shimmer3 inertial measurement unit (IMU). The accelerometer data were collected with a sampling frequency of 204.8 Hz. The sensor was attached sideways to their waist at belt high. Figure 1 shows a representation of a subject wearing the device.

Fig. 1. a) Coordinate system of the Shimmer3 inertial measurement unit sensor; b) Shimmer3 placement on a subject.

The twelve distinct ADLs performed were: walking, fast walking, running, fast running, jumping, high jumping, sitting, standing up, lying down, getting up from lying position, walking down the stairs and walking up the stairs. The three simulated falls on a 2 cm thick tatami mattress were: forward fall, sideways fall and backward fall. The twelve ADLs are subdivided in 10 classes (as reported in Table 1), which in turn are organized into 5 different groups (see Table 2). In total, the dataset was composed by 468 instances. Figure 2 shows the waveforms of accelerometer signals describing each activity class. 2.2 Deep Learning-Based Activity Classification Algorithm In this paper, a deep learning-based activity classification algorithm has been developed by using only accelerometer data. The algorithm was implemented in MATLAB(R) (version R2022b, TheMathWorks® Inc., Natick, MA, USA). The deep learning approach is composed of the following steps: (1) signal pre-processing, (2) spectrogram generation,

390

M. Sassi et al. Table 1. Activities of daily living (ADL) and types of simulated falls.

Class

Activity

Abbreviation # of instances % of instances

Class 1

Walking (walking and fast walking) MW

68

14.53

Class 2

Running (running and fast running) MR

68

14.53

Class 3

Jumping (jumping and high jumping)

MJ

68

14.53

Class 4

Walking down the stairs

WD

13

2.7778

Class 5

Walking up the stairs

WU

13

2.7778

Class 6

Forward fall

FF

34

7.265

Class 7

Sideways fall

FS

34

7.265

Class 8

Backward fall

FB

34

7.265

Class 9

Lying down

LD

34

7.265

Class 10 Other classes (sitting, standing up OT and getting up from lying position)

102

21.795

Table 2. Different groups including all classes and activities. Group

Activities included in group

Group 1 - Moving

Walking (walking and fast walking) Running (running and fast running) Jumping (jumping and high jumping)

Group 2 - Stairs

Walking down the stairs

Group 3 - Falls

Forward fall

Walking up the stairs Sideways fall Group 4 - Lying down

Lying down

Group 5 - Inactive

Other classes

and (3) CNN-based classification. Figure 3 shows the block diagram of the proposed deep learning based human activity classification algorithm. Signal Preprocessing. In the signal preprocessing step, the raw sensor data were low-pass filtered with a fourth order Butterworth filter with a cut off frequency of 8 Hz to attenuate the high frequency noise components introduced in the data. Then, Acceleration Vector Magnitude (AVM) was calculated from X-, Y-, and Z-axis acceleration according to the following formula:  (1) AVM (t) = ax2 (t) + ay2 (t) + az2 (t)

Spectrogram-Based Approach with Convolutional Neural Network

391

Fig. 2. Waveforms of accelerometer signals describing each activity. In sequence, from left to right: walking, running, jumping, walking downstairs, walking upstairs, forward fall, sideways fall, backward fall, lying down, other classes.

392

M. Sassi et al.

Fig. 3. Block diagram of the human activity recognition process.

The sizes of each signal were inconsistent. Since CNN takes fixed length input for training and testing, each acceleration signal has been adapted to the size of the signal with the longest length. Successively, the Z-score normalization (also called standardization) was applied to normalize the accelerometers’ signals in order to minimize the single magnitude variations across users. The Z-score normalization is defined as shown in Eq. 2: SN (t) =

S(t) − μ σ

(2)

where S(t) and SN (t) are the original and normalized acceleration or AVM signals, respectively, and t is the time steps in the interval. Whereas μ and σ are the mean and standard deviation of the signals, respectively. Spectrogram generation. After the procedure of signal normalization, each acceleration and amplitude signal were converted into time-frequency based 2D spectrogram. In particular, the short-time Fourier transform (STFT) was applied to generate those spectrograms, and to transforms the original time-domain signals to time-frequency domain signals [20]. The STFT is sequence of Fourier transforms of a signal across short overlapping windows [21]. It provides time-localized signal strengths where frequency of the signal varies over time: so time is on one axis, frequency on the other, and the brightness represents the intensity of a frequency component at each interval [20]. Spectrogram representation was used as it provides signal features as a function of frequency and time [22]. The equation of STFT of the normalized signals is defined as follows: STFT {x[k]} = XSTFT [m, n] =

L−1 

x[k]g[k − m]e−j

2π nk L

(3)

k=0

where x[k] is the normalized signals, g[k − m] is the window function with m as the center of the window function [23], n is the frequency axis, and L are the total number of data points in x[k]. In this paper, the STFT was computed segmenting the time signals using the Hamming window, whose definition is given in Eq. 4. The window size was set to 50 samples, with 10% overlap over time domain between segments.    2π m ,0 ≤ m ≤ M − 1 (4) g[k] = 0.5 1 − cos M −1 The STFT was calculated for each acceleration and AVM signal, resulting in feature size equal to 129 × 128. Figure 4 shows examples of the spectrograms for activities belonging to the five groups of moving (running), stairs (walking down the stairs), falls (backward fall), lying down, and other classes. Multilayer Convolutional Neural Network. Once the spectrograms of the X-, Y-, Z-axis normalized accelerometer signals and AVM signals were obtained through the

Spectrogram-Based Approach with Convolutional Neural Network

393

Fig. 4. Spectrogram examples of the X-, Y-, Z-axis acceleration data from five activities belonging to the five different groups. In sequence, from top to bottom: running, walking downstairs, backward fall, lying down, other classes.

spectrogram generation procedure, each spectrogram image was then considered as the input data of the convolutional neural network (CNN) classifier (see Fig. 5). In particular,

394

M. Sassi et al.

CNN takes as input the concatenated spectrograms signals from the triaxial accelerometer data (i.e., 129 × 128 × 3). A unique label was attributed to each spectrogram (see Table 1), providing ten activity classes in total.

Fig. 5. Architecture of the proposed CNN model.

There are five main layers in the CNN architecture: convolutional layer, activation layer, max-pooling layer, flatten layer, fully connected layer, and softmax layer [24]. The CNN has several convolutional layers that act as a feature extractor, meaning that they are capable to extract high-level features from the images input data. The overall architecture of the multilayer 2D-CNN developed is shown in Fig. 5 and described below (see Table 3). Each CNN layer is equipped by a ReLU (Rectified Linear Unit) activation function and a Max Pooling layer with pooling size P and stride D. A 20% dropout layer was added between each convolutional block to avoid overfitting. Then, the output of these several convolutional and max-pooling layers are flattened into a one-dimensional features vector, used as the input for the fully-connected layer. Finally, the output from these layers is passed to a softmax layer that computes the probability distribution over the predicted activity class. The final architecture of the CNN was determined by a trial-and-error process. The proposed structure shown in Fig. 5 contains 4 convolutional blocks, one flatten layer, one fully-connected layer, and one softmax layer. The first block consists of two consecutive convolutional layers where 16 filters are applied with kernel size = 11 × 11 and with a zero-padding. This is followed by a ReLU layer and a max-pooling layer with pool size P = 3 × 3 and stride D = 2 × 2, resulting output of 65 × 64 × 16. The second convolutional block takes the previous output and applies 32 filters with kernel size = 7 × 7 and with a zero-padding. This is followed by a ReLU layer and a max-pooling layer with pool size P = 3 × 3 and stride D = 2 × 2, resulting output of 33 × 32 × 32. The third and the fourth convolutional blocks take the previous output and apply 32 filters with kernel size = 7 × 7 and with a zero-padding. These are both followed by a ReLU layer and a max-pooling layer with pool size P = 3 × 3 and stride D = 2 ×

Spectrogram-Based Approach with Convolutional Neural Network

395

2, resulting output of 9 × 8 × 32. Then, the flatten layer reshape the dimensions of the input, providing a vector of 2304 × 1. Finally, the size of the features vector was reduced from 2304 × 1 to 10 × 1, in order to apply the softmax and the classification layers. Table 3. Architecture details of CNN. Layer

Feature map

Size

Kernel size

Stride

Activation

Image

1

129 × 128 × 3

–

–

–

2 × conv

16

129 × 128 × 16

11 × 11

1

ReLU

Max pooling

16

65 × 64 × 16

3×3

2

ReLU

2 × conv

32

65 × 64 × 32

7×7

1

ReLU

Max pooling

32

33 × 32 × 32

3×3

2

ReLU

1 × conv

32

33 × 32 × 32

7×7

1

ReLU

Max pooling

32

17 × 16 × 32

3×3

2

ReLU

1 × conv

32

17 × 16 × 32

7×7

1

ReLU

Max pooling

32

9 × 8 × 32

3×3

2

ReLU

Flatten

–

2304 × 1

–

–

–

Fully connected

–

10 × 1

–

–

–

Softmax

–

–

–

–

–

2.3 Training Protocol and Metrics Training and testing protocol. The entire dataset was divided into two portions: 80% as training portion (375 samples) used for training the neural network, and 20% as test portion (93 samples) used to assess its final performances. The training of the CNN was performed using the k-fold cross validation (CV) as validation method (Fig. 6). The total number of training samples were divided into k folds, and then the model was trained on samples of k − 1 folds and tested on the remaining fold. The k results obtained for all the experiments were then averaged to provide a single estimation of training performances. The model that achieved the best performances among the different experiments was then applied to predict labels of the test set. In particular, 2-folds, 5-folds and 10-folds cross validation strategies were implemented. Performance metrics. A confusion matrix (CM) is a table that enables the visualization of the performance of the algorithm. The rows of the CM correspond to the true classes, whereas the columns correspond to the predicted classes. The diagonal cells represent those samples that are correctly classified, while the off-diagonal values are the incorrectly classified samples. Various performance metrics can be extracted from the CM. Measures of accuracy (Acc), specificity (Sp), sensitivity (Se) or recall, precision (Pr), and Fβ -score were utilized to evaluate model’s performances, and defined as

396

M. Sassi et al.

Fig. 6. k-fold cross validation. Split the data into k folds (k = 5 as example). The model is trained on data of k − 1 folds, and tested on the remaining fold.

follows: Acc =

TP + TN TP + TN + FP + FN

(5)

Sp =

TN TN + FP

(6)

Se =

TP TP + FN

(7)

TP TP + FP  1 + β 2 · Pr · Se Fβ = β 2 · (Pr + Se) Pr =

(8)

(9)

where TP is the number of true positives, TN is the number of true negatives, FP is the number of false positives, and FN is the number of false negatives. In Eq. 9, β is a weighting factor that controls the degree of importance of sensitivity and precision. This parameter is a positive real number. In this paper β was set equal to 1, to give the same importance to both sensitivity and precision. Experimental settings. Data analysis was performed in MATLAB(R) . The weights of the proposed network were initialized by Xavier initialization. During the training phase, the parameters (weights and bias) in each layer of the network are learnt using back-propagation algorithm. In particular, the Adam Moment Estimation optimization method with a learning rate of 0.001 was deployed to update the network parameters for minimizing a categorical cross-entropy loss function [13, 25]. The forward and back propagation process are repeated until a stopping criterion is satisfied [16]. In this paper, the stopping condition was the achievement of the maximum number of epochs. The neural network was trained using 400 epochs.

Spectrogram-Based Approach with Convolutional Neural Network

397

3 Results and Discussion Table 4 compares results between the 1D-CNN model and the 2D-CNN model. In the 1DCNN model, the input of the neural network were the accelerometer time signals. From the table, we can find that the 1D-CNN model achieved an average accuracy of 75.3% and a F1 -score of 70.1%. Whereas the 2D-CNN model achieved an average accuracy of 86.02% and a F1 -score of 81.10%. According to the results, 2D-CNN outperforms over 1D-CNN in terms of classification performances on the test set. This shows that deep learning model developed on spectrogram of triaxial signals significantly outperformed that developed on 1D time domain signals. Table 4. Classification performances comparison of the proposed spectrogram-based (2DCNN) and acceleration time signals-based (1D-CNN) deep learning classification algorithm (implementing 5-folds cross validation). Metrics Performance Acc

Sp

Se

Pr

Fβ

1D-CNN

0.7527

0.9713

0.6675

0.7978

0.7012

2D-CNN

0.8602

0.9836

0.7989

0.8595

0.8109

Acc: accuracy. Sp = specificity. Se = sensitivity. Pr = precision. F β = F1 -score

To better investigate the approach of the proposed 2D-CNN model, a comparison of the performances was carried out when different inputs were given to the network: X-, Y-, Z-axis spectrogram data, and AVM spectrogram data. Table 5 summarizes and compares the classification performances of the algorithm using those different inputs. Even if the AVM combines the triaxial accelerometer data into one signal, given the spectrogram extracted from the AVM as input to the neural network does not improve classification performances. Using only the AVM signal, some activities’ information can be missed, such as on what axis the motion mainly occurs. The spectrogram extracted from the AVM signal does not reflect all the intensity of motions as the spectrograms extracted from the triaxial accelerometer signals. Therefore, the information content by reducing the size to one channel seems to be not sufficient for the differentiation of different classes. Indeed, implementing the 5-fold cross validation strategy, average values of performance metrics obtained using the triaxial spectrogram data and the AVM data as input respectively were: 86.02% and 73.87% for the accuracy metric, and 81.10% and 67.78% for the F1 -score metric. Classification performances decreased implementing 10-folds cross validation: 83.87% and 67.74% for the accuracy metric, and 80.10% and 54.71% for the F1 -score metric, when using the triaxial spectrogram data and the AVM data as input respectively. Table 6 summarizes and compares the classification performances of the algorithm implementing the different cross-validation strategies and using the concatenated X-, Y-, Z-axis spectrogram data as input of the network. The overall accuracies equal to 80.64%, 86.02%, and 83.87% were achieved by the 2-fold, 5-fold, and 10-fold crossvalidation, respectively. Values equal to 70.75%, 81.09%, and 80.10%, were obtained for

