Studies in Skin Perfusion Dynamics: Photoplethysmography and its Applications in Medical Diagnostics (Biological and Medical Physics, Biomedical Engineering) 9811554471, 9789811554476

Citation preview

Biological and Medical Physics, Biomedical Engineering

Vladimir Blazek Jagadeesh Kumar V. Steffen Leonhardt Mandavilli Mukunda Rao   Editors

Studies in Skin Perfusion Dynamics Photoplethysmography and its Applications in Medical Diagnostics

Biological and Medical Physics, Biomedical Engineering

BIOLOGICAL AND MEDICAL PHYSICS, BIOMEDICAL ENGINEERING This series is intended to be comprehensive, covering a broad range of topics important to the study of the physical, chemical and biological sciences. Its goal is to provide scientists and engineers with textbooks, monographs, and reference works to address the growing need for information. The fields of biological and medical physics and biomedical engineering are broad, multidisciplinary and dynamic. They lie at the crossroads of frontier research in physics, biology, chemistry, and medicine. Books in the series emphasize established and emergent areas of science including molecular, membrane, and mathematical biophysics; photosynthetic energy harvesting and conversion; information processing; physical principles of genetics; sensory communications; automata networks, neural networks, and cellular automata. Equally important is coverage of applied aspects of biological and medical physics and biomedical engineering such as molecular electronic components and devices, biosensors, medicine, imaging, physical principles of renewable energy production, advanced prostheses, and environmental control and engineering. Editor-in-Chief Bernard S. Gerstman, Department of Physics, Florida International University, Miami, FL, USA Series Editors Masuo Aizawa, Tokyo Institute Technology, Tokyo, Japan Robert H. Austin, Princeton, NJ, USA James Barber, Wolfson Laboratories, Imperial College of Science Technology, London, UK Howard C. Berg, Cambridge, MA, USA Robert Callender, Department of Biochemistry, Albert Einstein College of Medicine, Bronx, NY, USA George Feher, Department of Physics, University of California, San Diego, La Jolla, CA, USA Hans Frauenfelder, Los Alamos, NM, USA Ivar Giaever, Rensselaer Polytechnic Institute, Troy, NY, USA Pierre Joliot, Institute de Biologie Physico-Chimique, Fondation Edmond de Rothschild, Paris, France Lajos Keszthelyi, Biological Research Center, Hungarian Academy of Sciences, Szeged, Hungary Paul W. King, Biosciences Center and Photobiology, National Renewable Energy Laboratory, Lakewood, CO, USA Gianluca Lazzi, University of Utah, Salt Lake City, UT, USA Aaron Lewis, Department of Applied Physics, Hebrew University, Jerusalem, Israel

Xiang Yang Liu, Department of Physics, Faculty of Sciences, National University of Singapore, Singapore, Singapore David Mauzerall, Rockefeller University, New York, NY, USA Eugenie V. Mielczarek, Department of Physics and Astronomy, George Mason University, Fairfax, USA Markolf Niemz, Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany V. Adrian Parsegian, Physical Science Laboratory, National Institutes of Health, Bethesda, MD, USA Linda S. Powers, University of Arizona, Tucson, AZ, USA Earl W. Prohofsky, Department of Physics, Purdue University, West Lafayette, IN, USA Tatiana K. Rostovtseva, NICHD, National Institutes of Health, Bethesda, MD, USA Andrew Rubin, Department of Biophysics, Moscow State University, Moscow, Russia Michael Seibert, National Renewable Energy Laboratory, Golden, CO, USA Nongjian Tao, Biodesign Center for Bioelectronics, Arizona State University, Tempe, AZ, USA David Thomas, Department of Biochemistry, University of Minnesota Medical School, Minneapolis, MN, USA

Stuart M. Lindsay, Department of Physics and Astronomy, Arizona State University, Tempe, AZ, USA

More information about this series at http://www.springer.com/series/3740

Vladimir Blazek Jagadeesh Kumar V. Steffen Leonhardt Mandavilli Mukunda Rao •

•

•

Editors

Studies in Skin Perfusion Dynamics Photoplethysmography and its Applications in Medical Diagnostics

123

Editors Vladimir Blazek Medical Information Technology (MedIT) Helmholtz Institute for Biomedical Engineering RWTH Aachen University Aachen, Germany Steffen Leonhardt Medical Information Technology (MedIT) Helmholtz Institute for Biomedical Engineering RWTH Aachen University Aachen, Germany

Jagadeesh Kumar V. Department of Electrical Engineering Indian Institute of Technology Madras Chennai, Tamil Nadu, India Mandavilli Mukunda Rao Department of Physics Sri Sathya Sai Institute of Higher Learn Puttaparthi, Andhra Pradesh, India

ISSN 1618-7210 ISSN 2197-5647 (electronic) Biological and Medical Physics, Biomedical Engineering ISBN 978-981-15-5447-6 ISBN 978-981-15-5449-0 (eBook) https://doi.org/10.1007/978-981-15-5449-0 © Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are reserved 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 Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

As technology advances, novel methods of measurements become feasible. These methods infuse further understanding of the underlying principles of operation and behavior of systems. Comprehension of the principles of operation and behavior of the system further accelerates discovery of novel methods of measurement and fuels further advancements. This is more evident in the field of medical electronics. This book deals with the development of new non-invasive methods of measurement and the underlying physiological aspects of blood perfusion in extremities such as skin, earlobe, and fingers. Editorial team and the authors of the book collaborated to unravel the novel methods of determination of blood perfusion dynamics and unravels the relation between the measured (quantified) results and the physiological phenomena. The results outlined in this book elucidate the decade-long synergetic research carried by the authors as well as research undertaken by them in their individual capacity. The authors strongly feel that the book will be engrossing to readers who wish to understand blood perfusion dynamics and its relation to the underlying physiological phenomena. Equally, this book provides deep insights to the readers who are involved in research leading to advance the noninvasive optical measurements of skin perfusion dynamics using photoplethysmography (PPG) and photoplethysmography imaging (PPGI) and its relation to not only the physiological but also psychological activities. Authors are indebted to their respective organizations and funding agencies for supporting the research activities whose outcomes culminated as this book. Special thanks are due to the (i) Federal Ministry of Education and Research (BMBF), Germany, (ii) Alexander von Humboldt Foundation (AvH), Germany, (iii) German Academic Exchange Service (DAAD), DFG (German Research Foundation) and (iv) Department of Biotechnology (DBT), India. Happy reading! Editorial team and authors

v

Acknowledgements

The results presented here were originally initialized within the framework of the Indo-German project entitled “Studies of Neurological Induced Skin Perfusion Change Using Optical Sensors” which was jointly sponsored by the Bundesministerium für Bildung und Forschung (BMBF) through Deutsche Gesellschaft für Luft und Raumfahrt (DLR) in Germany and the Indian Institute of Technology at Chennai in India. Since this time, studies of skin perfusion phenomena have been in the focus of close research cooperation between RWTH Aachen University, Czech Technical University in Prague, IITM in Chennai and other institutions. Over time, different research activities and stays of project participants were partially sponsored by Ministry of Education, Youth and Sports of the Czech Republic (MSM grant 6840770012 “Transdisciplinary Research in the Area of Biomedical Engineering II”, CTU Prague). Furthermore, we acknowledge generous support by the German BMBF and BMWi (Bundesministerium für Wirtschaft und Energie) within the projects “In-Monit”, “Ohr-Biofeed-back” and “LAVIMO” as well as additional funding from the German institutions DAAD, and Alexander von Humboldt-Stiftung (AvH). All authors who have benefited from this support would like to express their sincere thanks to these institutions. Thanks are due to the Centre for Continuing Education, IIT Madras in Chennai for providing seed money in bringing out this book. The authors are also indebted to Springer Nature for their support in the publication of this book. Last but not the least, the editors say thanks to all the individual authors of different chapters and scientific co-workers in their groups and laboratories, graduate and doctoral students for their engagement and collaboration.

vii

Introduction

Aims and Results of Indo-European Research Activities “Studies of Neurological Induced Skin Perfusion Dynamics” In recent years, the investigation and understanding of the interaction between tissue perfusion, brain activity and human hemodynamic are receiving much attention. Such investigations have been generally confined to few premier research institutions where expensive and sophisticated facilities like PET, MRI, etc., are available. However, recent advancements in optoelectronics and computer technology have accelerated the development of new measuring systems and methodologies for use in this medical field. Specifically, transcutaneous assessment of skin perfusion changes (blood volumetric measurement) through optical-sensor-based photoplethysmography (PPG) has rapidly gained an important role. This basically non-invasive measuring procedure is devoid of harmful radiation and ionizing phenomena, simple in construction and connection to the measuring set-up and is easy to use in all areas of human body. The relatively low cost of these sensors has resulted in their use in various medical fields and clinical applications. In the past, though, the application of PPG had been limited by technical difficulties involving calibration of the data. The development of the quantitative PPG technique based on computer-aided data processing has removed this obstacle. In cooperation between the Indian Institute of Technology at Chennai and RWTH Aachen University, an Indo-German Project was initiated in 1996 for advanced studies in this research area. Measuring system designs, experimental details and some preliminary results obtained within the framework of this project are presented in this book. From the investigations carried out so far using the PPG sensors in conjunction with breathing sensors, it has been possible to monitor the 0.1–0.15 Hz rhythms in the arterial blood volumetric changes and to study the influence of breathing on them. These rhythms, which according to medical experts have relevance to psychosomatic conditions like stress or relaxation, can also be addressed to quantify the benefits or lack of ancient Indian practices like yoga and meditation. Using the PPG

ix

x

Introduction

techniques mentioned above, a first Indo-European project was initiated in 1996 for studying the effects of Indian relaxation techniques like pranayama, meditation, etc., in comparison with Western relaxation techniques like autogenic training. Many of the results obtained so far in this project have been presented at several international conferences and published in their proceedings. As a consequence, we now can constate that the Indian techniques of relaxation like yoga and meditation are very effective in generating low-frequency rhythms in the skin perfusion as monitored by the optical sensors. According to medical experts, these low-frequency rhythms have a very important influence on the human physiology and have potential therapeutic implications. We are now in the process of quantifying these rhythms under normal and controlled conditions with the help of data obtained from many subjects. We hope that these measurements will scientifically validate the efficacy of the ancient Indian relaxation techniques like yoga and meditation. The succeeding chapters, in full or part, describe the results of the research works carried out by the respective authors. Most of these results have been published by the authors themselves in different scientific forums. The aim of bringing all these results together under one unifying umbrella of this book is to provide an overall glimpse of the commonality in the techniques used for different and very much differing medical applications. Moreover, the value of some chapters should be seen in the historical context to understand the broad experimental and appreciate the innate clinical relevance of photoplethysmography. However, it should also be noted here that some chapters provide new results or extend the previously proven PPG applications to new, mesmerizing measurements. The authors hope that the next generation of scientists/engineers will carry this work forward in the years to come so that the general public, both in the east and in the west, will benefit from these practices in their lives. Mandavilli Mukunda Rao Jagadeesh Kumar V. Vladimir Blazek Steffen Leonhardt

Contents

1

2

3

4

5

Skin Perfusion Studies: Historical Notes and Modern Measuring Strategies Using Non-invasive Photoplethysmographic Sensor Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vladimir Blazek

1

Influence of Controlled Breathing (Pranayama) on Dermal Perfusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mandavilli Mukunda Rao

33

Pulse Oximetry for the Measurement of Oxygen Saturation in Arterial Blood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jagadeesh Kumar V. and K. Ashoka Reddy

51

Reflective Arterial Pulse Oximetry for New Measuring Sites and Long-Term Assessment of Dermal Perfusion . . . . . . . . . . . . . . Boudewijn Venema

79

Peripheral Venous Dynamics, Venous Oxygen Saturation and Local Oxygen Consumption Measured with an Extended Photoplethysmograpic Muscle Pump Test . . . . . . . . . . . . . . . . . . . . Vladimir Blazek and Claudia Blazek

93

6

Low-Frequency Blood Volume Rhythms in the Skin Perfusion Obtained by Optical Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Hans J. Schmitt

7

Synergetic Interpretation of Patterned Vasomotion Activity in Microvascular Perfusion: Application to Objective Recording of Subjective Responses to Pain . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Holger Schmid-Schönbein

8

A Self-Organized Rhythm in Peripheral Effectors: The Intermediary Rhythm Appears as 0.15 Hz-Band Activity . . . . 139 Volker Perlitz

xi

xii

9

Contents

Analyzing Pain and Stress from PPG Perfusion Signal Patterns . . . 151 Marcus Koeny

10 Photon-Tissue Interaction Modelled by Monte Carlo Method for Optimizing Optoelectronic Sensor Concepts . . . . . . . . . . . . . . . 163 Markus Hülsbusch and Vladimir Blazek 11 Remote Space- and Time-Resolved Skin Perfusion Detection Using Photoplethysmography Imaging . . . . . . . . . . . . . . . . . . . . . . 177 Nikolai Blanik 12 Selected Clinical Applications of Functional PPGI Perfusion Mapping in Dermatology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 Claudia Blazek and Vladimir Blazek 13 Concluding Remarks and New Horizons in Skin Perfusion Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Steffen Leonhardt Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

Editors and Contributors

About the Editors Dr. Vladimir Blazek is Professor Emeritus for Measurement Techniques and Senior Advisor at the Chair for Medical Information Technology, Helmholtz-Institute for Biomedical Engineering, RWTH Aachen University, Germany. His areas of research interest include optical communication technology, optoelectronics in medicine, biomedical sensors, tissue optics, photon-biotissue interaction, modelling and simulation of human hemodynamics, optical imaging, 2D/3D measuring concepts & functional signal analysis. He actively works in European integration & co-operation in the field of university education and is the Rector’s delegate for university cooperation between Aachen and Prague. He has over 300 publications in books, journals and conference proceedings, and close to 50 patents and patent registrations in Germany and abroad. Since 2014 is Dr. Blazek also a member of the Czech Institute of Informatics, Robotics and Cybernetics (CIIRC), established at the Czech Technical University (CTU) in Prague, Czech Republic, in 2013. Dr. Jagadeesh Kumar V. received his B.E. Degree in Electronics and Telecommunication Engineering from the University of Madras in 1978, and his M. Tech. and Ph.D degrees in Electrical Engineering from the Indian Institute of Technology Madras, in 1980 and 1986, respectively. He is presently a faculty member in the department of Electrical Engineering and dean at IIT Madras. He was a BOYSCAST Fellow at King’s College London during 1987-88. He was a DAAD fellow at the Technical University of Braunschweig, Germany during 1997. He worked as a Visiting Scientist at the RWTH Aachen University, Germany during 1999, 2010, 2013 and 2019. He holds two patents and has published 30 papers in international journals and presented papers at several conferences. His teaching and research interests are in the areas of Measurements, Instrumentation and Signal Processing.

xiii

xiv

Editors and Contributors

Dr. Steffen Leonhardt received his M.S. in Computer Engineering from the University of Buffalo (SUNY), NY, the Dipl.-Ing. in Electrical Engineering and the Dr.-Ing. degree in Control Engineering from Technical University of Darmstadt, Darmstadt, Germany, and the M.D. degree in Medicine from J. W. Goethe University, Frankfurt, Germany. Prior to returning to academia, he worked for almost 5 years at the R&D department of Dräger Medical GmbH Co KGaA, Lübeck, Germany. In 2003, he was appointed Full Professor and Head of the Philips endowed Chair for Medical Information Technology at the HelmholtzInstitute for Biomedical Engineering, RWTH Aachen University, Germany. In 2018, he received a honoray doctorate degree from Czech Technical University in Prague, Czech Republic, and was appointed Distinguished Professor at the Indian Institute of Technology Madras. Dr. Leonhardt´s research interests include physiological measurement techniques, personal health care systems, cyber-medical systems and feedback control systems in medicine. Dr. Mandavilli Mukunda Rao was a Post-doctoral Fellow at the University of Illinois, USA. He is a retired professor of of Electrical Engineering at the Indian Institute of Technology Madras where he was teaching in the area of Laser and Optical Communication. He has published a book on optical communication and several research papers in reputed journals. His research interests include Optical Non-invasive Diagnostics. He spent several years at the RWTH Aachen University in Germany, first as DAAD Fellow (1970-1972) and then as Alexander von Humboldt Fellow (1978-1979). He is currently an Honorary Research Professor at Sri Ramachandra University, Chennai.

Contributors Nikolai Blanik Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany Claudia Blazek The Private Clinic of Dermatology, Haut im Zentrum, Zurich, Switzerland Vladimir Blazek Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany Markus Hülsbusch Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany Marcus Koeny Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany Jagadeesh Kumar V. Indian Institute of Technology Madras, Chennai, India

Editors and Contributors

xv

Steffen Leonhardt Chair of Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany Mandavilli Mukunda Rao Indian Institute of Technology Madras, Chennai, India Volker Perlitz Simplana GmbH, Aachen, Germany K. Ashoka Reddy Indian Institute of Technology Madras, Chennai, India Holger Schmid-Schönbein Department of Physiology, RWTH Aachen University Hospital, Aachen, Germany Hans J. Schmitt Institute for High Frequency Technology, RWTH Aachen University, Aachen, Germany Boudewijn Venema Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany

Chapter 1

Skin Perfusion Studies: Historical Notes and Modern Measuring Strategies Using Non-invasive Photoplethysmographic Sensor Concepts Vladimir Blazek Abstract For adequate skin perfusion rhythmicity assessment, sensor technology must be used, which fulfills basic demands such as unobtrusiveness and continuous monitoring with spatial resolution photoplethysmography (PPG) and its modern camera-based imaging variant (PPGI) ideally meet these requirements, and they are generally accepted in the field of noninvasive medical diagnostics. PPG can work in reflective or transmissive mode and detects blood volume changes in the vascular plexus within the region of transilluminated tissue. The PPG and PPGI signals comprise a complex of pulsatile wave formations (AC) associated with cardiac, respiratory, and different nervous system activities, which are superimposed to a non-pulsatile baseline (DC) due to optical damping of bloodless tissue. Although most of the PPG applications in medical diagnostics are today devoted to recording cardiac rhythmicity, the strength and future of PPG lie in detecting the distributed and, in some cases, highly autonomous skin perfusion dynamics below the frequencies of the “central oscillators”, namely the heartbeat and breathing. This chapter presents selected activities and results of the bilateral and interdisciplinary longterm cooperation between IIT Madras in Chennai and RWTH Aachen University in Aachen in this exciting research field of the dermal perfusion dynamics.