398

M. Sassi et al.

Table 5. Classification performances comparison of the proposed spectrogram-based (2D-CNN) when different inputs were given to the neural network: X, Y, Z spectrogram, or AVM spectrogram. Input

K-fold

X-, Y-, Z-spectrogram AVM spectrogram

Metrics performance Acc

Sp

Se

Pr

Fβ

5-fold

0.8602

0.9836

0.7989

0.8595

0.8109

10-fold

0.8387

0.9823

0.8122

0.8331

0.8010

5-fold

0.7387

0.9702

0.6701

0.6920

0.6778

10-fold

0.6774

0.9640

0.5992

0.5377

0.5471

Acc: accuracy. Sp = specificity. Se = sensitivity. Pr = precision. F β = F1 -score

the F1 -score metric by the 2-fold, 5-fold, and 10-fold cross-validation, respectively. In addition, Fig. 7 shows the confusion matrices showing the number of correct and incorrect prediction using the different cross-validation strategies. In sequence, from left to right: confusion matrix for the 2-fold cross validation, for the 5-fold cross validation, and for the 10-fold cross validation. The number of misclassified samples on the test set reduces when the number of k folds increases in the cross-validation approach. For instance, there are 18 and 13 instances incorrectly classified for the 2-folds and for the 5 folds cross validation. These results demonstrate that the proposed deep learning approach can obtain good performance for classifying the activities. Table 6. Classification performances comparison of the proposed spectrogram-based deep learning classification algorithm by different cross-validation strategies. K-fold

Metrics performance Acc

Sp

Se

Pr

Fβ

2-fold

0.8064

0.9781

0.7079

0.7526

0.7075

5-fold

0.8602

0.9836

0.7989

0.8595

0.8109

10-fold

0.8387

0.9823

0.8122

0.8331

0.8010

Acc: accuracy. Sp = specificity. Se = sensitivity. Pr = precision. F β = F1 -score

Confusion matrices can help ascertain which activities are more likely to be misunderstood. Possible classification issues can be seen by examining the waveforms, as various activities - such as walking down the stairs (class 4) and walking up the stairs (class 5) - show similar waveforms (see Fig. 2). Also the spectrogram visual representation (see Fig. 4) can support obtained results. It can be seen from the above illustrations how accelerometer spectra are mostly shifted toward lower frequencies during all the activities. Some of the activities differ only for different variations of the movement, such as different speeds. For example, walking activities (class 1) can be misclassified with running activities (class 2).

Spectrogram-Based Approach with Convolutional Neural Network

399

Fig. 7. Confusion matrix using the proposed spectrogram-based deep learning classification algorithm implementing different cross-validation strategies. In sequence, from left to right: confusion matrix for the 2-fold cross validation, for the 5-fold cross validation, and for the 10-fold cross validation.

Examining the confusion matrices, there is always a misclassified sample between class 4 (walking down the stairs) and class 5 (walking up the stairs) with all the three cross validation strategies implemented. In addition to having very similar waveforms between them, this misclassification error could be related to a smaller number of samples associated with these two classes in the dataset (2.7778%). It is also possible to notice that the label of class 7 is incorrectly assigned to samples belonging to other classes (class 3, class 6, class 8, class 9). Potential source of this error could be related to the small differences between these tasks: these are mainly position changes or rapid movements. Moreover, these tasks have a lower average duration and therefore a lower intensity energy. On the opposite side, higher classification performances were obtained with activities of class 1, class 2, class 3, and class 10. These are the classes that have the largest percentage of samples in the entire dataset (see Table 1). Moreover, these activities are the ones that have a longer average duration and consequently a greater intensity energy. As mentioned before, misclassification issues between samples belonging to class 1 and to class 2 can be due the fact that they are very similar activities, differing mainly in the speed at which they are performed. It must also be considered that these activities were performed by different subjects, so each activity and the speed of execution are subject-dependent. Based on all the results shared above, the proposed DP algorithm can classify human activity recognition with good average values of performances. Even though encouraging results were obtained, the analysis should be performed on a larger dataset for achieve higher classification performances.

4 Conclusions Human activity recognition (HAR) is in the searchlight in many fields. For example, the automatic recognition of activities of daily living (ADL) is particularly crucial to assess the functional ability of an elderly person to live independently in everyday living. This study proposed a deep learning approach to classify ten classes of different human activities of daily living and different simulated falls. Our spectrogram driven deep learning model significantly outperformed deep learning model developed on 1Dtime series signal for classification of different human activity types and fall detection.

400

M. Sassi et al.

Besides, the model developed on spectrograms of triaxial accelerometer signals significantly outperformed the model developed on amplitude which shows importance of training time-frequency features across different axis. This study can lead to detection of different activity types and falls which can be highly beneficial for different health management, elderly care and holistic wellbeing. Future studies suggest the deployment of temporal context training for prediction under real-time conditions.

References 1. Erda¸s, Ç.B., Atasoy, I., Açıcı, K., O˘gul, H.: Integrating features for accelerometer-based activity recognition. Procedia Comput. Sci. 98, 522–527 (2016) 2. Ramamurthy, S.R., Roy, N.: Recent trends in machine learning for human activity recognition—a survey (2018) 3. Ferrari, A., Micucci, D., Mobilio, M., Napoletano, P.: On the personalization of classification models for human activity recognition. IEEE Access 8, 32066–32079 (2020) 4. Jurˇci´c, K., Magjarevi´c, R.: Physical activity recognition based on machine learning. Budapest University of Technology and Economics (2022) 5. Luo, F., Poslad, S., Bodanese, E.: Kitchen activity detection for healthcare using a lowpower radar-enabled sensor network. In: ICC 2019 - 2019 IEEE International Conference on Communications (ICC), 20–24 May 2019, pp. 1–7 (2019) 6. Gholamrezaii, M., AlModarresi, S.M.T.: A time-efficient convolutional neural network model in human activity recognition. Multimed. Tools Appl. 80(13), 19361–19376 (2021) 7. Vanus, J., et al.: Monitoring of the daily living activities in smart home care. HCIS 7(1), 30 (2017) 8. Urwyler, P., et al.: Recognition of activities of daily living in healthy subjects using two ad-hoc classifiers. Biomed. Eng. Online 14(1), 54 (2015) 9. Zhu, J., Chen, H., Ye, W.: Classification of human activities based on radar signals using 1DCNN and LSTM. In: 2020 IEEE International Symposium on Circuits and Systems (ISCAS), 12–14 Oct 2020, pp. 1–5 (2020) 10. Islam, M.M., Nooruddin, S., Karray, F., Muhammad, G.: Human activity recognition using tools of convolutional neural networks: a state of the art review, data sets, challenges, and future prospects. Comput. Biol. Med. 149, 106060 (2022) 11. Wang, J., Chen, Y., Hao, S., Peng, X., Hu, L.: Deep learning for sensor-based activity recognition: a survey. Pattern Recogn. Lett. 119, 3–11 (2019) 12. Gupta, N., Gupta, S.K., Pathak, R.K., Jain, V., Rashidi, P., Suri, J.S.: Human activity recognition in artificial intelligence framework: a narrative review. Artif. Intell. Rev. 55(6), 4755–4808 (2022) 13. Huang, J., Chen, B., Yao, B., He, W.: ECG arrhythmia classification using STFT-based spectrogram and convolutional neural network. IEEE Access 1 (2019) 14. Alzubaidi, L.A.-O., et al.: Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. 2196-1115 (Print) 15. Ignatov, A.: Real-time human activity recognition from accelerometer data using convolutional neural networks. Appl. Soft Comput. 62, 915–922 (2018) 16. Panwar, M., et al.: CNN based approach for activity recognition using a wrist-worn accelerometer. In: 2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 11–15 July 2017, pp. 2438–2441 (2017) 17. Muralidharan, K., Ramesh, A., Rithvik, G., Prem, S., Reghunaath, A.A., Gopinath, D.M.P.: 1D convolution approach to human activity recognition using sensor data and comparison with machine learning algorithms. Int. J. Cogn. Comput. Eng. 2, 130–143 (2021)

Spectrogram-Based Approach with Convolutional Neural Network

401

18. Haya, A.: Deep learning-based model architecture for time-frequency images analysis. Int. J. Adv. Comput. Sci. Appl. 9 (2018) 19. Erda¸s, Ç.B., Güney, S.: Human activity recognition by using different deep learning approaches for wearable sensors. Neural Process. Lett. 53(3), 1795–1809 (2021) 20. Nguyen, M.D., Mun, K.R., Jung, D., Han, J., Park, M., Kim, J.: IMU-based spectrogram approach with deep convolutional neural networks for gait classification. In: 2020 IEEE International Conference on Consumer Electronics (ICCE), 4–6 Jan 2020, pp. 1–6 (2020) 21. Haleem, M.S., et al.: Time adaptive ECG driven cardiovascular disease detector. Biomed. Signal Process. Control 70, 102968 (2021) 22. Abu Alsheikh, M., Selim, A., Niyato, D., Doyle, L., Lin, S., Tan, H.P.: Deep activity recognition models with triaxial accelerometers (2015) 23. Hsu, Y., Chang, H., Chiu, Y.: Wearable sport activity classification based on deep convolutional neural network. IEEE Access 7, 170199–170212 (2019) 24. Hussain, G., Maheshwari, M.K., Memon, M.L., Jabbar, M.S., Javed, K.: A CNN based automated activity and food recognition using wearable sensor for preventive healthcare. Electronics 8 (2019) 25. Kingma, D., Ba, J.: Adam: a method for stochastic optimization. In: International conference on learning representations (2014)

Biosensors and Bioinstrumentation

Effect of Cardiolite on Enzyme Activity Ajla Selimovi´c1(B) , Safija Herenda1 , Edhem Haskovi´c2 , Mirjana Ðermanovi´c3 , and Emina Opankovi´c4 1 Department of Chemistry, Faculty of Science, University of Sarajevo, Zmaja od Bosne 33-35,

71000 Sarajevo, Bosnia and Herzegovina [email protected], [email protected] 2 Department of Biology, Faculty of Science, University of Sarajevo, Zmaja od Bosne 33-35, 71000 Sarajevo, Bosnia and Herzegovina 3 Public Health Institute of the Republic of Srpska, Jovana Duˇci´ca 1, 78000 Banja Luka, Bosnia and Herzegovina 4 Clinical Center, University of Sarajevo, Clinic for Nuclear Medicine and Endocrinology, Bolniˇcka 25, 71000 Sarajevo, Bosnia and Herzegovina

Abstract. Modern electrochemical research provides a good model for investigating reaction mechanisms and for characterizing enzymatic reactions in biological systems. The objective of this study is to determine the impact of a specific cardiolite on the activity of the catalase enzyme. Cardiolite (Tc-99m sestamibi) is a radiopharmaceutical diagnostic tool, which is used for diagnostic imaging in nuclear medicine, i.e. for the detection of coronary artery disease and assessment of myocardial function. Cardiolite accumulates in the heart muscle in proportion to the blood flow delivered by the coronary arteries. Additionally, it can be used to evaluate pathologies in breast and parathyroid gland tissues. The immobilization of Cardiolite was performed on a glassy carbon (GC) electrode. The immobilization process involved attaching the enzyme to the carrier to reduce its mobility and create a biosensor. All experiments were conducted using a three-electrode system employing cyclic voltammetry and chronoamperometry techniques. The results obtained were presented in terms of kinetic parameters. Based on these results, the Michaelis-Menten constant and the maximum reaction rate were calculated. The analysis revealed that the cardiolite exhibits a competitive type of inhibition on catalase activity. Keywords: Enzymes · Immobilization · Chronoamperometry · Myocardium

1 Introduction Cardiolites are a group of radioactive markers (tracers) used in scintigraphic methods to assess blood flow to the heart muscle as well as to evaluate its strength [1, 2]. They are deposited in the heart muscle in accordance with the amount of blood flow delivered by the coronary arteries. Technetium (99mTc) sestamibi is large synthetic molecule, specifically a coordination complex consisting of the radioisotope 99mTc bound to six ligands, namely methoxyisobutylnitrile (MIBI) molecules, from which the name “sestamibi” is © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 405–413, 2024. https://doi.org/10.1007/978-3-031-49068-2_41

406

A. Selimovi´c et al.

derived. The molecular formula is je C36H66N6O6Tc [3]. Technetium (99mTc) sestamibi is a radiopharmaceutical diagnostic tool and it is used for imaging in nuclear medicine for to detect and assess myocardial function [4]. It is also in use for assessment pathologies in the tissues of the heart, breast and parathyroid glands and it is part of the radioactive diagnostic class of drugs characterized as a lipophilic cationic radiotracer [5, 6]. For diagnostic use, the medicine is administered by intravenous injection after reconstitution with sterile, non-pyrogenic, oxidant-free Sodium Pertechnetate Tc99m Injection [7]. After intravenous injection of the drug, technetium 99mTc sestamibi is taken by the myocardium, parathyroid gland and/or breast tissue [8]. Once distributed, radiotracers emit photons that can be captured with for imaging with SPECT or PET [9]. Technetium-99m is preferred for use due to its short half-life of six hours and its ability to distribute beyond the primary organ, resulting in lower radiation exposure to the patient compared to other radionuclides [8]. This radiopharmaceutical tool, alone od in combination with 99mTcO4 − is currently used to study the parathyroid glands. During the scintigraphy, injection of 99mTc provides the same functional information as 99mTcO4 −. Images are generally obtained after 10–30 min (early scanning) and 60–120 min (delayed scanning) [10]. In the experimental section of this paper, immobilization was carried out by the trapping method and nafion was used as a carrier. Cyclic voltammetry and chronoamperometry were employed to monitor and characterize the enzymatic reaction.

2 Materials and Methods Laboratory equipment: potentiostat/galvanostat Vertex one, Ivium Technologies; pH meter - PHYWE, Germany RS 232, analytical balance METLLER, technical balance KERN 440-47. Chemicals: Disodium phosphate (Na2 HPO4 ), Fischer Chemical, UK; Potassium dihydrogen phosphate (KH2 PO4 ), Semikem d.o.o., BiH; Hydrogen peroxide (H2 O2 ) p.a. 30%, Sigma-Aldrich Chemie GmbH, Germany; Catalase from bovine liver 2000– 5000 units/mg proteins Sigma-Aldrich Chemie GmbH, USA; Nafion (5% solution in alcohol and water), Sigma-Aldrich Chemie GmbH, USA, Perfluorinated resin solution containing Nafion Sigma-Aldrich Chemie GmbH, USA, Cardiolite Mon.Mibi Kit, Eczacıba¸sı-Monrol, Turkey. The following solutions that were used to determine the effect of cardiolite on catalase activity: – Phosphate buffer solution: The phosphate buffer was prepared by dissolving 3.407 g of potassium dihydrogen phosphate (KH2 PO4 ) in distilled water in a 500 mL volumetric flask and 4,450 g of disodium hydrogen phosphate (Na2 HPO4 ), in distilled water in a volumetric flask of 500 mL. The pH value of 7.24 was adjusted by adding 100 mL of KH2 PO4 and 150 mL Na2 HPO4 into a 250 mL volumetric flask. – Hydrogen peroxide solution: Basic hydrogen peroxide solution (H2 O2 ) concentration 0.1 M was prepared by adding 9.8 mL of 30% H2 O2 to a 100 mL volumetric flask and diluting with distilled water. – Catalase solution: The basic catalase solution with a concentration of 70.1 mg/mL was prepared in centrifuge microtube by dissolving 0.0701 g of pure catalase in 1 mL of phosphate buffer.

Effect of Cardiolite on Enzyme Activity

407

– Cardiolite solution: A cardiolite basic solution with a concentration of 13.71 mg/mL was prepared by dissolving the contents of one bottle of cardiolite (containing 24,675 mg substance) in 1.8 mL of distilled water. From the basic solution of catalase, concentration 0.701 g/mL, a new solution was prepared in centrifuge microtube combining of 10 µL of catalase, 10 µL of Nafion (enzyme carrier) and 980 µL of phosphate buffer. The resulting enzyme concentration was 7.01 × 10−3 g/mL. After preparing the surface of the GC electrode and the enzyme solution, the enzyme was immobilized by applying 10 µL of the solution to the electrode and allowing it to dry at room temperature until the solvent completely evaporated, making it ready for measurement. Measurements were preformed using a potentiostat in a three-electrode system consisting of a working electrode (GC electrode), reference electrode (saturated Ag/AgCl - electrode) and a counter electrode (Pt - electrode) which were immersed in the glass cell containing the aqueous phosphate buffer solution. Hydrogen peroxide was used as the substrate during this test. The success of the immobilization and the activity of the enzyme were monitored electrochemically using cyclic voltammetry techniques and chronoamperometry.

3 Results and Discussion In this part presents graphics and explanations of the test of the influence of the substrate concentration on catalase activity, the thickness of the enzyme layer on the surface of the immobilized GC electrode and chronoamperometric test of the kinetic parameters of the enzyme’s activity.

Fig. 1. Cyclic voltammogram of the GC electrode (blue line) and cyclic voltammogram of the immobilized electrode (red line)

Figure 1 shows cyclic voltammogram of the GC electrode with and without the enzyme. The potential limits values were adjusted from −0.1 to 0.7 V, with a scanning speed of 25 mV/s. After scanning, a voltammogram (blue line) was obtained. Subsequently, scanning was performed under the same conditions with the immobilized enzyme on the GC electrode. The cyclic voltammogram of this scanning exhibited higher current values, providing evidence of a successful immobilization.

408

A. Selimovi´c et al.

Fig. 2. Cyclic voltammogram of 10 cycles of the immobilized GC electrode y = -0,0059x + 0,054 R² = 0,9161

0.08 I(mA)

0.06 0.04 0.02 0 0

2

4

6

8

10

12

number of cycles

Fig. 3. Diagram of the linear dependence of the current peak (0.5 V) on the number of cycles

In Fig. 2, scanning 10 cycles on potentiostat using cyclic voltammetry technique at a scanning speed of 25 mV/s is depicted. Deviations in the appearance of the curve can be observed for each cycle as confirmed by the diagram of the linear dependence presented in Fig. 3. The current values at a potential of 0.5 V are shown in the diagram. The deviations in the curves, with higher or lower values as the number of cycles increases, indicate an uneven distribution of the enzyme on the surface of the GC electrode. In our case, the enzyme layer is relatively well-distributed over the surface of the electrode. During the examination of the influence of substrate concentration on catalase activity, scans were conducted with five different concentrations of hydrogen peroxide: 0.332, 0.662, 0.990, 1.31 and 1.64 mM. Figure 4 we illustrates an increase in the current peak in the area of hydrogen peroxide reduction al the potential of 0.41 V with an increase of concentration of substrate in solution. Figure 5 demonstrates that there is a sequence of increasing currents as the substrate concentration increases, providing evidence that the enzyme-substrate complex was formed. After conducting cyclic voltammetry, chronoamperometric tests of enzyme kinetics were performed by scanning at the constant potential of 0.9 V on potentiostat, with the addition of 100 µL of substrate every 50 s (Fig. 6). The amperometric biosensor based on GC electrode with immobilized catalase in the film of Nafion film exhibits a specific response to the presence of the

Effect of Cardiolite on Enzyme Activity

409

I(mA)

Fig. 4. Cyclic voltammogram of the immobilized GC electrode at different concentrations of the substrate, hydrogen peroxide

3.50E-03 3.00E-03 2.50E-03 y = -677,92x - 2,3418 2.00E-03 R² = 0,7899 1.50E-03 1.00E-03 5.00E-04 0.00E+00 0.00E+00 2.00E-01 4.00E-01 6.00E-01 8.00E-01 1.00E+00 1.20E+00 1.40E+00 -3

cH2O2(mmoldm )

Fig. 5. Linear dependence of the current on the concentration of the substrate

substrate. The addition of hydrogen peroxide causes an increase in current, followed by a decrease and formation of a current peak. The reaction continues until the substrate binds to all active sites. As the reaction approaches its completion, a decrease in the intensity of the peaks occurs, indicating the formation of a saturated enzyme-substrate complex. Subsequent addition of substrate does not significantly alter the peaks’ intensity. Figure 7 presents the Lineweaver-Burk diagram, which shows the correct sequence of increasing currents with increasing concentration of hydrogen peroxide in the solution. Linearity is observed in the concentration range from 0.332 to 1.32 mM. The kinetic parameters determined in this experiment are the maximum velocity (Vmax ) and Michaelis-Menten constant (KM ), with values of Vmax = 3.80 × 10–3 mol/dm3 s and KM = 0.26 mM. The obtained value of the constant indicates that a higher concentration of the substrate is required to achieve the maximum reaction rate, and the immobilized enzyme becomes saturated. To test the influence of cardiolite concentration on enzyme activity, measurements were conducted with the presence of cardiolite at concentrations of 0.49 and 0.32 g/mL,

410

A. Selimovi´c et al.

1/I (mA-1)

Fig. 6. Amperometric response of the immobilized GC electrode (7.01 × 10–3 g/mL catalase) with the successive addition of the substrate, hydrogen peroxide (uninhibited reaction)

500 450 400 350 300 250 200 150 100 50 0

y = 5E-05x + 0,2647 R² = 0,9738

0

0.5

1

1.5 2 2.5 1/cH2O2 (mmo1-1dm3 )

3

3.5

Fig. 7. Lineweaver-Burk diagram for determining the maximum velocity and Michaelis-Menten constant for uninhibited reaction

while keeping a fixed concentration of immobilized catalase at 1.7 × 10–3 g/mL. Different substrate concentrations were added at intervals, with 100 µL of hydrogen peroxide added every 50 s. Chronoamperograms (Fig. 8) compare the amperometric response of immobilized catalase with and without the presence of cardiolite. This figure illustrates how increasing the concentration of cardiolite affects the current response with each addition of the substrate. After the amperometric measurement, a graph is plotted to show the reciprocal value of the current strength against the reciprocal value of the hydrogen peroxide substrate concentration. The diagram demonstrates a linear dependence of 1/I on 1/c(H2 O2 ) when different amounts of inhibitors are added. It can be concluded that larger amounts of inhibitors result in smaller current values. From the diagram in Fig. 9, it is possible to determine the type of inhibition. The segment on the ordinate does not change, but the slope of the line increases. The results indicate that the maximum reaction speed remains unchanged with the addition of the

Effect of Cardiolite on Enzyme Activity

411

Fig. 8. Diagrams of the amperometric response of the immobilized GC electrode with the successive addition of the substrate and cardiolite in concentration 0.49 g/mL (green line) and 0.32 g/mL (orange line)

inhibitor, but the inhibitor increases the KM values. This suggests that it is a competitive type of inhibition. The competitive mechanism occurs when the substrate and the inhibitor compete for the same active site on the enzyme. As a result of this competition, a large amount of substrate is replaced by the inhibitor when binding to the active site on the enzyme, which is why the maximum reaction speed remains unchanged and has a unique value of 0.0038 mol/dm3 s (Table 1).

A. Selimovi´c et al.

1/I,mA–1

412

y = 235,74x + 261,66 R² = 0,7728 y = 134,38x + 260,72 R² = 0,8266 y = 66,431x + 257,53 R² = 0,9625

1/cH2O2, mmol-1dm3

Fig. 9. Lineweaver-Burk diagram for determining maximum velocity and Michaelis-Menten constant with and without the presence of different concentration of cardiolite. Grey line—0 g/mL; orange line—0.32 g/mL; red line—0.49 g/mL.

Table 1. KM and Vmax values without and with the presence of different concentrations of cardiolite C of cardiolite (g/mL)

Vmax (mol/dm3 s)

KM (mM)

0

0.0038

0.26

0.32

0.0038

0.45

0.49

0.0038

0.52

4 Conclusion In this paper, the effect of cardiolite on enzyme activity was investigated using electrochemical methods to determine the kinetics of enzymatic reactions. Electrochemical methods enable the monitoring of fast reactions, such as electron transfer between enzymes and substrates in enzyme-catalyzed reactions, and the characterization of their mechanisms. Based on the results obtained from the experimental part using cyclic voltammetry and chronoamperometry methods, the following conclusions can be drawn: 1. Increasing the substrate concentration leads to a linear increase in the current in the current in the area of hydrogen peroxide reduction, which indicating the formation of an enzyme-substrate complex (ES). 2. After recording 10 cycles of the immobilized GC electrode, data on the thickness of the enzyme layer on the electrode surface were obtained. Analysis of the current values in the anodic oxidation region revealed a relatively even distribution of the enzyme layer on the surface of the GC electrode.

Effect of Cardiolite on Enzyme Activity

413

3. Analysis of the data obtained from chronoamperometry confirmed that catalase follows the Michaelis-Menten kinetic model. 4. The addition of larger amounts of inhibitors resulted in smaller current values and a slowdown of the reaction, as observed from the linear dependence diagram 1/I = f(1/c(H2 O2 )). 5. Lineweaver-Burk diagram showed that the curves intersect at the ordinate, indicating that cardiolite exhibits a competitive inhibitory effect on catalase. 6. The KM values obtained indicate that the Michaelis-Menten constant increases with the addition of cardiolite, while the maximum reaction rate remains the same.

References 1. Schenke, S.A., Görges, R., Seifert, P., Zimny, M., Kreissl, M.C.: Update on diagnosis and treatment of hyperthyroidism: ultrasonography and functional imaging. Q. J. Nucl. Med. Mol. Imaging 65(2), 102–112 (2021). https://doi.org/10.23736/S1824-4785.21.03333-1 2. Mochula, A.V., Zavadovsky, K.V., Lishmanov, Y.B.: Method for studying the myocardial blood flow reserve by load dynamic single-photon emission computed tomography. Bull. Exp. Biol. Med. 160(6), 864–866 (2016). https://doi.org/10.1007/s10517-016-3328-z 3. Compound, N.C.f.B.I.P. and T.m.s.R.F.: Summary for CID 22617237. https://pubchem.ncbi. nlm.nih.gov/compound/Technetium-_99mTc_-sestamibi (2023) 4. Wishart, D.S., et al.: DrugBank 5.0: a major update to the DrugBank database for 2018. Nucl. Acids Res 46(D1), D1074–D1082 (2018) 5. Chesser, A.M., et al.: Technetium-99m methoxy isobutyl isonitrile (MIBI) imaging of the parathyroid glands in patients with renal failure. Nephrol. Dial. Transplant. 12(1), 97–100 (1997) 6. Delmon-Moingeon, L.I., et al.: Uptake of the cation hexakis(2-methoxyisobutyli-sonitrile)technetium-99m by human carcinoma cell lines in vitro. Cancer Res. 50(7), 2198–2202 (1990) 7. Cardiolite: J.D.c.h.w.d.c.p.c.h (2022) 8. Green, C.H.: Technetium-99m production issues in the United Kingdom. J. Med. Phys. 37(2), 66–71 (2012) 9. Lee, W.W.: Clinical applications of technetium-99m quantitative single-photon emission computed tomography/computed tomography. Nucl. Med. Mol. Imaging 53(3), 172–181 (2019). https://doi.org/10.1007/s13139-019-00588-9 10. Signore, A.a.C.L.: Gamma camera imaging of benign thyroid diseases. In: Signore, A. (ed.) Nuclear Medicine and Molecular Imaging, pp. 45–58. Elsevier, Oxford (2022)

Evaluation of Printed Coplanar Capacitive Sensors for Reliable Quantification of Fluids in Adult Diaper Muhammad Tanweer1(B) , Liam Gillan2 , Raimo Sepponen3 , Ihsan Oguz Tanzer1 , and Kari A. Halonen1 1 Department of Electronics and Nanoengineering, Aalto University, 02150 Espoo, Finland

[email protected]

2 VTT Technical Research Centre of Finland Ltd., 02150 Espoo, Finland 3 Department of Automation and Electrical Engineering, Aalto University, 02150 Espoo,