1.1 Introduction The Englishman William Harvey (1578–1657) was the first physician to correctly describe the dynamics of blood circulation after many false interpretations and working hypotheses presented by others. Nevertheless, he considerably underestimated the pumping function of the heart. He assumed that half an ounce (18 g) of blood is normally pumped per minute by the left ventricle; he then multiplied this amount by 1000 heartbeats per half an hour and concluded that 1000 oz of blood has V. Blazek (B) Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_1

1

2

V. Blazek

to be pumped in one hour. It appeared logical to him that such a large amount of blood could not be regenerated in any way within our body—an idea based on the erroneous theories of blood circulation proposed by Galen of Pergamon (physician/philosopher c. AD 129 to c. 201) that were popular at that time. Nevertheless, he forged ahead and gained valuable insight into the closed human circulatory system based on simple quantitative examination. Today, we know that the driving force of our blood circulation is based on the difference in pressure between the arteries and veins. The pulsed high pressure in the arteries funnels the blood until the blood pressure in the veins reaches a steady-state low pressure. The human vascular system (with an integral length of hundreds of kilometers or more) circulates about 6 L of blood through our body in just one minute, powered by our heart (weighing only 350 g) acting as a pump and impulse generator. The transportation capacity of such an extensive vascular network (without losses) is about 10,000 L per day. The phenomenon of rhythmic fluctuation due to arterial blood pressure was experimentally discovered in the eighteenth century. In 1726, the Reverend Stephen Hales (1677–1761), an English clergyman who pioneered quantitative experimentation in plant and animal physiology was the first to observe the magnitude of arterial blood pressure and its pulsation in an invasive manner. Figure 1.1 illustrates the experiment he conducted to determine the arterial blood pressure of a horse. After the first continuous recording of blood pressure, an extensive series of investigations have taken place, all dealing with the problem of how to explain the rhythmic fluctuation of blood flow. However, even until now, a complete understanding of the underlying mechanisms that control the formation of such rhythms is still lacking. A practical, non-invasive way to acquire information on peripheral venous and/or arterial hemodynamics is by use of optoelectronics paired with quantitative photoplethysmography (PPG). An optoelectronic biosensor concept is introduced that is capable of identifying rhythms from several sensors in combination with acquisition of data on, for example, respiration, ECG and body movement. Depending on the area measured and the purpose of measurement, the optoelectronic sensor can be used in either reflection or transmission mode. The data can then be analyzed using high-time resolution and displayed in time and frequency domains [3]. This chapter presents various possibilities related to the PPG measuring concept by means of examples and perfusion protocols. Current research focuses not only on the so-called central rhythms and their correlation with heartbeat and respiratory rate, but also on the perfusion frequency range around 0.1 Hz. However, assessment and interpretation of these perfusion rhythms are often hindered by the fact that these patterns have a very strong spatial variability and are highly transient [4]. Nowadays, new information about the rhythmic phenomena in skin perfusion is available based on the relationship between blood volume rhythms and respiratory dynamics, evaluation of the hemodynamic-related effect of autogenous training, analysis of the pulse waveform parameters, and/or pulse wave transient time. Using new camera-based sensor and signal processing strategies, a recently developed photoplethysmographic setup allows contactless measurements of cutaneous perfusion with spatial resolution.

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

3

Fig. 1.1 A spectacular observation by Hales, measuring arterial blood pressure by means of a glass tube placed in the carotid artery of a horse lying down [1, 2]

1.2 Rhythmic Phenomena in Dermal Perfusion: A Brief Introduction Before discussing our experimental results and PPG findings, we present a brief overview of rhythmical skin perfusion phenomena based on physiologically-related publications. To this chapter makes no claim to completeness, but seeks only to classify these phenomena and thus facilitate the understanding of the perfusion time series and graphs portrayed in the following chapters. When human skin is irradiated with infrared light, a large proportion of the photons injected into the tissue is scattered or reflected, and the remaining part is absorbed. If the photons interact with haemoglobin (red blood cells), then the absorption increases. Since the blood volume in the skin is not constant, the amount of reflected light varies inversely with the abundance of blood in the irradiated part of the skin. Thus, variations, e.g., vascular cross-section, cause the rhythmic abundance of blood, making these rhythms particularly easy to find and important for diagnostic

4

V. Blazek

Fig. 1.2 Schematic presentation of some typical rhythmical patterns in the skin perfusion (modified after [5] and H. Schmid-Schönbein, Aachen, personal communication)

purposes. These are the following well-known rhythmical fluctuations [5–18], (see also Fig. 1.2).

1.2.1 Heart Rate (Pulse-Synchronous Rhythms) During a heartbeat, changes of blood pressure and blood circulation occur below the skin. The observed/imaged waveform changes with the distance from the heart. The arteries in the heart’s vicinity are distensible; they accumulate blood during the contraction of the heart and dispense it again over the entire period. This is called the “Windkessel effect”. The resting frequency of 1–1.2 Hz is significantly higher than the other investigated fluctuations. Moreover, in the air-chamber integer multiples of the heartbeat—the harmonics— are generated by reflections. Above these frequencies, the measurement only picks up disturbances which must be filtered out.

1.2.2 Breathing Frequency (Respiratory-Synchronous Rhythms) Many large vessels, arteries and veins lie in the vicinity of the lungs. The lungs rhythmically press on this system with a pressure of a few mmHg. This affects the arterial system with a central pressure of 100 mmHg only minimally. But on the veins, with a mean pressure of 1 mmHg to 4 mmHg this breathing influence is experienced. After each inhalation and during diastole, the venous system pumps more blood back into the right ventricle than in the exhaled state in order to increase the pumping and the arterial blood pressure. Strength and shape of the blood pressure will be

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

5

influenced by the form of respiration. One distinguishes between thoracic breathing, diaphragmatic breathing, and mixed forms of both. Another important factor is that through the respiration-induced effect known as respiratory sinus arrhythmia (RSA), occasional phase jumps in the heart rate are noticed, which always have a certain phase relationship with the respiration rate. The exact sequence is unclear, but there seems to be a synchronization mechanism between respiration and heart activity. Finally, there exist vasomotor effects (vascular muscle movements) synchronous to breathing, which are driven by synchronized neuronal activity. It would therefore be conceivable that the vasomotor centre in the brain is connected through the respiration centre. However, it is not yet clear if both these alleged centres could be driven by nerve impulses from the same region of the brain stem. The respiratory rate at rest is usually at 0.2 Hz to 0.4 Hz (in the frequency range of the middle PPG spectrum (see) in the logarithmic scale).

1.2.3 Vasomotor Rhythms The investigation of the vasomotor system really began with the work of French physiologist Claude Bernard (1813–1878), who demonstrated that a section of the sympathetic nervous system in a rabbit caused dilatation of the vessels of ear and that simulation of the peripheral cut end of the nerve caused constriction [6, 7]. His famous quotation may be worth noting in this context: I consider the hospital the antechamber of medicine; it is the first place where the physician makes his observations. But the laboratory is the temple of the science of medicine

Each human organ is perfused with oxygenated blood through a capillary vascular network. This takes care of the nutrient intake, the residual material dissipation and temperature regulation. The structure is as follows: branching off from the arteries are the arterioles (diameter see). From this point on, one speaks of “microcirculation”. The venules collect the blood at the opposite ends of the microcirculation system and pass it back into the veins. Between arterioles and venules are two types of connections: the arteriovenous anastomosis, a short circuit which is closed when required, and the metarteriole, the main blood-flow path through the capillary network. At the junctions are ring muscles (sphincter precapillaris), which regulate blood flow through the capillaries. The walls of the capillaries consist of a single cell-layer separating the blood vessel (intravascular layer) from the cell gap (interstitium). The pressure decreases about 50% in the arterioles. The vessel diameter is therefore well-regulated. According to Hagen-Poiseuille’s law, the resistance R of a tube like an artery is defined as [17]: R=

8·v·l , with: v = viscosity, l = length and r = radius of the artery. (1.1) π · r4

6

V. Blazek

This mechanism regulates the blood volume flow in the vascular plexus. Arterioles and precapillary sphincter are considered to be resistance vessels. Several different control mechanisms exist, which are all interrelated and together affect the blood distribution: • neural (nervous) influence (from the brain, almost exclusively based on sympathetic activity), • systemic humoral influence (control by hormones), • local metabolic influence, • hormonal influence (due to inflammation or allergies), • self-regulation (mainly in the brain and in the kidneys), • influence through local chemo-receptors (in skeletal muscles, with the nerves, then signaling the blood distribution). Skin perfusion is also supported by influences from the temperature controlloop. The local mechanisms are mostly more dominant than global mechanisms. The interrelation of the above-mentioned mechanisms is currently not clear: they all together, maybe in different ways and intensities contribute to the generation of the vasomotor rhythms. Especially waves with a duration of 5–7/min (fast wave types of third order, also called Traube-Hering-Mayer (THM) waves) are prevalent in blood pressure recordings, while waves with a periodic length of about 1/min are dominant in peripheral perfusion (skin, muscle). These dominant rhythms are called 10 s- and 1min-rhythms according to their preferred frequency. The 10 s- and the breathing-rhythm show coupling in form of relative coordination according to Golenhofen and Hildebrand [5]. The mechanism of this coupling is still debated after many years of research. Since the 1 min-rhythm of dermal perfusion is characterized by compensation of blood-content variations in connected vessel districts nearby, the central pressure fluctuations probably don’t play a part in these rhythms. The frequencies of all these fluctuation patterns are usually at see and are at the bottom of a PPG signal spectrum.