Finland

Abstract. Advancements in printed technology have led to the development of economical and sustainable electronic solutions for wearable medical devices in the healthcare sector. Printed capacitive sensors in planar geometry are widely used in the development of smart diapers for detecting urination events, quantifying detection, and quantifying voided volumes. However, factors such as the effect of sodium electrolyte variation, body weight effect on a wet diaper and gravitational effect on wet diapers impair the quantification of voided volume with capacitive sensors. In this study, a printed capacitive sensor for quantifying human body fluids in adult diaper was evaluated to analyze these effects. Silver and carbon inks were used to print the parallel-plate capacitive electrodes on a flexible substrate in a coplanar geometry. In-diaper quantification measurements were performed at various concentration levels in pseudo urine with small incremental levels at the adult human urination flow rate. The impact of human body weight on quantification measurements using a wet diaper was studied. The gravitational pull effect of wetness was evaluated for on-human-torso use in both standing and lying positions. It was observed that a printed coplanar capacitive sensor alone is insufficient to reliably quantify the voided volume in diapers. Keywords: Capacitive sensors · Sodium concentration · Printed sensors · Printed electronics · Voided volume quantification · Smart diapers · Wearable medical devices

1 Introduction The population is aging globally, and the aging level is on the rise, especially in eastern European and Asian regions, where the share of aged people is larger with lower life expectancy due to dominant health issues [1]. Urinary incontinence – the lack of voluntary control of the bladder – is prevalent in older people with an overactive bladder, causing embarrassment [2, 3]. Disposable diapers © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 414–422, 2024. https://doi.org/10.1007/978-3-031-49068-2_42

Evaluation of Printed Coplanar Capacitive Sensors

415

are one of the most common products used by patients with a loss of bladder control. The global market for absorbent hygiene products such as adult diapers was USD 13.4 billion in 2020, and only USD 0.3 billion in 2016. It is expected to grow to USD 22.7 billion between 2021 and 2028 with a CAGR of 4.5% [4]. A wet diaper, if not replaced at appropriate intervals, may cause infections such as decubitus ulcers, incontinence-associated dermatitis, and lower urinary tract infections in geriatric patients [5]. Care providers continuously check the level of wetness every hour to inspect the need for a diaper change, which not only leads to discomfort for the care receiver but also makes the care provider’s job labor-intensive. Moreover, care providers are instructed to fill in urination tracking events and voided volumes in adult diapers of geriatric patients to improve their quality of life. The conventional method of measuring the voided volume and flowrate is by manually weighing diapers or using a container with a stopwatch to maintain the frequency volume charts and paper diaries of patients facing urinary incontinence. This adds an extra burden to healthcare staff and many discrepancies have been noticed in such measurements [6, 7]. Recent advancements in the internet-of-things (IoT) have led to the development of state-of-the-art wearable biomedical solutions for body area sensor networks by enabling the use of healthcare devices not only by healthy individuals but also by the care receiving vulnerable members of society to measure and record their daily routine activity events. The healthcare parameters are continuously ambulatory, measured at sensor nodes, monitored on smart devices, and recorded on the cloud for further investigation of chronic diseases and health performance using deep learning, big data, and artificial intelligence (AI) based algorithms to enhance digital healthcare [8]. The development of smart diapers using various measurement techniques under the umbrella of the Internet-of-Things has brought solutions to several challenges faced by not only absorbent hygiene product users of early age but also the elderly and patients with overactive bladders and lower urinary tract symptoms (LUTS). In literature, various sensors have been developed to detect and quantify the voided volume of body fluids in diapers to maintain the digital voiding diary of geriatric patients. For example, a study in [9] presented the use of a pair of ultra-high-frequency radio frequency identification (RFID) tags inside a diaper as moisture sensors to detect wetness in a diaper. Capacitive sensors with coplanar geometry have been widely adopted for urination event detection. A planar capacitive sensor development technique was studied in [10], where copper tape was used to implement a coplanar sensor with variation in the interelectrode distance. The sensor is deployed on the outer surface of the diaper, and discrete components are used to implement the front-end electronic system for data acquisition. Water volume measurements were performed using a sensor on the torso in the supine and lateral positions. Another study [11] presented the development of a portable device to measure the voided volume using coplanar capacitive sensors. A set of three coppertape-based capacitive sensors was deployed on the outer surface of the urine collection container. The study in [12] presented the design of a low-cost wireless incontinence sensing system with an array of six capacitive and resistive sensors in planar geometry using conducting polymer in silicon with limited of flexibility and stretchability. The sensor was implemented by sandwiching the absorbing pads of the diaper between several interdigitated electrodes.

416

M. Tanweer et al.

Recent developments in printed electronics technology have made it possible to produce sensors and electronic components on flexible, economical, and sustainable substrates [13]. A coplanar capacitive sensor was implemented using various conductive materials to collect the specifications for developing printed capacitive sensors using economical and sustainable materials for human fluid detection and voided volume quantification in adult diapers [14]. To the best of our knowledge, previous works are missing the effect of sodium electrolyte concentration variations in voided volume on the coplanar capacitive sensor model. Secondly, the wet diaper faces a gravitational pull which affects the capacitive sensor measurements. Finally, the wet diaper encounters weight stress owing to body movement, which causes a change in the capacitive sensor readings. All these factors may lead to false quantification of the voided volume if only a capacitive sensor is used for the measurements.

Fig. 1. Voided volume measurement setup with the impedance analyzer using the printed coplanar capacitive sensor.

This study focuses on sensor design and development in biomedical engineering for IoT-based wearable medical devices where screen-printed coplanar capacitive sensors from silver and carbon inks are evaluated for reliable quantification of voided volume in adult diapers. The capacitive sensors of the two electrodes in coplanar geometry were developed based on the specifications provided by [14] to evaluate the aforementioned effects of electrolyte variation, gravity, and weight stress on capacitive sensor measurements while quantifying voided volume in an adult diaper. Figure 1 shows the measurement setup used in this study. The voided volume was imitated using tap water and pseudo-urine with various concentrations of sodium chloride electrolyte. An interesting

Evaluation of Printed Coplanar Capacitive Sensors

417

behavior of the printed capacitive sensors to the variation in electrolyte concentration was observed. Experiments revealed that the precise quantification of voided volume in adult diapers is affected by the variation in the concentration of electrolytes, and measurement with capacitive sensors is insufficient to produce reliable results. The materials and methods used in this study are discussed in detail in Sect. 2. The results are elaborated in Sect. 3, and the conclusion and future recommendations are provided in Sect. 4. A list of all the abbreviations used in this article is provided in Appendix I.

2 Materials and Methods Printed capacitive sensors depicted in Fig. 2 were designed with two electrodes in parallel coplanar geometry with dimensions of 10 mm x 200 mm and a 3 mm interelectrode separation. The sensors were fabricated on polyethylene terephthalate (PET, Dupont Teijin Films, Melinex ST506, 125 µm thick) substrate using a flatbed screen printer (EKRA, E2) to pattern silver ink (Asahi LS-411AW, oven-dried at 120 °C for 20 min) or carbon ink (Sun C2171023D1, oven-dried at 130 °C for 30 min), before encapsulation with insulator ink (Loctite EDAG PF 455B E&C, two printed layers, ultraviolet (UV) light cured using a Fusion UV light systems D-bulb for 2 min per layer). Electrically conductive adhesives (Chemtronics CW2400, oven-dried at 90 °C for 5 min) were used to make the connections.

Fig. 2. Screen-printed capacitive sensors in a coplanar geometry using silver and carbon inks.

Sodium electrolyte concentration varies in urine from 30 to 140 mmol/l for children below the age of 16 years and from 80 to 240 mmol/l for adults, according to the Laboratory of Helsinki University Hospital. Pseudo urine is prepared for this study with different concentrations of electrolytes to analyze the effect on the quantification

418

M. Tanweer et al.

measurements with printed capacitive sensors in coplanar geometry. Sodium chloride solution was prepared as pseudo-urine using table salt and tap water. Solution with concentration of 0, 10, 30, 60, and 90 mmol/l were used to evaluate the behavior of the printed coplanar capacitive sensors inside the adult diapers. Capacitance measurements of the printed sensors were performed using an impedance analyzer E4990A (Keysight Technologies Inc. with fixture HP16664A), and the recorded data were processed to compile the results using MATLAB software. Commercially available adult diapers by Caroli (W. Pelz GmbH & Co. KG) were used to conduct in-diaper sensor measurements in this study.

Fig. 3. The figure depicts the weight load on the entire diaper and partial diaper for voided volume quantification on the test bench.

The experimental setup involved three scenarios of measurements using the impedance analyzer: (i) sodium electrolyte concentration effect, (ii) weight effect on wet diaper, and (iii) gravitational effect on measurement using a humanoid torso. In the first scenario, the effect of sodium electrolyte concentration on the printed silver and carbon sensors was measured using diaper pads on the test bench. Pseudo-urine is poured into the diaper with an incremental volume of 50 ml for each measurement. All measurements were recorded with a delay of 30 s after the introduction of the liquid to the diaper to ensure stable readings. Pseudo urine (300 ml) was used as the total voided volume to measure each concentration level of 0, 10, 30, 60, and 90 mmol/l.

Fig. 4. Measurement setup to analyze the gravitational pull effect with on-torso diapers in both standing and lying positions.

Evaluation of Printed Coplanar Capacitive Sensors

419

The second scenario assessed the effect of weight on a wet diaper. An incremental weight from 1 kg to 10 kg was applied to the entire pad and partial section diaper to analyze the impact on the capacitive sensor measurements. Figure 3a depicts the measurement setup with a weight load on the entire pad, and Fig. 3b shows the partially loaded diaper pad. In the third scenario the gravitational impact of a wet diaper on a capacitive sensor was measured using a human torso in both the standing and lying positions. The measurements were recorded at 1, 5, and 10 min after introducing voided volume to the diaper worn by a human torso. Figure 4 shows the measurement setup of the torso in the standing and lying positions for measuring the gravitational effect of the wet diaper position on the capacitive sensors.

3 Results The variation in the electrolyte concentration of human fluid affects the sensing capabilities of coplanar capacitive sensors, leading to incorrect measurements of the voided volume. The silver- and carbon-printed sensors show incremental capacitive behavior with the addition of every 100 ml fluid introduction to the diaper for all sodium concentration levels in pseudo urine. However, readings at the same volume with different concentrations present meaningful changes in the capacitance for both the carbon and silver sensors. The capacitance reading of the silver sensor is twice for 100 ml fluid when the concentration increases from 0 mmol/l to 10 mmol/l, and it is four times higher with concentration levels of 60 mmol/l and 90 mmol/l. Figure 5 shows the measurement of the behavior of silver and carbon capacitive sensors for the concentration change of sodium electrolytes in pseudo-urine. The change in the sensor capacitance is even higher for higher voided volumes when the concentration is greater.

Fig. 5. The figure depicts the effect of electrolyte concentration on the capacitance measurement to quantify the voided volume with coplanar sensors printed with silver and carbon ink.

420

M. Tanweer et al.

Figure 5 also shows the capacitive sensor behavior over the frequency spectrum of 100 kHz to 1 MHz. The higher sheet resistance of the carbon ink impairs capacitive measurement at high frequencies. However, at low frequencies, it measures the capacitance close to the silver ink sensor measurements, making it comparatively economical, sustainable, and a suitable candidate for printed capacitive sensors. Capacitive sensor measurements in wet diapers are also prone to body movements when the sensor faces parts of the diaper squeezed owing to partial weight loading. Figure 6 shows the behavior of the capacitive sensor in a wet diaper when an incremental weight is applied on part of the diaper or on the entire diaper. It was found that the capacitance of the sensors increases when part of the diaper or the entire diaper is under a weight load, and it increases with the increase in weight either partially or on the whole diaper. It was observed that the capacitance increase was threefold with a just 1 kg point load on a wet diaper. This increase is significant enough to lead to false quantification of the voided volume.

Fig. 6. Weight loading effect on wet diaper measurements with load implementation on partial and entire diapers.

Finally, the gravitational pull on wet diaper is also a factor that impairs accurate quantification measurements using a capacitive sensor alone. In this study, an amount of 100 ml liquid was introduced into diapers on the human torso in both standing and lying positions. Figure 7 depicts the gravitational pull effect on a wet diaper with 100 ml liquid over a time span of 10 min. The effect of the gravitational pull was measured after 1, 5, and 10 min. In the first 5 min, there was a significant drop in the measured capacitance, which stabilized after 10 min. Gravitational pull has a greater effect in the standing position than lying. After 10 min another 100 ml liquid was introduced to the diaper, and the change was significantly detected in both positions hence making coplanar capacitive sensor a suitable candidate for intermittent urination events detection.

Evaluation of Printed Coplanar Capacitive Sensors

421

Fig. 7. The figure presents the gravitational pull of wet diaper on the measurements of capacitive sensors in both standing and lying positions.

4 Conclusion The development of smart diapers with urination detection solutions can make the work of care providers easier; however, the precise quantification of human fluids in diapers still poses challenges. Capacitive sensors can indicate the wetness of a diaper and detect intermittent urination events. However, the results of this study show that the electrolyte concentration in urine significantly affects the readings of the coplanar capacitive sensors. In addition, the gravitation pull and body weight over capacitive sensors introduce artifacts into the measurements. Therefore, it is observed that the capacitance does not provide reliable information for the quantification of urine. In future work, more accurate quantification of fluids might be assisted by the incorporation of additional sensors to enable correction for variations in electrolyte concentration, changing gravitation pull, and applied pressure. A multi-sensor node using sustainable printed electronics technology combined with AI-based data processing algorithms is proposed for the future developments. Acknowledgement. Technical assistance towards the printed sensors from Olli Halonen, Kim Eiroma, Mika Suhonen, Tapio Mäkelä, and Asta Pesonen is gratefully acknowledged. The research work is executed under the Urisens project funded by Business Finland (grant agreement number 1696/31/2022) and the EHIR project funded by the Academy of Finland.

Appendix – I List of Abbreviations AI: Artificial Intelligence CAGR: Compound Annual Growth Rate IOT: Internet of Things

422

M. Tanweer et al.

LUTS: Lower Urinary Tract Symptoms RFID: Radio Frequency Identification USD: United States Dollar

References 1. Balachandran, A.: Population ageing in Europe and Asia: Beyond traditional perspectives (2020). https://doi.org/10.33612/diss.135497884 2. National Institute on aging. Urinary Incontinence in Older Adults. https://www.nia.nih.gov/ health/urinary-incontinence-older-adults. Accessed 15 Feb 2023 3. Batmani, S., Jalali, R., Mohammadi, M., et al.: Prevalence and factors related to urinary incontinence in older adult women worldwide: a comprehensive systematic review and metaanalysis of observational studies. BMC Geriatr. 21, 212 (2021). https://doi.org/10.1186/s12 877-021-02135-8 4. Adult Diapers Market Size, Share, Growth Report 2030. Zion Market Research. https://www. zionmarketresearch.com/report/adult-diapers-market. Accessed 15 Feb 2023 5. Bender, J.K., Faergemann, J., Sköld, M.: Skin health connected to the use of absorbent hygiene products: a review. Dermatol Ther (Heidelb). 7(3), 319–330 (2017). https://doi.org/10.1007/ s13555-017-0189-y. Epub 2017 Jun 30. PMID: 28667496; PMCID: PMC5574741 6. Mehta, S., Geng, B., Xu, X., Harmanli, O.: Current state of bladder diary: a survey and review of the literature. Int Urogynecol J. (2022). https://doi.org/10.1007/s00192-022-053 98-w. Epub ahead of print. PMID: 36322174 7. Stone, A.A., Shiffman, S., Schwartz, J.E., Broderick, J.E., Hufford, M.R.: Patient noncompliance with paper diaries. BMJ 324(7347), 1193–1194 (2002 May 18). https://doi.org/10. 1136/bmj.324.7347.1193.PMID:12016186;PMCID:PMC111114.,lastaccessed2016/11/21 8. M. Tanweer and K. A. I. Halonen, “Development of wearable hardware platform to measure the ECG and EMG with IMU to detect motion artifacts,” 2019 IEEE 22nd International Symposium on Design and Diagnostics of Electronic Circuits & Systems (DDECS), ClujNapoca, Romania, 2019, pp. 1-4https://doi.org/10.1109/DDECS.2019.8724639 9. Tajin, M.A.S., Mongan, W.M., Dandekar, K.R.: Passive RFID-based diaper moisture sensor. IEEE Sensors J. 21(2), 1665–1674 (2021). https://doi.org/10.1109/JSEN.2020.3021395 10. Konno, S., Kim, J., Nakajima, K.: Development of capacitive sensor for diaper absorption volume. Adv. Biomed. Eng. 9, 106–111 (2020). Released on J-STAGE May 23, 2020, Online ISSN 2187–5219, https://doi.org/10.14326/abe.9.106 11. Nikoli´c, G., Stožer, A., Kramberger, I.: Development of a portable device for urodynamic data acquisition suitable for home use. In: Badnjevic, A., Gurbeta Pokvi´c, L. (eds.) CMBEBIH 2021. CMBEBIH 2021. IFMBE Proceedings, vol. 84. Springer, Cham (2021). https://doi.org/ 10.1007/978-3-030-73909-6_10 12. Ngo, H.-D., et al.: A novel low-cost wireless incontinence sensor system (screen-printed flexible sensor system) for wireless urine detection in incontinence materials. Proceedings 2(13), 716 (2018). https://doi.org/10.3390/proceedings2130716 13. Hakola, L., Immonen, K., Sokka, L., V¨alim¨aki, M., Smolander, M., Mantysalo, M., Tanninen, P., Lyytik¨ainen, J., Leminen, V., Naji Nassajfar, M., Horttanainen, M. Venetjoki, P.: ‘Sustainable materials and processes for electronics, photonics, and diagnostics. Proceedings of the Electronics Goes Green 2020+, pp. 45–52 (2020) 14. Tanweer, M., Sepponen, R., Tanzer, I.O., Halonen, K.A: Development of capacitive sensors to detect and quantify fluids in the adult diaper. In: Chen, Y., Yao, D., Nakano, T. (eds.) Bio-inspired Information and Communications Technologies, pp. 237–245. Cham, Springer Nature Switzerland (2023). https://doi.org/10.1007/978-3-031-43135-7_23

Biosensor for Rapid Methods for the Detection of Viruses Sara Deumi´c1(B)

, Aida Lavi´c1 , Neira Crnˇcevi´c1 and Amar Deumi´c2

, Emina Pramenkovi´c1

,

1 International Burch University, Francuske Revolucije bb, 71 210 Ilidža, Bosnia and

Herzegovina {sara.deumic,neira.crncevic}@ibu.edu.ba, {aida.lavic, emina.pramenkovic}@stu.ibu.edu.ba 2 Medical Device Inspection Laboratory “Verlab”, Ferhadija 27, 71 000 Sarajevo, Bosnia and Herzegovina [email protected]

Abstract. Infectious diseases caused by viruses are a big problem and a threat to public health. Rapid diagnosis of these diseases is critical for effective clinical outcomes. In vitro diagnostics often require centralized laboratories and large and expensive devices. Recently, biosensors have been developed for the rapid detection of viral particles. This review compiles different types of biosensors and their principles. The focus has been placed on biosensors used for the detection of Influenza A and HIV. The method of detection, detection time, sensitivity, and accuracy of these biosensors were explained and used for the comparison. The properties of gold nanoparticles, DNA aptamers, and antibodies against viral proteins enabled them to be the most efficient materials for designing biosensors for Influenza A. Electrochemical, piezoelectric, mechanical, and optical sensors for the detection of HIV are described. Some of their advantages and disadvantages are mentioned. This review analyzed 26 papers which are dated from 2010 to 2020. A significant amount of these studies has been focused on the application of nanoparticles for improving the accuracy of sensing methods. Keywords: Biosensors · Rapid methods · Virus detection · HIV · Influenza

1 Introduction Viruses are obligate intracellular parasites, meaning that they require a living host cell in order to replicate [1]. They change and mutate very rapidly thus the same vaccines and antiviral drugs cannot be used over and over again. Viral infection by a new virus or a new strain of a known virus can cause a worldwide pandemic very quickly, due to its high mutation rate [2]. There is always an emerging need to develop and produce new vaccines each time a new virus or a new strain of a known virus emerges [1]. This is not as easy as it seems. To produce a new vaccine, they must first be detected and isolated. Knowing the structure and composition of the virus is crucial for its detection or any further action regarding the development of the corresponding vaccine. Precise and © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 423–429, 2024. https://doi.org/10.1007/978-3-031-49068-2_43

424

S. Deumi´c et al.

accurate detection is of significant importance. Early detection of viruses is necessary, but the methods used are not always fast and efficient [3]. In recent years, there has been an increased number of infectious diseases caused by viruses such as SARS-Cov, influenza A/H5N1, influenza A/H1N1, Dengue virus, HIV, and new encephalitis viruses. A quick and well-timed diagnosis of these diseases is of utmost importance. Several laboratory diagnostic methods are applied to confirm the identity of the pathogen: serology (immunofluorescence techniques, neutralization test, and enzyme-linked immunosorbent assay (ELISA)) [4], indirect or direct examination (inoculation, animal tests, electron microscopy, antigen detection, molecular techniques (PCR)) [5]. However, these diagnostic techniques require a pre-treated sample, biological products, standard biosafety laboratories, and time-consuming analyses to yield a reliable answer [6]. In recent decades, biosensors have seen a lot of use in rapid virus detection, due to their easy operation and transport [7]. They require no reagent and provide results in a matter of minutes. Biosensors are analytical devices that convert a biological response into an electrical signal. They are composed of a bioreceptor (recognition element), a transducer, and a signal processor. Based on the type of transducer, biosensors are classified into four main groups: optical, piezoelectric, electrochemical, and thermal [8]. The biosensor is defined as the medical device that measures biological processes’ changes and converts them into an electrical signal. Manufacturers do device risk assessment during pre-market testing based on pre-market information such as design, clinical studies, and test-model assumptions, which may not adequately reflect real-world scenarios. Pre-market testing is completed entirely in a controlled setting where the product is entirely shielded from outside impacts, including changes in usage patterns, temperature and humidity variations, and electromagnetic field affects from nearby devices [9]. Users now anticipate having more than just a device. Governments all over the world are focused on cost reduction, including medical devices, despite the fact that there is a huge demand for healthcare, but they also want to improve patient outcomes. In other words, users of medical devices nowadays need more capability, as seen by enhanced classification, clinical outcome prediction, and other tasks requiring intelligence [10]. There are numerous instances today when software is employed to give improved healthcare to patients. These software programs for diagnosing different diseases, which might aid doctors in making accurate diagnoses, usually depend on object-oriented methodology. Healthcare institutions need to handle healthcare technology more effectively and efficiently since they rely on a huge number of medical equipment to deliver high-quality healthcare [11]. In this review, we start our Results section with an analysis of these biosensors and their principles. We present two tables that show different types of biosensors used for Influenza A and HIV detection. Nanotechnology is mentioned as a promising way to design these biosensors. Nanoparticles are used as recognition biomolecules in order to improve the speed and sensitivity of biosensors. To sum up this review, we mention biosensors as a great tool used for the rapid detection of viral infections.

Biosensor for Rapid Methods for the Detection of Viruses

425

2 Methods and Materials This review paper is based on the analysis of 26 scientific, research papers [1–26]. The literature search was conducted in Pub Med, PMC, as well as Google Scholar scientific database. Keywords used for the literature search were biosensors, rapid methods, virus detection, influenza virus detection, and HIV detection. Research papers used for this review study are dated from 2010 to 2020. The list of criteria for the literature search is given in Table 1. Finally, out of 26 papers, 11 fit the criteria and were used for the subsequent analysis. Table 1. Criteria for literature search. Criteria

Included

Excluded

Publication

Research papers Journals

Grey literature

Availability

Full text available

Only abstract

Date of publication

2010–2020

Everything before 2010

Language

English

Non-English

Biosensors

Virus detection

Other use

Viruses

HIV Influenza

Other viruses

Biosensors were analyzed by 4 parameters: biosensor principle, bioreceptor, dynamic range, detection limit, and detection time.

3 Results The results of this review study are presented in Tables 2 and 3. The main concern for the detection of viruses is how to detect them when they are isolated. Gold (Au) has proven to be the best material for this purpose. It has been established that gold nanoparticles (AuNPs) have a lot of use as biosensors, due to their optical properties, high surface area, and their inertness [12]. The ability to detect viruses rapidly is crucial for biosensors. Other methods are time-consumable and not very efficient. Thus, these useful properties of AuNPs are utilized by various biosensors. Using nanoparticles can improve the specificity and rapidity of detection [13]. For example, AuNPs were used by 66% of the biosensors for influenza detection, which all showed to be fast and sensitive [14–17]. Some of them showed faster detection time than others, but when compared to other methods they are all time-efficient. Sayhi et al. reported a detection time of 160 s for their biosensor. The use of MNPs rendered the process of purification unnecessary, as it is quite time-consuming and requires a significant amount of solvent. This kind of biosensor is able to quantify a virus without any immobilization, making it less time-consuming, more stable, and yield higher reproducibility

426

S. Deumi´c et al.

than others [14]. Out of all of the papers, this one had the lowest detection time. Even though this biosensor is the fastest, it did not show the best sensitivity. They stated that the limit of detection (LOD) for this biosensor is 8 HAU [14]. Bai et al. presented a biosensor with a higher sensitivity than the one mentioned earlier. Their biosensor showed a LOD of 0.128 HAU. They also stated that the detection time is 1.5 h, which is the most time-consumable out of all of the others [15]. Jiang et al. presented a biosensor with one of the lowest LOD, which is 1.35 copies/µl [18]. Lee et al. stated that their system did not show the lowest LOD (1 pM). However, this approach has been shown to have high efficiency, due to the ability to change the parts of the DNA 3WJ. It is possible to apply the dual aptamer-tagged bio-probe for the determination of other subtypes of AI viruses. It is also possible to use electrochemical and colorimetric dual-detection methods [17]. The electrochemical principle is the most commonly used principle. The electrochemical principle was the method used in [14, 17, 19]. 2 out of 6 papers used the optical principle [15, 16]. Colorimetric and optical biosensors showed to be the most sensitive ones, with the ability to detect small concentrations of the virus, such as 1.35 copies/µl and 0.128 HAU, respectively [15, 18]. In general, electrochemical biosensors showed to be the best choice for getting the most sensitive biosensor. The LOD for these biosensors were 8 HAU, 1 pM, and 2–1 HAU/50µl, for [14, 17, 19], respectively. The optical biosensor was proven to be the most time-consumable (1.5 h) [15]. The accuracy was not stated for all of the biosensors. The accuracy in these papers was determined through linear regression analysis, providing a linear regression equation and strength of association as the R2 value. Lee et al. stated that the R2 value was 0.9659 and that the relative standard deviation (RSD) was from 1.36 – 6.23%, showing the high reproducibility of this biosensor [16]. Lin et al. had a similar result for the R2 value, 0.96, showing the linear relationship between impedance change and the concentration of the H5N1 virus. They stated that their impedance immunosensor had an accuracy of ~90% [19]. The two most accurate biosensors mentioned had R2 values of 0.99 and 0.996. The first showed that the refraction index (RI) was linearly related to the concentration of AIV [15]. For the second, a regression equation was used to analyze the relationship between the color response and H5 influenza virus concentration [18]. All of the biosensors have high accuracy and based on the R2 value we can say that the colorimetric biosensor has the highest accuracy, with an R2 value of 0.996. The comparison of the properties of these biosensors is given in Table 2. When discussing biosensors for HIV detection, Zheng et al. constructed a sandwich HIV p24 amperometric biosensor using a gold electroplating-modified electrode. Electroplating nano-gold on a glassy carbon electrode (GCE) surface increased the conductivity and reversibility of the electrode and it fixed massive antibodies and maintained their biological activity [20]. This biosensor displayed excellent response to p24 within the concentration range of 0.01–100 ng/mL. Examinations indicated a 98.6% recovery rate from human serum samples. This indicated that amperometric methods could be used to monitor HIV [20]. Another biosensor was developed to measure HIV-originated target sequences using an exonuclease III (Exo III)-assisted target recycling amplification strategy. Gold nanocluster-modified graphene electrodes (GR/AuNCs) had aptamer, which was then

Biosensor for Rapid Methods for the Detection of Viruses

427

Table 2. Comparison of biosensors for the detection of Influenza virus. Biosensor principle

Bioreceptor

Dynamic range Detection limit

Detection time

Paper

Electrochemical

Anti- M2 antibody and Fetuin A

8 – 128 HAUa

8 HAU

160 s

[14]

Optical

Aptamers specific against H5N1

0.128 – 1.28 HAU

0.128 HAU

1.5 h

[15]

Optical

GBP – H1a

Not given

Not given

Not given

[16]

Electrochemical

DNA3WJ/pAuNPs

1 1 pM – 1000000 pM

Not given

[17]

Colorimetric

PCDA/DMPC with 1.35 – 13.5 anti- HA mAb copies/µl

1.35 copies/µl

20 min

[18]

Electrochemical

Monoclonal antibodies

2–1 HAU/50 µl

$1000 per instrument for all of them [25]. Generally, electrochemical biosensors are widely chosen due to their low cost, simplicity of construction, portability, low background noise, and their need for very small sample volumes [26]. The comparison of the properties of these biosensors is given in Table 3.