1.2.4 Perfusion Rhythms and the Vegetative Nervous System All major vascular and organ functions are regulated and controlled by the autonomic nervous system. This consists of two parts, the sympathetic and the parasympathetic. They branch out from the spinal canal at the top of the chest and are regulated by the medullary control centres in the brain. The sympathetic nervous system drives the body and accelerates body activities. Parasympathetic activity inhibits body activity. During physical exercise, the parasympathetic nervous system adjusts itself more quickly to the new situation compared to the sympathetic and the organism experiences oxygen- and nutrient-deficiency; this is moderated by reserves in the blood. The above-mentioned rhythms are mainly triggered by sympathetic nerves. During anaesthesia, the decaying sympathetic becomes active first. The previously

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

7

made measurements on subjects who are under anaesthesia or just awakening as well as from anaesthetized animals are therefore not meaningful. Since the frequency spectrum and probably also the operating and control parameters of all these loops overlap, causal investigations hardly make sense. They should include the entire contiguous vegetative scheme. Many vegetative rhythms fluctuate in frequency, if they are assessed over long periods of time. Mostly, they are synchronized with each other at integer harmonics.

1.2.5 Rhythmic Perfusion Patterns in Peripheral Venous Network Related to Active Body Exercise The main application of photoplethysmography to map venous blood flow has been en vogue for more than 30 years and provides functional evaluation of total blood displacement in the venous system of the upper extremities. To obtain a PPG for venous blood flow analysis, the PPG sensor is fixed about 10 cm above the inner ankle using a double adhesive ring. The seated patient performs the classical, worldwide standardized muscle pump test (MPT): 8 dorsal extensions in 16 s [19–24]. In an adult with healthy leg veins, the venous blood pool that has been pumped by the dorsal flexions uphill, back to the heart will not flow back due to of the venous valves. As a result the venous blood volume in the foot/leg part reduces which causes the PPG signal to rise. From the venodynamic behavior during such active MPT exercise, several functional parameters like venous refilling time and venous drainage can be calculated (see Chap. 5).

1.3 Photoplethysmography—Historical Notes For many years, classical photoplethysmography has been one of the most popular non-invasive methods for functional monitoring of peripheral vascular (venous and/or arterial) status. After ground-breaking works by Cartwright [25], Haxthausen, Mathes [26], and Molitor et al. [27], in 1937 Alrick B. Hertzman (1898–1991), a physiologist at St. Louis University School of Medicine, discovered the relationship between the intensity of backscattered polychromatic light and blood volume in the skin [28]. The instruments consisted of three essential components which are still essential in modern systems: a light source, a light detector and a registration unit. He called the device Photoelectric Plethysmograph and summarized his findings as follows ([29], p. 336) (Fig. 1.3): The volume pulse of the skin as an indicator of the state of the skin circulation at rest Amplitude of volume pulse as a measure of the blood supply of the skin

8

V. Blazek

Fig. 1.3 Hertzman´s photoplethysmographic sensor with polychromatic light source, optical filter and photocell detector positioned over the skin of the hand [29]

Following these discoveries, many enhancements of the basic measuring principle have been developed: however, these could only make their way into daily usage in clinics and cardiovascular labs because of the continuous rapid developments in photonic components and microprocessor technology.

1.4 Photoplethysmography—Biophysical Fundamentals In the skin, induced infrared light is most strongly absorbed by blood, particularly by its hemoglobin content, and not by surrounding skin tissues (Fig. 1.4). Therefore, the amount of reflected infrared light increases in the measurement window when the surface area of blood vessels decreases. A reduction of blood vessel surface area is caused by decreased blood volume [3, 20]. It follows that the PPG signal, which detects the amount of reflected infrared light, displays these changes in blood volume in the cutaneous and partially the subcutaneous vessel plexus. In addition to the very small, periodically changing arterial signal, the sensor signal consists of a high constant part (light scattering in tissue) and a quasi-static vein signal (Fig. 1.5). It can clearly be seen from Fig. 1.5 that the PPG signal is firstly composed of a very large static component that is due to a large part of measuring light passing only through skin, tissue, or bone (without interaction with blood vessels). The second biggest part of the detected light is attenuated/modulated by venous blood volume changes in the transilluminated tissue volume. This component varies slowly due to respiration, vasomotor and vasoconstriction activity, and also thermoregulation [30–34]. The smallest PPG signal component is proportional to the number of photons passing arterial and terminal micro-vessels; this component will mainly possess peripheral blood volume pulse dictated by the heart beat (central oscillator in

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

9

Fig. 1.4 Typical optical properties of human skin and blood in the visible and near-infrared area of the spectrum (left). The diagram shows the reflection spectra of anaemic skin in vivo, a 0.3 mm epidermal layer and of a 0.12 mm thick blood layer on glass. The difference in reflectivity between tissue and blood leads to a high optical contrast between skin and dermal vessel plexus. The optical attenuation of epidermis is lowest at wavelengths of about 930 nm and increases with decreasing wavelengths. Illustration of a transilluminated skin vessel plexus in the measurement area explaining the correlation between blood volume and PPG signal (right), modified after [20]

Fig. 1.5 Synthesis of the rPPG signal. The intensity of the backscattered light (subsurface reflection) encodes the PPG signal and depends partially on the blood volume in arterial and venous vessels in the measuring zone. A separation of the venous and arterial perfusion components are possible by selective post-processing of the signal, modified after [21]

10

V. Blazek

Fig. 1.6 Typical PPG sensors in reflection (left) and transmission (right) mode. Each sensor consists in its minimal configuration of one light source (LED) and one light detector (Si photodiode). Pulse oximetric sensors, a sub group of PPG sensors, consists of mostly two selective light sources and one detector, which is sensitive for both working wavelengths

our cardiovascular system). Last two PPG signal components depend on (possible) illumination artefacts (e.g., ambient light coupling into the sensor) and quantization noise in the sensor close A/D signal conversion: SPPG (x, y, t) = ST (x, y) + SV (x, y, t) + S A (x, y, t) + Serr (x, y, t) + Sqn (x, y, t)

(1.2)

Examples of early PPG sensors working in reflective (rPPG) or transmitive (tPPG) mode is shown in Fig. 1.6.

1.5 Detecting Light Attenuation Changes in Biotissue as a Function of Blood Volume Biological tissue is a highly scattering and non-homogenous material concerning electromagnetic radiation at frequencies of about 300 THz (near-infrared). Mainly the spreading of photons with this energy content inside the tissue is of high interest in therapeutic and diagnostic applications of medical optoelectronics [35]. A typical skin cross-section is shown in Fig. 1.7a. When optical radiation is sent into the tissue, some photons are reflected directly at the skin surface (Fig. 1.7b), another fraction will be distributed in the tissue through absorption or scattering, while the remaining photons will travel into deeper layers, either straight through (ballistic photons) or after a number of scattering collisions [20, 35]. Typical values for the absorption and scattering coefficients range from 0.05 mm−1 to 0.15 mm−1 (µa ) and 3 mm−1 to 10 mm−1 (µs ) in the near-infrared wavelength range (800 nm to 1000 nm) in skin tissue. Monte Carlo simulations show that the free photon path between two collisions is around 0.24 mm and the mean scattering-to-absorption probability ratio approximately 50 [36, 37].

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

11

Fig. 1.7 Light propagation in dermal tissue. a Schematic structure of skin perfusion. 1: epidermis, 2: capillaries, 3: plexus superficialis, 4: vasa communicantia, 5: plexus profundus, 6: subcutis. b Possible photon paths in tissue. c Calculated relative sensitivity of a reflective optical sensor as a function of skin depth (modified after [20] and [35])

Therefore, the optical attenuation in skin can be calculated to about 7000 dB/m. Because of this, detector and light source of optical sensors are often placed in a single encasement next to each other on the skin surface. These sensors work in reflection mode. Transmission mode sensors is only used at fingertips or earlobes, where the distance between source and detector is not too large [35]. The effective measurement depth of the reflective PPG sensors as well as its sensitivity can be adjusted—besides by changing the wavelength—by varying its geometry. Distance (a) and axis alignment of both components as well as the beam angles of the opening (numerical aperture NA) affect these characteristics. For example a typical rPPG sensor (900 nm wavelength) with a = 6 mm, right-angled positioning and NA = 0.09 (α = ± 5°), has its main detection area between 0.1 mm and 3.1 mm skin depth [20, 35]. (decrease of maximum sensitivity to 1/e, Fig. 1.7c). The resulting measurement volume is about 100 mm3 . In this case, only around 120 photons per million reach the detector and can be used for further signal processing. A sensor sensitivity profile can be also calculated when the light intensity at different depths locations are regarded. I (z 0 ) = S(z 0 ) = Iges

 ymax  xmax −ymax

−xmax Iq (x,

y, z 0 ) · Id (x, y, z 0 ) · dx · dy · dz Iges

.