4 Conclusion Due to the lack of a universal vaccine for viral diseases, dealing with both newly discovered and already known viruses is a very arduous task. Biosensors have proven to be a great tool used in this particular field of rapid detection of viral infections and diseases as information processing from an analyte to the necessary result has proven to be quite fast. In this work, we have summarized previous research explaining on which biosensor principles can be used for early and rapid detection of infectious diseases such as HIV and several varieties of the influenza virus. We have also shown and commented on how the studies, works, and findings of Lee, Sayhi, et al. have contributed to the topic of biosensors in the field of rapid detection of viral diseases. Their approaches to virology have proven to be a great guideline for the future applications of biosensors in virus detection.

References 1. Chan, Y.K., Gack, M.U.: Viral evasion of intracellular DNA and RNA sensing. Nat. Rev. Microbiol. 14, 360–373 (2016) 2. Marston, H.D., Folkers, G.K., Morens, D.M., Fauci, A.S.: Emerging viral diseases: Confronting threats with New Technologies. Sci. Transl. Med. 6 (2014) 3. Afzal, A., Mujahid, A., Schirhagl, R., Bajwa, S., Latif, U., Feroz, S.: Gravimetric viral diagnostics: QCM based biosensors for early detection of viruses. Chemosensors. 5, 7 (2017) 4. Lee, B.W., Bey, R.F., Baarsch, M.J., Simonson, R.R.: ELISA method for detection of influenza A infection in swine. J. Vet. Diagn. Invest. 5, 510–515 (1993) 5. Hoet, A.E., Chang, K.-O., Saif, L.J.: Comparison of Elisa and RT-PCR versus immune electron microscopy for detection of bovine torovirus (Breda virus) in calf fecal specimens. J. Vet. Diagn. Invest. 15, 100–106 (2003) 6. Mokhtarzadeh, A., et al.: Nanomaterial-based biosensors for detection of pathogenic virus. TrAC, Trends Anal. Chem. 97, 445–457 (2017) 7. Grieshaber, D., MacKenzie, R., Vörös, J., Reimhult, E.: Electrochemical biosensors - sensor principles and Architectures. Sensors 8, 1400–1458 (2008)

Biosensor for Rapid Methods for the Detection of Viruses

429

8. Dalal, A., Mohan, H., Prasad, M., Pundir, C.S.: Detection methods for influenza A H1N1 virus with special reference to biosensors: a review. Biosci. Rep. 40 (2020) 9. Badnjevic, A.: Evidence-based maintenance of medical devices: current shortage and pathway towards solution. Tech. Health Care. 1–13 (2022). https://doi.org/10.3233/thc-229005 10. Badnjevi´c, A., Avdihodži´c, H., Gurbeta Pokvi´c, L.: Artificial intelligence in medical devices: past, present and future. Psychiatria Danubina. 33, S336–S341 (2021) 11. Gurbeta, L., Badnjevi´c, A.: Inspection process of medical devices in healthcare institutions: software solution. Heal. Technol. 7, 109–117 (2016). https://doi.org/10.1007/s12553-0160154-2 12. Yu, X., Jiao, Y., Chai, Q.: Applications of gold nanoparticles in biosensors. Nano LIFE 06, 1642001 (2016) 13. Draz, M.S., Shafiee, H.: Applications of gold nanoparticles in virus detection. Theranostics 8, 1985–2017 (2018) 14. Sayhi, M., et al.: Electrochemical detection of influenza virus H9N2 based on both immunomagnetic extraction and gold catalysis using an immobilization-free screen printed carbon microelectrode. Biosens. Bioelectron. 107, 170–177 (2018) 15. Bai, H., Wang, R., Hargis, B., Lu, H., Li, Y.: A SPR aptasensor for detection of avian influenza virus H5N1. Sensors 12, 12506–12518 (2012) 16. Lee, T., et al.: Fabrication of electrochemical biosensor consisted of multi-functional DNA structure/porous au nanoparticle for Avian Influenza Virus (H5N1) in chicken serum. Mater. Sci. Eng., C 99, 511–519 (2019) 17. Lee, K.G., et al.: Development of a plastic-based microfluidic immunosensor chip for detection of H1N1 influenza. Sensors 12, 10810–10819 (2012) 18. Jiang, L., et al.: Development and evaluation of a polydiacetylene based biosensor for the detection of H5 influenza virus. J. Virol. Methods 219, 38–45 (2015) 19. Lin, J., et al.: An impedance immunosensor based on low-cost microelectrodes and specific monoclonal antibodies for rapid detection of avian influenza virus H5N1 in chicken swabs. Biosens. Bioelectron. 67, 546–552 (2015) 20. Zheng, L., et al.: A sandwich HIV P24 amperometric immunosensor based on a direct gold electroplating-modified electrode. Molecules 17, 5988–6000 (2012) 21. Shafiee, H., Lidstone, E.A., Jahangir, M., Inci, F., Hanhauser, E., Henrich, T.J., Kuritzkes, D.R., Cunningham, B.T., Demirci, U.: Nanostructured optical photonic crystal biosensor for HIV viral load measurement. Sci. Rep. 4 (2014) 22. Lu, C.-H., et al.: Sensing HIV related protein using epitope imprinted hydrophilic polymer coated quartz crystal microbalance. Biosens. Bioelectron. 31, 439–444 (2012) 23. Wu, S., He, Q., Tan, C., Wang, Y., Zhang, H.: Graphene-based electrochemical sensors. Small 9, 1160–1172 (2013) 24. Gilbert, M., Kirihara, J., Mills, J.: Enzyme-linked immunoassay for human immunodeficiency virus type 1 envelope glycoprotein 120. J. Clin. Microbiol. 29, 142–147 (1991) 25. Lifson, M.A., et al.: Advances in biosensing strategies for HIV-1 detection, diagnosis, and therapeutic monitoring. Adv. Drug Deliv. Rev. 103, 90–104 (2016) 26. Huang, Y., Xu, J., Liu, J., Wang, X., Chen, B.: Disease-related detection with electrochemical biosensors: a review. Sensors 17, 2375 (2017)

Implantable and Ingestible Biosensors Neira Crnˇcevi´c(B)

, Damilola M. Ajayi , Tarik Zubˇcevi´c , Sara Deumi´c , and Haris Koli´c

International Burch University, Francuske Revolucije bb, 71 210 Ilidža, Bosnia and Herzegovina {neira.crncevic,tarik.zubcevic,sara.deumic}@ibu.edu.ba, {damilola.mildred.ajayi,haris.kolic}@stu.ibu.edu.ba

Abstract. Implantable and ingestible biosensors are used in medicine and research to monitor patients’ health over extended periods. This review paper focuses on applying these biosensors, their strengths, limitations, stability, and risks. The use of these biosensors has the potential to reduce the costs in the healthcare system and improve the quality of patient treatment. All biosensors mentioned in this review paper allow healthcare workers to monitor patients’ health in real-time and improve their overall quality of life. The ingestible micro bioelectronic device (IMBED) and the My-treatment-medication (MyTMed) have relatively high accuracy and provide information about the gastrointestinal health of the patient. Their high sensitivity is vital for detecting small traces of chemicals in the gastrointestinal tract. Using the ingestible micro-bio-electronic device provides faster detection of gastrointestinal bleeding, especially in gastro intestinal anatomical parts that are hard to reach. My-treatment-medication provides healthcare workers a way to gather information about medication adherence, which was previously based only on the patient’s subjective answers. The implantable cardioverter defibrillator and the glucose biosensor prevent heart failure and track the glucose values in the patient’s body. Both of them exhibit high accuracy and sensitivity. The irregular heart rate, which can lead to cardiac arrest, can now be monitored and prevented using the implantable cardioverter defibrillator. The glucose biosensor can improve the quality of life of patients who have diabetes. Keywords: Biosensors · IMBED · MyTMed technology · Glucose detection

1 Introduction A biosensor is a device that measures biological or chemical reactions by generating signals proportional to the concentration of an analyte [1]. These biosensors must be able to display target analytes and distinguish them from potentially interfering species. Also, these devices must retain the appropriate sensitivity for monitoring the analyte in the range of concentration found in the sample. In general, only biological molecules display the required sensitivity and specificity to provide recognition systems for electrochemical devices. Although the history of biosensors dates back to 1906, the first true biosensor was created by Leland C. Clark Jr in 1956 for oxygen detection and his invention of the oxygen electrode bore his name: “Clark Electrode.“ [1]. The field of implantable and © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 430–437, 2024. https://doi.org/10.1007/978-3-031-49068-2_44

Implantable and Ingestible Biosensors

431

ingestible biosensors has rapidly grown for the past ten years. In the past, the market for continuous monitoring devices was minimal, especially for glucose monitoring devices. As of 2018, the biosensor market is around $18.6 billion, with ingestible biosensors averaging $198.2 million. An implantable device can be introduced fully or partially into the human body through surgical or medical methods and is intended to remain there after the procedure [3]. The implantable cardioverter defibrillator can monitor the heart rate continuously. The use of this biosensor can prevent cardiac arrest [4]. Implantable biosensor devices are produced in different shapes, weights, and sizes to make them compatible with varying human activities and reduce user discomfort. The implantable glucose biosensor measures the insulin and glucose rate in the blood. Since diabetes mellitus affects 422 million people worldwide, so this biosensor is more needed than ever before. This biosensor can potentially improve the quality of life of patients with diabetes and prevent conditions that arise from diabetes, such as strokes, kidney failures, heart attacks, and even death [6]. Ingestible biosensors represent electronic devices the size of a medical capsule that can telecommunicate and help healthcare professionals in disease diagnostics and continuous patient monitoring regardless of the patient’s physiological state [7]. The sizes of ingestible biosensors can vary depending on their application. The basic structure of the ingestible sensor is a gelatin capsule, an inert circuit, and a radiofrequency emitter. The medication of choice could be put inside it. The stomach acid dis solves the capsule after the patient swallows the tablet. This releases the medication, and the hydrogen ions inside the gastric juice activate the radiofrequency emitter. Non adherence to medications is associated with significant mortality, accelerated disease progression, and increased healthcare costs. My/Treatment/Medication (MyTMed) is a novel adherence monitoring system that directly measures medication adherence/non adherence [8]. My-treatment-medication monitors the medication adherence of a patient and the time at which the patient took the medication. This biosensor is helpful since nonadherence to medication is associated with accelerated disease progression, increased mortality rates, and increased costs to the healthcare system. The target groups for this biosensor are HIV-positive patients who suffer from chronic diseases and have schizophrenia [9, 10]. The ingestible micro-bio-electronic-device detects bleeding in the gastrointestinal tract and gut markers thiosulfate and acyl-homoserine lactone. The main limitation of these biosensors is that they cannot obtain accurate information about gastrointestinal bleeding in the patient. This is because the pH of the environment can interfere with the biosensor and cause it to obtain results with lower accuracy. This limitation is present in the ingestible micro-bio-electronic-device [10]. This paper reviews the application of implantable and ingestible biosensors, focusing on their strengths and limitations.

2 Methods and Materials 26 original, full versioned articles in English language concerning implantable and in gestible biosensors were examined. These papers were collected from the following scientific databases: Google Scholar, Research Gate, Science Direct, Web of Science

432

N. Crnˇcevi´c et al.

and Pubmed. All papers were published in the period 2015–2020. The objective of this review study was to analyze the application of implantable and ingestible biosensors. All reported biosensors were assessed by 21 parameters such as: accuracy, sensitivity, specificity, size, usage, success rate, ethical issues, battery, location of usage in the human body, connection abilities, application support, monitoring in real-time, recognition element, testing, shelf life, safety, toxicity, stability of the device.

3 Results The results of this study show that implantable biosensors are mostly applied in medical science, such as cardiology, where they are used to control or treat abnormal heartbeats. On the other hand, the glucose biosensors detect insulin and glucose rate in the blood [11]. The ingestible biosensors are more used in internal testing and diagnosis. The results of the review of the applications of implantable and ingestible biosensors are presented in Table 1. The analysis shows that implantable biosensors, like implantable cardioverter defibrillators and implantable glucose biosensors, are an important class of biosensors [11]. In contrast, for ingestible biosensors the ingestible microbio-electronic device (IMBED) and My-treatment-medication (MyTMed) biosensor are used as representatives of the ingestible biosensors.

4 Discussion Out of 6 analyzed papers that involved the ingestible biosensor micro-bio-electronic device (IMBED), three reports that they are very sensitive. Their sensitivity is reported to be 83.3%, and it improved over time to 100% at 120 min. These values were reported during the experiments regarding gastrointestinal bleeding, which were conducted on pigs. The photon flux can be captured as low as 5 × 104 photons/s incident on the 0.29 mm2 area of detectors [10, 14]. The micro-bio-electronic device’s accuracy decreases as the gastrointestinal tract’s pH decreases. The low pH levels interfere with the results. This is why the pigs are given a bicarbonate-glucose neutralization solution to decrease the acidity of the environment where the biosensor will be located. This limitation is reported in 4 papers regarding this topic. The four papers also suggest that the biosensor is very specific since they have specifically engineered a strain of Escherichia coli called Nissle 1917 that can detect heme in the blood. It represents the recognition element for this biosensor [10, 14, 16, 17]. Out of 8 analyzed papers that involved the ingestible biosensor micro-bio-electronic device (IMBED), one paper specified the dimension of this biosensor. The length of the biosensor is 3.5 cm. The small size enables the patient or animal to ingest the biosensor with slight discomfort, but this is still much less painful than the traditional methods of detecting gastrointestinal bleeding. In these 3.5 cm, engineers have succeeded in incorporating all of the components needed for its proper functioning while still having a small battery with 2.7 V [7]. Three papers regarding the implantable biosensor micro-bio-electronic device suggest that the biosensor cannot be used yet in humans for common medical practices.