(1.3)

Last but not least, the knowledge of the basic optical skin parameters (absorption coefficient µa (λ), scattering coefficient µs (λ) and anisotropy factor g(λ)) makes the determination of the light penetration depth in tissue possible (see Chap. 9).

12

V. Blazek

1.6 Comparing Measured Arterial rPPG and tPPG Signal Waveform with Blood Pressure Waveform, Generated with Arterial Tree Model in the Finger Tip A computer model of hemodynamics in the human vascular system has been developed at the RWTH Aachen University in 1993 [38, 39] to investigate blood pressure, blood flow and blood volume waveforms in major branches of the arterial tree as well as in smaller digital arteries and arterioles of microvascular networks. The basic units of the simulation model are vascular segments which are considered as thickwalled viscoelastic tubes with fixed coupling to the surrounding tissue. Blood motion in moving elastic tubes is well described by the Navier–Stokes equation, which is equivalent to conservation of momentum, and by the continuity equation, which is equivalent to conservation of mass. A linear approximation Navier–Stokes equation leads to a relation between pressure gradient and flow. With an expression of pressure diameter relation, utilizing the shell theory, hemodynamics is described in completely analogous terms of electrical transmission line theory. The growing algorithm used in the RWTH model of the human vascular tree is based on the assumption that each dichotomous branching fulfils several bifurcation rules while the entire tree fulfils the criterion of minimum blood volume and several boundary conditions formed by anatomical constrains. The model grows successively by adding new vessels to the pre-existing tree; each new vessel is connected to the optimum side with respect to the growth criterion. A detailed model of finger arteries of the index finger was published in [40]. This spherical network consists of 4000 arterial segments. In a study, photoplethysmographic measurements in reflection (rPPG) and transmission (tPPG) mode are compared with peripheral blood pressure waveform. Subsequent to the size of vessels, the calculated blood pressure waveforms are similar to the photoplethysmograms detected in t- and r-sensor mode as shown in Fig. 1.8. These results clearly demonstrate that using the T-mode sensor technology, dominantly microvessel hemodynamics are recorded. With the R-mode PPG sensors, however, more hemodynamics in terminal microvessels are recorded (see Chap. 9). In the evaluation of peripheral arterial pulse shape, it is, therefore, necessary to always declare with which sensor modality the analyzed photoplethysmograms were recorded (see Sect. 1.11.5 in this chapter).

1.7 Principle of Quantitative Photoplethysmography The perfusion signal intensity detected by a PPG invariably depends on the following parameters: • The intensity of the input light, • the wavelength of the input light,

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

13

Fig. 1.8 Simulation study comparing photoplethysmographic measurement of arterial blood volume pulse in the fingertip (left) with blood pressure waveforms in large and small arteries (right), modified after [38]

• the epidermal attenuation, which depends on the individual skin color and texture, • the attenuation that depends on the individual dermal (arterial and/or venous) blood volume (initial and dynamic part), and on • the sensitivity factor of the detector electronics. The variations in these parameters manifest in the detected PPG signal. The amplitude will then become not only dependent on the blood volume sensed by the sensor but also on these parameters. For example, the PPG amplitudes as a function of blood volume for different skin colors are depicted in Fig. 1.9: for identical blood volume change of V, a larger amplitude change is obtained for a person with fair skin compared to the signal obtained from a person with dark color skin. Hence, if

Fig. 1.9 The dependence of non-calibrated PPG signals on the color of the skin and the individual initial blood volume in the assessed tissue compartment. The same blood volume change V produces different PPG signal R (left).Quantitative PPG: the same starting point for each skin type as a result of automatic calibration routines (right), [20]

14

V. Blazek

Fig. 1.10 Block schematic of the hardware to obtain “quantified PPG signals”, [41]

a meaningful measurement on blood volume changes are to be made, then all these dependencies have to be quantified or explicitly determined. A novel strategy of handling these dependencies by way of introducing a simple signal calibration before making a measurement was introduced at RWTH Aachen University in 1986. The method of obtaining quantitative photoplethysmography is depicted in a block schematic form in Fig. 1.10. Here, the PPG sensor is put in a feedback control loop. The input to the control is the detected PPG signal and the control parameter is the power supplied to the light source (it is almost universal now that the light source is implemented with a LED and hence, the control parameter is invariably the current through the LED). Before starting a measurement cycle, the calibration phase is invoked. In the calibration phase, the measured PPG signal is compared to a pre-selected value. If the detected PPG is more than the pre-set value, then the intensity of the source is reduced (by decreasing the LED current). On the other hand, if the received PPG signal is less than the pre-set value, then the intensity of the source is increased. This process is repeated by the controller until the received PPG is equal to the pre-set value. It is already shown by the author that a successive approximation algorithm that starts with half the full-scale intensity as the initial condition achieves the desired final intensity condition in a minimum number of iterations. Once the condition that the detected PPG amplitude is equal to the pre-set value is met, the calibration phase ends and the measurement cycle begins. At this calibrated condition, the received PPG is normalized. If a person possesses a dark skin, then the intensity of the source is increased so that the received PPG has

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

15

the same magnitude as that obtained from a person with normal skin. If the person to be tested has a fair skin, the intensity of the source is reduced suitably so that the received signal is again on par with that obtained from a person with a normal skin. Thus the PPG characteristics if forced to have the variation as depicted in the central loop of Fig. 1.10 and any measurements made on the detected PPG with this condition is independent of the interfering parameters listed above and thus the detected PPG becomes “quantified”.

1.8 “Historical” PPG Multi-Sensor System Designed for the Indo-European Project Our final experimental system is shown in Fig. 1.11. Five PPG channels are used to simultaneously measure different locations of the human body as well as a microprocessor-controlled, non-invasive monitoring of neurologically induced skin perfusion dynamics. Two of the five PPG channels are equipped with interfaces for both standard electrical and special, metal free fibre-optical sensors. Comparative studies and dynamic control during NMR-imaging are possible applications for such sensors. Two other channels are prepared to include recently developed optical sensors with heating facilities. They can locally warm up the skin to increase the blood volume and paralyze local vasomotion of the dermal vascular plexus. The processing of the PPG signals includes an automated calibration process, suppression of other light sources (noise) and filters to distinguish between major frequencies [35]. Furthermore, two breath-monitoring channels (one for each nostril) and an electrocardiograph are applied to the system. All measured bio-signals are transferred

Fig. 1.11 Monitoring skin perfusion dynamics with a non-invasive multi-sensor measuring system, simultaneously reading ECG, temperature, respiration as well as multiple PPG channels [42, 43]

16

V. Blazek

to a computer after A/D conversion to visualize and store these data. With our measurement setup all functional settings and sensor controls can be managed via PC.

1.9 Monitoring of Skin Perfusion Dynamics Under Controlled Conditions in Time and Frequency Domain This kind of post processing of PPG signals was first implemented and published in 1985 with the aim to analyze pulsatile perfusion signals in both time and frequency domain [44, 45]. Our “historical” results are shown in Fig. 1.12. The recording of the time domain (Fig. 1.12a) shows periodical waveforms that are synchronous with heartbeats as well as other low frequent PPG signal fluctuations. These correspond to respiration and other neurological (local or centrally induced) vasomotor activities. Those rhythmical perfusion patterns can also be recognized in the recording of the FFT (Fig. 1.12b). The heartbeat as a “hemodynamic pump” is dominant at a frequency of about 1.1 Hz (66 BPM), and the respiratory frequency is detectable as well (at 0.2 Hz or 12 breaths per minute respectively). Apart from cardiac and respiratory rates, autonomous perfusion changes (vasomotor patterns) can also be observed at frequencies below 0.2 Hz [20, 35]. Some selected results from our Indo-German project “Studies of Neurological Induced Skin Perfusion Studies”, which focused on the endogenous effect of yoga on dermal perfusion, are shown in Fig. 1.13. PPG sensors that were positioned on the forehead and chest of a person practicing a yoga relaxation exercise measured interesting periodical perfusion patterns (also shown in Fig. 1.12). Compared to the PPG signal from the chest, the signal amplitude from the forehead region is larger. This indicates that microcirculation has to be relatively stronger in the forehead. The FFT-analysis reveals a 0.15 Hz rhythm formation in the forehead signal

Fig. 1.12 Skin perfusion rhythms, taken at the right forefinger with a reflective PPG sensor. Registrations 10 min after starting the examination in the time domain (a) and frequency domain (b)

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

17

Fig. 1.13 Investigation of the rhythmical phenomena in the skin. Left: Positioning of the PPG and respiratory sensors by the subject (courtesy of Prof. M. Mukunda Rao, India). Right: Two channel time domain registration and FFT analysis of skin perfusion rhythms, recorded in the forehead and chest regions, modified after [51]

beside the heartbeat (approx. 1 Hz) and breathing frequency (approx. 0.35 Hz). Other research groups [46–50] concluded that these low frequency “relaxation” patterns have an important bearing on human physiology. Furthermore, potential therapeutic implications, e.g., in psychosomatic medicine can be derived from those findings [4, 35] (see Chap. 2).

1.10 Common Hardware and Software Requirements for Optimized Skin Perfusion Monitoring 1.10.1 Design of an “Intelligent” PPG Sensor Interface with “As Soon As Possible” Direct High-Resolution Data Conversion Using different PPG devices in combination with the application of different PPG sensors working in reflection or transmission mode leads to different perfusion pattern characteristics (i.e., different current levels and AC to DC relations). Therefore, commercially available PPG systems do not allow optimized sensor control and perfusion recordings with adequate signal quality [52]. To compensate this, a customized intelligent interface was recently developed in our group at the Philips Chair for Medical Information Technology (MedIT), RWTH Aachen University. Instead of ubiquitous DC compensation and subsequent detection with medium resolution A/D converters, only minimal analog pre-processing and direct conversion with 24-bit resolution can be chosen for this project up to now. An integrated microcontroller controls the multi-channel illumination and reads the data coming from the A/D converter (Fig. 1.14). In addition to multi-channel

18

V. Blazek

Fig. 1.14 Measuring setup and realization of electronic sensor interface with direct high resolution data conversion

PPG recordings, other sensors (such as respiration, temperature) can be integrated to support multi-wavelength detection of blood volume rhythms in skin perfusion. After first pre-processing steps, such as compensation of environmental light, the photoplethysmographic data is sent to a PDA or laptop for further signal processing via a wireless link [52]. To permit long-term application while using a self-reliant power supply, the electronic sensor control is optimized for low current consumption. It draws about 48 mW on average (excluding the Bluetooth link) while performing measurements at 200 Hz/channel and providing LED currents of 60 mA (red) and 15 mA (IR) for example. It is expected that optimizing the utilized illumination wavelength will result in further energy savings [53–58].

1.10.2 Advanced Skin Perfusion Signal Processing and Visualization in Multidimensional Space Advanced signal processing is carried out on a PDA or laptop. Basic algorithms for heart beat detection, SpO2 calculation as well as analysis of the heart rate variability has already been implemented. Further tasks include analysis of slow perfusion rhythms and the assessment of cardiac risk and possible alarm functions [52]. When trying to further analyze the perfusion patterns with the classical FFT, not much new information is revealed. It is possible to recognize differences at low frequencies; however, the resolution is quite limited. The frequency spectrum cannot reveal much-advanced information; the reason being that the Fourier transform is not well suited for the analysis of transient signals. It is not possible to judge only from the power spectrum of a signal if an oscillation is stationary or occurs only during a limited time and at which instance in time. To assess non-stationary characteristics of a signal a joint time-frequency representation of the signal is needed [4]. This problem is illustrated in Fig. 1.15.

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

19

Fig. 1.15 Examples of transient signal analysis. Top: two different synthetic signals. Middle: Fourier Transform of sample signals. Bottom: Wavelet transform of the same signals. The advantage of the multidimensional time-frequency analysis can clearly be seen

The Wavelet transform of a signal leads to a three-dimensional time-frequency representation where the spectral evolution over time can be assessed directly. It is actually a family of transformations where a signal g(t) is transformed by an analyzing function ψ(t). ψ(t), called the “mother wavelet”, can be chosen from a collection of functions. All of these have to meet certain restraints (see [59]), especially its localization in time and frequency domain.

20

V. Blazek

To analyze the signal g(t), the mother wavelet is shifted across the time axis (by parameter b) and also scaled by different factors a. Thus a family of basic functions a,b (t) = |a|

−1/2



t −b ψ a



is obtained. The continuous Wavelet transform is defined as:  ∞ ∗ g(a, ˜ b) = g(t)a,b (t)dt

(1.4)

(1.5)

−∞

Utilizing this transformation one can obtain a higher dimensional representation of the signal g(t), where the dimension b is responsible for the time information and the other dimension a for the scaling information, which is inversely proportional to the frequency. The original function can be recovered from g(t) ˜ by the inverse transform ¨ da db (1.6) g(a, ˜ b)a,b (t) 2 g(t) = Cψ−1 a where the normalizing coefficient Cψ is determined by the shape of the mother wavelet:  ∞ 2 ˆ  (1.7) Cψ = ψ(ω) |ω|−1 dω −∞

(ψˆ designates the Fourier transform of ψ). To fully describe the Wavelet transform, the mother wavelet ψ(t) also has to be specified. An often applied function is the Morlet Wavelet, which is a wave modulated by a Gaussian unit of width (see Fig. 1.16): t2

ψ(t) = e 2 (cos(ω0 t) − i sin(ω0 t)).

(1.8)

The parameter ω determines the time-versus-frequency resolution, the relation  between scaling and frequency becomes f = 2π ω0 a. When using the Morlet wavelet, the resemblance to the windowed Fourier transform becomes apparent. The Gaussian can be interpreted as the windowing function. In contrast to the windowed Fourier transform the width of the function is not fixed, but scaled together with the wave function. So, for every frequency, the same number of oscillations is taken into account, i.e., if we search for slow rhythms of 0.1 Hz, the window function will be ten times wider than if we would search for 1 Hz components. Thus, a very broad frequency range spanning multiple decades with the Wavelet transform is admissible. As a representative example, Figs. 1.17 and 1.18 show typical in-ear-perfusion rhythmicity in the time and Wavelet domains.

1 Skin Perfusion Studies: Historical Notes and Modern Measuring … Fig. 1.16 Morlet mother wavelet consisting of a complex wave modulated by a Gaussian

Fig. 1.17 Typical rPPG recordings monitored in the earlobe (without any hardware filtering)

21

22

V. Blazek

Fig. 1.18 Wavelet Transform of a PPG perfusion signal sequence from the earlobe, additionally the black line is the instantaneous frequency as the inverse of the HRV signal computed by classical detection of maxima in the time signal

Post-processing of the PPG recordings are facilitated by the fact that no readjustment of DC-compensation has to be performed by the microcontroller. Thanks to the wide dynamic range of the implemented electronics used, transmission sensors can also be attached without further hardware or software modification. Our current PPG measuring & data acquisition concept offers the following possibilities: • intelligent front-end sensor concept (using a miniaturized PPG device, without any controls, executable and manageable via the associated software), • long term perfusion studies (also 24/7), • multi-channel and multi-wavelength design, • full signal-recording (recording of the complete PPG signal without signal distortion by the filter in the chain), • 200 measurement values per channel and second, • 24-bit digitalization precision, • future-proof PC-connection via USB, • bus-powered via USB-Bus (no battery needed), • data transfer monitoring via LED signalization, • size: 95 mm (L) × 90 mm (W) × 35 mm (H), weight: 90 g, • excellent price-performance ratio, • on-screen interactive system operation, • simultaneous display of 2 channels, • selection of the port the device has been connected to, • data saving in data files, offering the possibility of reloading, • printable examination protocol, • displayed graphs can be saved as pictures.

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

23

1.11 Recent Developments and Additional Application Fields of PPG Sensing Modalities The analysis of heterogenous rhythmical phenomena in dermal perfusion requires sophisticated assessment strategies [60–62]. It is possible to acquire undistorted vital signals across a broad frequency range using modular multi-sensor concepts and the correlations between the sensors. These sophisticated assessment strategies reveal—alongside known central rhythms—local oscillations around 0.1 Hz, showing endogenous influence ability. Local changes in perfusion patterns can further be assessed by novel optical remote sensing techniques like photoplethysmography imaging (short PPGI; this acronym is a registered trademark since 1997) in a completely contactless manner, while simultaneously providing results with high spatial resolution. Advanced joint-time-frequency signal processing allows the observation and visualization of local slow perfusion rhythms in healthy and/or adjacent skin regions. Signals across a wide frequency range can be examined, including a good resolution at low frequencies. At the same time, the temporal evolution of different frequency components in the skin perfusion dynamics over time can also be revealed (see Chaps. 11 and 12). Some of the current hardware and software developments should be mentioned in the following short review, which will open up new and possibly can revolutionize today´s application fields of the photoplethysmography in functional vascular diagnostics and smart home care.

1.11.1 Distributed Micro Sensor Solutions Multi-body side PPG sensor solutions have been proposed for peripheral vascular disease detection. Comparing arterial pulse recordings simultaneously taken at the right and left ear lobes, index fingers, and great toe sides, the considerable similarity in bilateral body paths under physiological conditions were demonstrated in [63]. This study also described that cross-correlation analysis quantified the degree of similarity in normal subjects compared to dissimilarity in a patient with unilateral arterial diseases. A fascinating platform, integrating a collection of independent sensor types (such as electro-optical, temperature and strain gauge), wireless powering components and components for RF communications, all integrated on a thin and flexible sheet was published in [64]. Such distributed microsensor solutions also can be used for assessment and minimization of motion artefacts in PPG recordings or as a human/machine interface in common body alarm devices.

24

V. Blazek

1.11.2 Biofeedback System Solutions for Clinical, Home, and Outdoor Health Monitoring The demographic changes in most industrialized countries lead to new strategies in the health system. Therefore, all of us are encouraged to assume a more active role into prevention, stress reduction with endogenous and/or active exercises, training, early therapy and rehabilitation. PPG and other non-invasive sensor concepts and portable 24/7 biofeedback systems may take over a private role responsible even for monitoring vital body signs supplying support and alarming in extreme physical or mental stress or times of crises [60]. With user feedback, e.g., in form of a mobile phone, a technological/biological regulation loop will be realized whereby the subject/patient will be trained to manage vegetative functions to reduce stress and to follow telemedical support in case of emergency (see also Chap. 4).