Implantable and Ingestible Biosensors

433

Table 1. Results of the review of most important properties of implantable and ingestible biosensors. Biosensors

Ingestible

Implantable

Ingestible microbioelectronic device [7, 10, 14–17]

My-treatment-medication [8, 9, 18, 20]

Implantable cardioverter defibrillator [2, 11, 13, 22]

Implantable glucose biosensor [4, 11, 21, 23–26]

Accuracy

High (100%)

Relatively high (87.3%)

Relatively high (80%)

Relatively high (70%)

Sensitivity

Very sensitive

Very sensitive

Very sensitive

Very sensitive

Specificity

Very specific

Very Specific

Very specific

Very specific

Type

Electrochemical

Electrochemical

Electrochemical

Amperometric

Size

3.5 cm

1–1.5 cm

3.26 mm2

0.4 * 1.3 cm2

Wide usage

Not yet

Not yet

Yes

Yes

Success

Successful

Successful

Successful

Successful

Ethical issues

None

Privacy issues

None

None

Battery

Needed

Needed

Needed

Not needed

Location

Upper GI tract

Abdomen

Abdomen or chest

Forearm area

Laptop of cellphone connection

Yes

Yes

No

Yes

Custom application

Yes

Yes

No

Yes

Real-time monitoring

Yes

Yes

Yes

Yes

Tag

Blood vessels in heart

Glucose oxidase

Recognition element Echerichia coli Nissle 1917 Human testing

No

Yes

Yes

Yes

Animal testing

Pigs, mice, horses’ blood

No

Dogs

Rats

Safety

Very safe

Very safe

Safe

Safe

Toxicity

Non-toxic

Non-toxic

Non-toxic

Non-toxic

Invasiveness

Minimally-invasive

Non-invasive

Invasive

Invasive

Device stability

Highly stable up to 36 h

Highly stable

Highly stable for weeks, months or years

Highly stable

Major Strength

Detection of GI bleeding and abnormalities GI environments

Accurate detection of medication adherence

Prevention of irregular heart rates

Accurate detection of insulin and glucose levels in the blood

(continued)

This is because it is too dangerous for a human to ingest a neutralization solution tested on pigs. Without the neutralization solution, the results would not be accurate enough. However, experiments in pigs, mice, and horse blood showed very high accuracy. There are no ethical issues surrounding the application of this biosensor. [7, 10, 14]. Five papers out of 8 covering the topic of the ingestible micro-bio-electronic device discuss the availability to monitor the gastrointestinal tract in real time. The biosensor can be connected to a laptop or a cellphone. Only Android devices can be used for this. It also has a custom app, which is used for easier monitoring in real time [7, 10, 14, 17, 18],

434

N. Crnˇcevi´c et al. Table 1. (continued)

Biosensors

Major limitation

Ingestible

Implantable

Ingestible microbioelectronic device [7, 10, 14–17]

My-treatment-medication [8, 9, 18, 20]

Implantable cardioverter defibrillator [2, 11, 13, 22]

Implantable glucose biosensor [4, 11, 21, 23–26]

Accuracy of detection decreases in highly acidic environments

Requires the external receivers to be adhered to the patient’s body

Restricts use of electronics because of electrical interference

Easily damaged during implementation

Cost-effectiveness

Yes

Yes

Yes

Yes

Potential risk

Low pH of environment interferes with the results

Tag retention in the gastrointestinal tract and privacy issues

Risk of infection at implant site

Risk of damage during implantation

Four papers regarding ingestible My-treatment-medication (MyTMed) were re viewed. The average reported accuracy for ingestible biosensors in analyzed papers [8] and [19] is 87.3%. This value was obtained from an experiment that included 10 participants. The paper [8] states that My-treatment-medication can monitor medication adherence in real-time. It also mentions certain ethical issues regarding this biosensor. This ingestible biosensor can be connected to a laptop with a cloud-based web server. The SMS and MMS signals can be sent to a cell phone. It also has a custom app that can be used by the patient to monitor their medication adherence and be reminded when to take the medication. However, this biosensor has certain ethical issues. Privacy issues are one of these concerns. Because of this, an adequate security system needs to be set up to avoid intrusions and unauthorized access to the patient’s information [8]. A major advantage of My-treatment-medication device when compared to ingestiblemicro-bio-electrical device is that the accuracy of the biosensor does not decrease in a highly acidic environment. My-treatment-medication device can properly emit radio frequencies at pH 2. This has been reported in the paper [9]. Out of 4 papers regarding My-treatment-medication device, paper [19] states that this biosensor can drastically decrease the mortality rates of patients who suffer from chronic conditions. This biosensor has the potential to gain a large patient acceptance, and it is not difficult to use it. In the paper [8], it is stated that 96% of patients who have bipolar disorder and schizophrenia have managed to complete a pilot study in which they had to use this device. Also, HIV-positive drug users from this study stated that they are willing to use this device if patient confidentiality is highly regulated [8, 19]. 4 out of 4 papers reviewed concerning implantable cardioverter defibrillators were independent of external devices such as laptops and mobile phones. To function properly, they only need to be connected to the bio element to activate them and give out the required signals [2, 11, 13, 22]. Paper [2] estimated 6 to 10 years as the acceptable time for an implantable cardioverter defibrillator can be implanted in patients. This makes this biosensor suitable for long-term use and economically reasonable. Among the ten papers used in this review for implantable biosensors, articles [25] and [2] give information concerning the size of the implantable cardioverter defibrilla

Implantable and Ingestible Biosensors

435

tor and the implantable glucose device. Since both biosensors have to be implanted into the body, it is important that these biosensors are small and lightweight enough to be implanted into the patients without weighing them down and causing unnecessary bur dens for the patients. Any unnecessary weight could lead to other illnesses or even affect the functioning of the devices in the area in which they have been implanted. For this reason, the implantable defibrillator cardioverter is approximately 64 mm × 51 mm × 13 mm, and the implantable glucose device is 0.4 × 1.3 cm2 . In papers [4, 25], and [26], it is stated that the accuracy of the glucose biosensor is directly proportional to the amount of serum/bio receptor interacting with the biosensor. In the case of an implantable glucose biosensor, the device constantly interacts with glucose oxidase, the recognition element. Hence, the accuracy tends to increase and be more reliable in contrast to glucose biosensors that are not implanted into the body. Paper [25] states that the implantable glucose biosensor can be connected to a computer or laptop device to give out reading through piezoelectric voltage. As the concentration of glucose increases, the reading on the computer increases too. Paper [11] states that all implantable biosensors have been tested and are still being tested on veterinary patients and animal models. Implantable cardioverter defibrillators have been tested specifically on dogs. In contrast, implantable glucose biosensors were tested on rats for the purpose of determination of the safety and working procedure of these biosensors in the body. In paper [24], in a study on 51 diabetic patients to determine the accuracy of implantable glucose biosensor reading for insulin and glucose level, 70% of the sensor reading satisfied the ISO criteria for accuracy, an acceptable value, and approved for use medically. In paper [23], in another study carried out on 71 diabetic patients over 180 days, the analysis showed that 99.2% of samples were in the clinically acceptable error zones. 81% of hypoglycemic events were detected within 30 min, with no device-related serious adverse events during the study. According to paper [13], a study was carried out on patients over the age of 65 who received implantable cardioverter defibrillators after surviving sudden cardiac arrest. The result showed that over 80% of the patients survived the past two years, showing the device’s efficiency. In paper [22], the implantable cardioverter defibrillator complication rate can be as high as 10%. These complications include vascular injuries, cardiac perforation, pneumothorax, and hemothorax. The complications can be very dangerous and can increase the patient mortality rate. Paper [25] talks about the risks of the implantable glucose biosensor application. One of these risks is that the device can break during implantation, which can lead to inaccurate results or no results at all. This would mean that the device could not be used at all.

5 Conclusion The research on implantable and ingestible biosensors shows that using ingestible and implantable biosensors in research and medicine represents the future and a new model of the healthcare system. Although this field of technology is fairly new, it has promising results. The use of these biosensors provides faster and more accurate diagnosis,

436

N. Crnˇcevi´c et al.

reduces the waiting lists in hospitals, and has the potential to improve the overall quality of life of patients. Based on the analyzed papers, the ingestible micro-bio-electronic device (IMBED) has tremendous potential due to its high accuracy, sensitivity, stability, and specificity. This electrochemical biosensor still has not been tested on humans, but because of its properties and ability to detect abnormalities in the GI tract, this will likely happen in the near future.

References 1. Bhalla, N., Jolly, P., Formisano, N., Estrela, P.: Introduction to biosensors. Essays Biochem. 60(1), 1–8 (2016) 2. Dharun, V.S., Simon, S.G., Shihab, S & Thahan. S.: Implantable cardioverter defibrillator. Int. J. Appl. Eng. Res. 10(59) (2015) 3. Ugalmugle, S., Swain, R.: Biosensors market by type. Globalmarketinsight (2019) 4. Heo, Y.J., Kim, S.H.: Toward long-term implantable glucose biosensors for clinical use. Appl. Sci. 9(10), 2158 (2019) 5. Chen, C., Zhao, X.L., Li, Z.H., Zhu, Z.G., Qian, S.H., Flewitt, A.J.: Current and emerging technology for continuous glucose monitoring. Sensors 17(1), 182 (2017) 6. Vaddiraju, S., Tomazos, I., Burgess, D.J., Jain, F.C., Papadimitrakopoulos, F.: Emerging synergy between nanotechnology and implantable biosensors: a review. Biosens. Bioelectron. 25(7), 1553–1565 (2010) 7. Bhatt, A.S.: Digesting new developments in biosensors. N. Engl. J. Med. 379(7), 686–688 (2018) 8. Chai, P.R., Rosen, R.K., Boyer, E.W.: Ingestible biosensors for real-time medical adherence monitoring: MyTMed. In: Proceedings of the ... Annual Hawaii International Conference on System Sciences. Annual Hawaii International Conference on System Sciences, pp. 3416– 3423 (2016) 9. Guk, K., et al.: Evolution of wearable devices with real-time disease monitoring for personalized healthcare. Nano Materials (Basel, Switzerland) 9(6), 813 (2019) 10. Mimee, M., Nadeau, P., Hayward, A., Carim, S., Flanagan, S., Jerger, L., Collins, J., McDonnell, S., Swartwout, R., Citorik, R.J., Bulovi´c, V., Lu, T.K.: An ingestible bacterial-electronic system to monitor gastrointestinal health. Science 360(6391), 915–918 (2018) 11. Gray, M., et al.: Implantable biosensors and their contribution to the future of precision medicine. The Veterinary J. 239, 21–29 (2018) 12. DeVries, L.M., Dijk, W.A., Hooijschuur, C.A.M., Leening, M.J.G., Hemel, V.: Utilisation of cardiac pacemakers over a 20-year period: results from a nationwide pace maker registry. Neth Heart J. 25(1), 47–55 (2017) 13. Masoudi, F.A., Go, A.S., Magid, D.J., Cassidy-Bushrow, A.E., Gurwitz, J.H., Liu, T.I., Reynolds, K., Smith, D.H., Reifler, L.M., Glenn, K.A., Fiocchi, F., Greenlee, R.T.: Age and sex differences in long-term outcomes following implantable cardioverter-defibrillator placement in contemporary clinical practice: findings from the cardiovascular research network. J. Am. Heart Assoc. 4(6), e002005 (2015) 14. Gastone Ciuti, R., Caliò, D., Camboni, L., Neri, F., Bianchi, A.A.: Frontiers of robotic endoscopic capsules: a review. J. Micro-Bio Robot. 11, 1–18 (2016) 15. Patel, P.: Gulp! ACS Publications, Electronics Down the Hatch (2017) 16. Min, J., Yang, Y., Wu, Z., Gao, W.: Robotics in the gut. . Adv. Therapeutics 3(4), 1900125 (2020) 17. Kalantar-Zadeh, K., Ward, S.A.: Future is ready for swallowable sensors. Hepato Biliary Surg. Nutr. 8(3), 267–269 (2019)

Implantable and Ingestible Biosensors

437

18. Inda, M.E., Mimee, M., Lu, T.K.: Cell-based biosensors for immunology, inflammation, and allergy. J. Allergy Clin. Immunol. 144(3), 645–647 (2019) 19. Brown, M.T., Bussell, J.K.: Medication adherence: WHO cares?. Mayo Clinic Proc. 86(4), 304–314 (2011) 20. Aldeer, M., Javanmard, M., Martin, R.P.: A review of medication adherence monitoring technologies. Appl. Syst. Innov. 1(2), 14 (2018) 21. Chen, C., Zhao, X.L., Li, Z.H., Zhu, Z.G., Qian, S.H., Flewitt, A.J.: Current and emerging technology for continuous glucose monitoring. Sensors (Basal), 17(1), 182 (2017) 22. Kempa, M., Budrejko, S., Zienciuk-Krajka, A., Daniłowicz-Szymanowicz, L., Królak, T., Opielowska-Nowak, B., Kwiatkowska, J., Raczak, G.: Subcutaneous implantable cardioverter defibrillators for the prevention of sudden cardiac death: five-year single-center experience. Kardiologia Polska (Polish Heart J.), 78(5), 447–450 (2020) 23. Kropff, J., et al.: Accuracy and longevity of an implantable continuous glucose sensor in the PRECISE study: a 180-day, prospective, multicenter, pivotal trial. Diab. Care 40(1), 63–68 (2017) 24. Leelarathna, L., Nodale, M., Allen, J.M., Elleri, D., Kumareswaran, K., Haidar, A., Caldwell, K., Wilinska, M.E., Acerini, C.L., Evans, M.L., Murphy, H.R., Hovorka, R.: Evaluating the accuracy and large inaccuracy of two continuous glucose monitoring systems. Diab. Technol. Therapeutics, 15(2), 143–149 (2013) 25. Zhang, W., et al.: Self-powered implantable skin-like glucometer for real-time detection of blood glucose level in vivo. Nano-Micro Lett. 10, 32 (2018) 26. Hammadi, A.M., Humadi, A.F., Mahmood, A.L.: New optical fiber biosensor method for glucose in serum. Mater. Sci. Eng. 745 (2020)