1.11.3 Photoplethysmographic System Solutions for Sustainable Biomedical Care Diagnostic measuring methods based on biomedical technology are in widespread use today, typically involving patients’ visits to a doctor or a hospital, where measurements are then made. However, in certain circumstances, patients are too far away from the required facilities—be it while traveling in remote regions, or even inside an orbiting spaceship. A range of sustainable sensor concepts, systems solutions, and procedures—some already on the brink of clinical implementation—have been developed for such situations. For this purpose PPG is a simple in use and basically low-cost modality that can be used for long-term monitoring of patients with cardiovascular risk by their outdoor and working activities also via telemedical channel. If needed, a doctor can establish contact with patients to guide them through a ‘cinematographic’ recording of the relevant parameters, inform them about their condition, suggest appropriate measures, or dispatch on-site medical assistance, such as the well-known Flying Doctors in Australia.

1.11.4 Advanced Examination Strategies for Venous Saturation Detection and Arterio-Venous Oxygen Consumption The photoplethysmographic detection of peripheral arterial oxygen saturation (pulse oximetry) has become one of the established and well-proven methods in the monitoring of adults and children and is clinically used daily. The oxygen saturation indicates what percentage of the total hemoglobin is loaded with oxygen in the

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

25

investigated volume of blood. Pulse oximetry devices that usually detect the heart synchronous arterial blood volume pulse on fingers and toes (more recently, in the ear canal) is nowadays present in any ambulances and on any intensive care unit; however, they are used with equal success for long-term monitoring of serious injuries and patients with hemodynamic instability, increased cardiovascular risk, respiratory failure and other risk conditions. The venous oxygen saturation might recently be calculated from peripheral venous blood volume fluctuations that occur in a specific sequence leg exercise (producing so called venous pulse). Since the arterial oxygen saturation is calculated too, the local oxygen consumption can be determined. Normally, the arterial blood saturation value SpO2 should be around 98% and the peripheral venous saturation (SvO2 ) value at about 78%, so that the peripheral local saturation consumption is 20% in this case (see Chap. 5).

1.11.5 Analyzing Pulsatile Component of the PPG Waveform On the arterial PPG signal, several main pulse wave features can be observed (Fig. 1.19). Some of these evaluation parameters have already been formulated by Hertzman, others have been recently proposed and their clinical significance is currently under investigation. It should be noted at this point that the shape of the photoplethysmographically registered blood volume pulse depends on some experimental boundary conditions, including the nature of the PPG sensor (reflective, transmitive). Therefore, especially those derived pulse shape parameters appear to be diagnostically relevant. Among

Fig. 1.19 A fragment of recorded arterial rPPG pulse wave measured on second finger (left) and description of some pulse wave form parameters (right): (t 1 ) pulse rise time, (t 2 ) time to pulse peak, (t 3 ) time to minimum between pulse and dicrotic peak, (t 4 ) time to dicrotic peak, (t 5 ) pulse time, (t IW ) interwave time in 2/3 of the pulse peak, (R) arterial pulse amplitude (AC component of the PPG signal), (R0 ) DC component of the PPG signal, (R1 ) inflection point amplitude, (R3 ) interims minimum amplitude, (R4 ) dicrotic peak amplitude and (MPA) maximal pulse acceleration

26

V. Blazek

others, the so called tissue perfusion index (TPI) and Oliva Roztocil index (ORI) as extracted from the peripheral arterial pulse wave with suitable electronic filtering, amplification and signal post-processing permits relevant insights into the peripheral hemodynamical status of the monitored subject [61, 65, 66]. The slower (not heartsynchronous) rhythms in the dermal perfusion, as important as they are for the holistic assessment of complex local perfusion dynamics, first remain unconsidered here. The value of the photoplethysmographic DC signal component and the derived (normalized) pulse shape parameters TPI, ORI, RCT (relative crest time) and interwave distance (IWD) can be calculated as follows: 1 R0 = tp



t+tp

R(t)· dt,

(1.9)

t

TPI =

R , R 0 · tp

(1.10)

tIW , tp

(1.11)

t2 and tp

(1.12)

tIW , tp

(1.13)

ORI = RCT = IWD =

1.11.6 New PPG Horizons for Detecting Pain and Stress In addition, also respiratory activity describing parameters (RR, RRV) and other associated parameters like stress and pain sensitive evaluation can be derived from the peripheral arterial waveform recordings. In the last case, mostly the frequency selective components in the HRV-Fourier spectrum was analyzed (see Chap. 11). Currently used algorithms for pain assessment like, for example, the analgesia nociception index (ANI) or surgical stress index (SSI) [67, 68] analyze the heart rate, its amplitude and variability based on the ECG or PPG. But also the ORI described below seems to be sensitive to pain [60, 61, 65, 69, 70]. In order to be able to differentiate relaxation from stress, analyses of the heart rate and respiration rate is needed. Nevertheless, stress is a physiological phenomenon with very high inter- and intra-individual variation. Therefore, a quantification of a stress and pain level is challenging. In general, pain assessment is stress assessment as well. Newest research focuses on variations in respiration and heart frequency since this is more promising for such psychosomatic questions and subjective feeling factors [60]. As shown in Fig. 1.20, cardiac activity detected from peripheral photoplethysmogram (tPPG recorded from the second finger on the left hand) react significantly

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

27

Fig. 1.20 Experimental setup for detection of pain related skin perfusion changes

during the pain provocation test (50 µg capsaicin i.c. application, see figure top right) and influences HR and its variability.

1.11.7 Remote PPG Sensing Solutions In contrast to the conventional PPG with skin-attached discrete sensors, photoplethysmography imaging (PPGI) system, introduced at the RWTH Aachen University in 1997 [71–76], operates remotely, i.e., not in contact with the tissue. The measured large area of the skin is illuminated by quasi monochromatic LED light of selected wavelengths and is filmed by the camera from a distance of typically 50 cm. This detects small fluctuations in the tissue brightness, which are synchronous with the venous and/or arterial blood-volume dynamics. The fluctuations are caused by parts of the light which are reflected or transmitted towards the camera by passing the skin tissue. The core of the PPGI is an imaging strategy capable of contactless recording, processing, and displaying of image sequences of the selected skin area, so as to visualize the skin vessels and analyze dermal perfusion. The selected body area is mostly illuminated by monochromatic light (multiple LED panels), but observation of skin perfusion dynamics is also possible by ambient illumination. The size of

28

V. Blazek

the observed skin/body region and the spatial resolution can be arbitrarily chosen, depending on the utilized camera lens and distance between camera and measuring object. To minimize motion artifacts, which can be created by the separation of the camera sensor from the skin, different algorithm have been developed in the last decade [77, 78]. Different research groups has already shown that robust and unobtrusive estimation of the HR and the HRV is possible by the PPGI, also by using low cost cameras [79–81].

1.12 Conclusion Sophisticated assessment strategies are required to analyze complex rhythmical changes in dermal skin perfusion. The selected multi-sensor concepts allow to acquire undistorted perfusion signals from human skin in vivo across a broad frequency range. A comparative analyses of different sensor signals reveal that—next to the known central rhythms (correlated to the heart beat and respiration)—certain local oscillations, especially around 0.1 Hz, show high endogenous autonomy and influenceability. Using multi-wavelength PPG sensor concepts, peripheral venous saturation behavior can be detected. Probably already in the near future, such concepts will attain universal acceptance, analogous to the status enjoyed by arterial pulse oximetry. With the PPG equipment available on the market so far, the low frequency rhythms are invariably filtered out, due to the focus on capturing heart rate data for studying the heartbeat and heart rate variability (HRV). From these vital signs other PPG parameters can finally be calculated, which can throw insights into sensing stress or pain. With the present PPG technology, it also may be possible to investigate and understand the scientific basis for ancient Indian practices such as yoga and meditation, based on deep and controlled breathing. For time-related interdisciplinary research on the dermal rhythmic phenomena, the most important task continues to be clarification of the medical importance of an experimentally collected blood volume rhythmic. It would be desirable to undertake the entire gamut of possible experiments, in order to connect ancient Indian yoga teachings and classical western medicine on an evidence-based level. Another emerging perspective is a camera-based photoplethysmography imaging for functional, contact-less, spatially resolved visualization of the dermal venous and/or arterial perfusion behavior, for example, in the area of wounds. Thus, the steady successful progress in PPG as a non-invasive technique with old traditions and new perspectives leads to global clinical and also out-of-hospital health monitoring systems with increasing relevance.