Characteristics of Ocular Following Responses (OFRs) in Children with Stereodeficiencies Aleksandar Miladinovi´c1(B) , Christian Quaia2 , Miloš Ajˇcevi´c3 , Laura Diplotti1 , Simone Kresevic3 , Stefano Pensiero1 , and Agostino Accardo3 1 Department of Ophthalmology, Institute for Maternal and Child Health-IRCCS “Burlo

Garofolo”, Trieste, Italy [email protected] 2 Laboratory of Sensorimotor Research, Department of Health and Human Services, National Eye Institute, National Institutes of Health, Bethesda, MD, USA 3 Department of Engineering and Architecture, University of Trieste, Trieste, Italy

Abstract. Ocular following responses (OFRs) are reflexive eye movements triggered by sudden motion of textured patterns in the visual field. OFRs have been extensively studied in both humans and non-human primates and are mediated by cortical neurons sensitive to disparity. Recent studies have demonstrated that OFRs can be recorded non-invasively using a video-based eye tracker. We examined the characteristics of OFRs in children with stereodeficiencies. Eight children between 5 and 12 years old with stereodeficiency associated with either unilateral refractive amblyopia or strabismic amblyopia were enrolled. To provide comparison we also included eight age/sex matched healthy children. We found that the amplitude of the OFRs observed in children with stereodeficiencies is similar to amplitudes of children with typical stereo vision and those documented also in healthy adults. By confirming the existence of the reflex in children with stereodeficiencies, this study lays the groundwork for further exploration of how stimulus interocular correlation affects OFRs in children. Keywords: Ocular following response · Stereopsis · Video-oculography · Eye-tracking

1 Introduction Ocular following responses (OFRs), reflexive eye movements triggered by sudden motion of (usually large) textured patterns in the visual field, have been extensively studied in both humans and non-human primates [1–4]. These eye movements, which occur at ultra-short latencies of around 70 ms in humans, are the initial component of the optokinetic nystagmus response that supports the translational vestibulo-ocular reflex system in the stabilization of gaze [5]. Recent studies have shown that OFRs can also be induced, albeit at longer latencies, by relatively small stimuli [6, 7], suggesting that they may also contribute to the initial phase of smooth pursuit eye movements. The strong dependency of OFRs on various stimulus properties, such as size, contrast, and spatial and temporal frequency content, has made them a powerful tool for This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply 2024 A. Badnjevi´c and L. Gurbeta Pokvi´c (Eds.): MEDICON 2023/CMBEBIH 2023, 94, pp. 438–446, 2024. https://doi.org/10.1007/978-3-031-49068-2_45

Characteristics of Ocular Following Responses (OFRs)

439

investigating the visual motion processing in humans. Recently, it has been inferred that, since OFRs are sensitive to interocular correlations [8], they must be mediated by disparity-tuned cortical neurons, and in principle can thus be used as a diagnostic tool for identifying stereo-anomalies [9]. Since OFRs are typically quantified over a very short time window (the so-called open-loop period), their magnitude is however very small, making recording them difficult and their functional role in clinical settings questionable. Nonetheless, OFRs have been recorded non-invasively using video eye trackers, and we have recently been able to recorded them in healthy young children in a clinical setting [10]. As far as we know, they have however never been recorded in children with stereodeficiencies. In this preliminary study we verified that it is indeed possible to measure OFRs in such a patient population, providing the basis for further investigations into their potential clinical applications.

2 Materials and Methods 2.1 Study Population Between June and August 2022, the Ophthalmology Department at the Institute for Maternal and Child Health-IRCCS Burlo Garofolo (Trieste, Italy) enrolled eight children (5M/3F aged 9.4 ± 2.9 yo) who underwent a complete orthoptic and ophthalmological examination. An ophthalmologist first examined each child to determine the presence of strabismus, evaluated stereo vision and visual acuity through non-cycloplegic refraction. The study included patients between 5 and 12 years old with stereodeficiency (Titmus stereoacuity > 200 arcsec) associated by either unilateral refractive amblyopia or strabismic amblyopia. Exclusion criteria were organic visual loss and any other cause of visual impairment besides refractive and strabismic amblyopia. In addition, eight age/sex matched healthy children were also enrolled. 2.2 HR-VOG and Behavioral Paradigm The experimental setup involved seating the children in a dimly lit room and stabilizing their heads using padded chin and forehead supports, along with a loosely tied headband. The children faced a monitor (ASUS VG248QE), which was positioned 50 cm away from the corneal vertex and set to a resolution of 1920x1080 pixels and a vertical refresh rate of 144Hz. The chair height was adjusted to align the subject’s eyes with the center of the screen. The developed HR-VOG (High-Resolution Video-oculography) system includes a high-resolution camera (FLIR Grasshopper 3 GS3-U3-51S5M-C) with a resolution of 2448 × 2048 pixels, fitted with a C-Mount 50mm f/2.8 lens, which records in the nearinfrared spectral range. To block the visible spectrum and capture images only in the near-infrared range, a Hoya R72 IR filter with a cut-off wavelength of 720 nm was placed in front of the lens. To ensure adequate lighting conditions, we utilized three custom-built infrared (IR) LED illuminators (one positioned on each side of the subject, and one placed in front and below the subject) as shown in Fig. 1A. For each trial, we acquired three frames: t0 = 0 ms (fixation cross offset/motion onset), t1 = 80 ms, which

440

A. Miladinovi´c et al.

corresponds to the typical ocular following latency in humans to stimuli of similar size and contrast used in previous studies [6, 7], and t2 = 160 ms, which marks the end of the open-loop period of the movement.

Fig. 1. A) The custom designed HR-VOG used for recording. B) The stimuli used which at t0 begins to drift rapidly (50°/s) upward or downward for a duration of 200 ms.

At the start of each trial, the screen was filled with a mid-luminance (7.0 cd/m2 ) binocular blank background. A central fixation cross was then presented to the dominant eye only, to prevent stereodeficient participants from alternating the eye used to fixate across trials. Stimuli were presented within a square aperture (28 deg side) centered on the screen and had a mean luminance of 7.0 cd/m2 . Vertical OFRs were induced using low-pass filtered horizontal 1D random line stimuli. Each stimulus was generated by randomly assigning a high or low luminance value (symmetric around the mean luminance) to consecutive pairs of pixel rows (0.06deg), and the resulting stimulus was then filtered with low-pass spatial filter. The gain of the filter was zero above 0.75 cycles per degree (cpd) and one below 0.375 cpd, with the transition following a raised-cosine function. The root mean square contrast of the stimulus was set to 22%, which kept the Michelson contrast below 100% and prevented saturations (Fig. 1B). The drift speed of the stimuli was approximately 50°/s. 2.3 Data Analysis After obtaining a reference frame at time t0, our algorithm extracted the displacement of the head marker and pupil center at frames t1 and t2 with sub-pixel resolution, using the reference frame as a guide. To calculate the eye displacement (eye-in-head) during the fixation epoch (frames t1 vs t0) and movement epoch (frames t2 vs t1), we subtracted the head marker displacement (head-in-space) from the pupil displacement (eye-in-space). A procedure was applied to automatically exclude from the analysis trials in which fixation was poor (for example, because of the presence of saccades, microsaccades, or large head movements). The algorithm determined that neither the head nor the eyes were moving during the fixation period, while only the eyes were moving during the movement period. Accordingly, the displacement of the head (in space) and the eye (in space and relative to the head) during the fixation period, and the displacement of the head (in space) during the movement period were assessed. Any trials in which any of

Characteristics of Ocular Following Responses (OFRs)

441

these measures were identified as outliers were excluded. After that, the values of each measure were sorted in ascending order across all trials and identified outliers as those with a value larger than three times the 68th percentile value. On average, this procedure resulted in excluding 15% of the trials (ranging from 5% to 25%). We then computed the difference (mean and SD) of the vertical displacements induced by upward and downward motion, which are referred to as the vertical OFRs. The vertical eye displacements (eye-in-head) induced by upward and downward moving stimuli were compared for each subject using a paired t-test and and Mann-Whitney U test with a 0.05 significance level. All statistical analyses were performed in Python using the statsmodels and scipy packages. 2.4 Ethics Statement Each participant’s parent or legal guardian provided written informed consent prior to their child’s involvement in the study. They were informed that the test would not directly benefit their child, that participation was entirely voluntary and unrelated to any clinical care the child was receiving at the eye clinic that day, and that their decision to decline participation would not impact their child’s clinical care at the hospital, either currently or in the future. The equipment used in the study was demonstrated to the parent, and a few sample trials were conducted to illustrate the visual stimulation and what was expected of the child. Additionally, the parent was present throughout the testing session. This research was approved by the Institutional Review Board of IRCCS Burlo Garofolo and follows the principles of the Declaration of Helsinki.

3 Results In Fig. 2, scatter plots displaying the vertical eye displacements for upward and downward drifting stimuli, recorded from one representative subject (subject 3) with stereodeficiency, are presented. The left column illustrates the displacement during fixation (t1 vs t0, 0–80 ms from stimulus onset), while the right column shows the displacement during the movement-epoch (t2 vs t1, 80–160 ms from stimulus onset, also known as the open-loop period where the magnitude of ocular following movements is typically assessed). In Fig. 3, the individual trials are represented by small dots, and the mean is represented by a larger dot (the bars indicating ±1 standard-error-of-the-mean are also displayed). The displacement was minimal and unrelated to the direction of motion of the stimuli during the fixation epoch. However, during the movement epoch, the eye displacements were substantial (averaging around 0.15°-0.25° in each direction). In order to provide the qualitative comparison with age/sex matched children, in Figs. 4 and 5 are reported displacement recorded in a representative single healthy subject (subject 9) and mean displacements in all subjects, respectively. None of the subjects displayed significant displacement during the fixation epoch (Table 1). However, during the movement epoch, all subjects showed significant differences in vertical displacement induced by 1D drifting stimuli moving in the upward and downward directions (Table 2). The same was observed in healthy group (Tables 3 and 4).

442

A. Miladinovi´c et al.

Fig. 2. Example of eye displacements recorded in a subject with stereodeficiency during the fixation (left panels) and movement (right panel).

Fig. 3. Displays scatter plots of the mean eye displacements during the fixation (left) and movement (right) epochs across all stereodeficient subjects.

4 Discussion We investigated the characteristics of OFRs in children with stereodeficiencies, associated with refractive amblyopia and strabismic amblyopia. In this study, the OFRs were successfully recorded in response to stimuli in all eight subjects, which indicates that non-invasive recording of OFRs is feasible in school year children with stereodeficiencies. Our results demonstrate the existence of OFRs also in stereodeficient children, enabling the future development of OFRs based diagnostic tools in this population. We also provided the qualitative comparison with age/sex matched healthy children. Our findings show that the amplitude of the OFRs in children with stereodeficiencies is comparable to the amplitude observed in healthy controls.

Characteristics of Ocular Following Responses (OFRs)

443

Fig. 4. Example of eye displacements recorded in a healthy subject during the fixation (left panels) and movement (right panel).

Fig. 5. Displays scatter plots of the mean eye displacements of healthy subjects during the fixation (left) and movement (right) epochs across all healthy subjects.

In addition, nonetheless we used a slightly different stimuli, the data was also comparable to those reported in our previous study [10]. Our next step will be to manipulating independently the visual stimulus presented to each eye, in both normal and stereodeficient children, which might reveal differences of clinical relevance. Furthermore, the non-invasive recording of OFRs using video-based eye trackers provides a promising approach for evaluating visual motion processing in clinical settings, which may have implications for the early identification and treatment of visual processing disorders. In conclusion, our pilot study provides evidence that OFRs can be recorded noninvasively in children with stereodeficiencies and opens the door to the future use of this system for stereodeficiency diagnosis.

444

A. Miladinovi´c et al.

Table 1. Responses to fixation cross offset/motion onset of stereodeficient subjects (y: Vertical eye displacement in 0-80ms time window (deg); N: number of correct trials; p: unpaired t-test; p_np: Mann-Whitney U test) Subj

Age/Sex

1

12/M

2

y ± SD (N) UP

y ± SD (N) DW

p

FIX ± SD

p_np

0.036 ± 0.152 (18)

0.036 ± 0.157 (18)

0.991

0.937

0.001 ± 0.212

7/F

−0.132 ± 0.357 (14)

0.036 ± 0.348 (14)

0.236

0.346

0.168 ± 0.513

3

11/F

−0.065 ± 0.141 (12)

0.777

0.696

0.019 ± 0.198

4

11/M

−0.072 ± 0.103 (12)

−0.066 ± 0.122 (16)

0.903

0.798

0.006 ± 0.172

5

12/M

−0.096 ± 0.271 (12)

−0.071 ± 0.245 (13)

0.816

0.532

0.025 ± 0.382

6

5/M

0.072 ± 0.116 (11)

0.074 ± 0.156 (10)

0.979

0.916

0.002 ± 0.193

7

6/M

−0.035 ± 0.199 (18)

−0.065 ± 0.142 (14)

0.645

0.459

0.030 ± 0.217

8

11/F

−0.049 ± 0.127 (18)

−0.038 ± 0.140 (19)

0.816

0.891

−0.043 ± 0.183 (14.8)

−0.022 ± 0.183 (14.1)

Avg

←0.084 ± 0.149 (9)

0.011 ± 0.201 0.020 ± 0.267

Table 2. Responses to vertically drifting images of stereodeficient subjects (y: Vertical eye displacement in 80-160ms time window (deg); N: number of correct trials; p: unpaired t-test; p_np: Mann-Whitney U test) Subj

Age/Sex

y ± SD (N) UP

y ± SD (N) DW

p_np

OFR ± SD

1

12/M

0.192 ± 0.119 (18)

−0.116 ± 0.104 (18)