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

29

References 1. S. Hales, Haemostaticks, in Statistical Essays II, Vol. 11, (Innings Manby, London, 1733) 2. www.britannica.com/EBchecked/topic/252340/Stephen-Hales 3. V. Blazek, New Insights in the phenomenon of distributed dermal perfusion rhythmicity using computer aided optoelectronic sensor measuring strategies. Acta Mech. Slovaca 14(2), 42–50 (2010) 4. V. Blazek, Biomedical Technology—2011 and beyond. Communications 1, 5–12 (2011) 5. K. Golenhofen, G. Hildebrandt, Die Beziehung des Blutdruckrhythmus zu Atmung und peripherer Durchblutung . Pflügers Archiv, Bd. 267 Nr 1, 27–45 (1958) 6. C. Bernard, Influence du grand sympathique sur la sensibilité et sur la calorification. Comptes Rendus De La Société De Biologie, Paris 3, 163–164 (1851) 7. https://www.whonamedit.com/doctor.cfm/846.html 8. E. Hering, Ueber den Einfluss der Atmung auf den Kreislauf. I. Mittheilung . Sitz. Ber. Kais. Akad. Wiss. Wien, math.-nat. Kl., 2 Abt. 60, 829–856 (1869) 9. L. Traube, Über periodische Tätigkeitsänderungen der vasomotorischen und HemmungsNervensysteme. Centralbl. Med. Wiss. 56, 880 (1865) 10. R.J.S. McDowall, The nervous control of the blood vessels. Physiol. Rev. 15, 98–174 (1935) 11. E. von Holst, Die relative Koordination als Phänomen und als Methode zentralnervöser Funktionsanalyse. Erg. Physiol. 42, 228–306 (1939) 12. K. Matthes, Kreislaufuntersuchungen am Menschen mit fortlaufend registrierenden Methoden. G. Thieme Verlag Stuttgart (1951) 13. H.P. Koepchen, P.H. Wagner, H.D. Lux, Über die Zusammenhänge zwischen zentraler Erregbarkeit, reflektorischem Tonus und Atemrhythmus bei der nervösen Steuerung der Herzfrequenz. Pflügers Archiv, 273(5), 443–465 (1961) 14. B.W. Hyndman, R.I. Kitney, B.M. Sayers, Spontaneous rhythms in physiological control systems. Nature 233(5318), 339–341 (1971) 15. H. Haken, Synergetic, an Introduction (Springer Verlag, Berlin, 1983) 16. R.I. Kitney, T. Fulton, A.H. McDonald, D.A. Linkens, Transient interactions between blood pressure, respiration and heart rate in man. J. Biomed. Eng. 7, 217–224 (1985) 17. K. Golenhofen, Physiologie. (Urban & Schwarzenberg, München), ISBN 3-541-16351-8 (1997) 18. B.R. Lotric, Coupling among subsystems that regulate blood flow. Ph.D. Thesis, University of Ljublana (1999) 19. V. Blazek, Optoelektronische Sensorkonzepte für nichtinvasive Kreislaufdiagnostik. Opto Elektron. Mag. 7, 212–219 (1991) 20. V. Blazek, U. Schultz-Ehrenburg, Quantitative photoplethysmography. Basic facts and examination tests for evaluating peripheral vascular functions. VDI Verlag Düsseldorf, ISBN 3-18-319220-9 (1996) 21. V. Blazek, Funktional leg vein diagnostics with the quantitative digital photoplethys-mography (D-PPG)—Proven and recent assessment methods of the peripheral venous and arterial situation, in Sklerotherapy in practice, ed. by K. Hübner. (Viavital Verlag Essen, 2007), pp. 49–58 22. U. Schulz-Ehrenburg, V. Blazek, Innovations in venous photoplethysmography, in Dermatology, progress & perspectives. ed. by W. Burgdorf, S.I. Katz (Partenon Publ. Group. Ltd., New York, 1993), pp. 849–851 23. V. Blazek, C. Blazek, Funktionelle Beinvenendiagnostik mit der quantitativen Photoplethysmographie PPG)—bewährte und neue Untersuchungstests der peripheren venösen und arteriellen Hämodynamik, in Praktische Sklerotherapie, eds. by K. Hübner, F.X. Breu (Viavital Verlag Köln, 2013), pp. 64–76. ISBN 978-3-934371-49-1 24. V.J. Kumar, V. Blazek, Photoplethysmography for non-invasive diagnostics, in Proceeding of the International Conference on Smart Instrumentation, Measurement and Applications (ICSIMA) (2013), ISBN 978-1-4799-0843-1/13

30

V. Blazek

25. C.M. Cartwright, Infrared transmission of the flesh. J. Opt. Soc. Amer. 20, 81–84 (1930) 26. K. Matthes, Untersuchungen über die Sauerstoffsättigung des menschlichen Arterienblu-tes. Arch. Exper. Pathol. Pharmakol. 179, 698–711 (1935) 27. H. Molitor, M. Kniazuk, A new bloodless method for continuous recording of peripheral circulatory changes. J. Pharmakol. Exp. Ther. 57, 6–18 (1936) 28. A.B. Hertzman, C.R. Spealman, Observations of the finger volume pulse recorded photoelectrically. Amer. J. Physiol. 119, 334–340 (1937) 29. A.B. Hertzman, The blood supply of various skin areas as estimated by the photoelectric plethysmograph. Amer. J. Physiol. 124, 329–340 (1938) 30. A.C. Burton, the range and variability of the blood flow in the human fingers and the vaso-motor regulation by body temperature. Am. J. Physiol. 127, 437–453 (1939) 31. A.C. Burton, R.A. Taylor, A study of the adjustment of peripheral vascular tone to the requirements of the regulation of body temperature. Am. J. Physiol. 129, 565–577 (1940) 32. A.B. Hertzman, J.B. Dillon, Applications of photoelectric plethysmography in pereipheral vascular disease. Am. Heart J. 20, 750–761 (1940) 33. A.B. Hertzman, L.W. Roth, The vasomotor components in in the vascular reactions in the finger to cold. Am. J. Physiol. 136, 680–691 (1942) 34. W. Karlen, S. Raman, J.M. Ansermino, G.A. Dumont, Multiparameter respiratory rate estimation from the photoplethysmogram. IEEE Trans. Biomed. Eng. 60(7), 1946–1953 (2013) 35. V. Blazek, H.J. Schmitt, M.M. Rao, Studies of neurological induced skin perfusion changes using optoelectronic devices: Aims and results of an Indo-German project. Proc. Od SPIE 3668, 566–573 (1999) 36. T. Mühl, Berechnung von Lichtintensitäten in streuenden Medien. Ph.D. Thesis, RWTH Aachen University (1988) 37. H.J. Schmitt, V. Blazek, T. Mühl, Light scattering in biological tissue, in Proceeding V. Tropical College on Applied Physics, (Kuala Lumpur, Malaysia, 1990) 38. V. Blazek, E.W. Gaelings, H.J. Schmitt, Dynamik der Blutbewegung im Gefäßsystem des Menschen. RWTH-Themen 1/93 “Technik in der Medizin”, (1993), pp. 26–27 39. E.W. Gaelings, W. Rütten, V. Blazek, Computer modelling of growth and hemodynamics of vascular networks, in Advances in vascular imaging and photoplethysmography. Proc. CNVD´95, (VDI Verlag Düsseldorf, 1996) eds. by U. Schulz-Ehrenburg, V. Blazek (FortschrittBerichte, Reihe 20, Nr. 221, 1996) pp. 1–7. ISBN 3-18-322120-9 40. E.W. Gaelings, Numerische simulation hämodynamischer Processe in vaskulären Netzen. Ph.D. Thesis, RWTH Aachen University (1996) 41. V. Blazek, H.J. Schmitt, Quantitative Photoplethysmographie—Eine nicht-invasive ScreeningMessmethode für die funktionelle Gefäßdiagnostik. Minim. Invasive Med 5(3), 123–128 (1994). ISSN 092-9042 42. U. Fornefeld, Rechnerunterstützte Analyse optisch abgetasteter Perfusionsrhythmik der Haut, Master Thesis, Institut für Hochfrequenztechnik, RWTH Aachen University (1999) 43. M. Hülsbusch, V. Blazek, Rhythmical phenomena in dermal perfusion—proved assessment strategies and new discoveries. Adv. Electr. Electron. Eng. 3(4, 3), 112–117 (2005) 44. V. Blazek, T. Mühl, H.J. Schmitt, H. Schmid-Schönbein, Optoelektronische Erfassung und rechnerunterstützte Analyse der Mikrozirkulations-Rhythmik. Biomed. Techn. 30(Erg.-B), 121–122 (1985) 45. T. Mühl, V. Blazek, H.J. Schmitt, H. Schmid-Schönbein, Measuring system with fiber sensors for the recording an Fourier-Analysis of dermal microcirculation oscillations. in Proceeding of SPIE 713 Optical Fibers in Medicine (1986), pp. 121–125 46. M.M. Rao, V. Blazek, H.J. Schmitt, Investigations on the influence of breathing and brainactivity using optical sensors. Proc. SPIE 2982, 53–64 (1997) 47. V. Perlitz, H. Schmid-Schönbein, A. Schulte, J. Dolgner, E. Petzold, W. Kruse, Effektivität des Autogenen Trainings. Therapiewoche 26, 1536–1544 (1995) 48. H. Schmid-Schönbein, S. Ziege, The high pressure system of the mammalian circulation as a dynamic self-organizing system, in Rhythms in Physiological Systems. ed. by H. Haken, H.P. Koepchen (Springer Series in Synergetics, Berlin, Springer, 1991), pp. 77–96

1 Skin Perfusion Studies: Historical Notes and Modern Measuring …

31

49. H. Schmid-Schönbein, S. Ziege, R. Grebe, V. Blazek, R. Spielmann, F. Lienzenich, Synergetic interpretation of patterned vasomotion activity in microvascular perfusion: Discrete effects of myogenic and neurogenic vasoconstriction as well as arterial and venous pressure fluctuations. Int. J. Microcirc. 17, 346–359 (1997) 50. M. Mück-Weimann, D. Hager, K. Streubel, R. Kalb, T. Rechlin, Some new investigations regarding the inspiratory gaps response, in Frontiers in Computer-Aided Visualisation of Vascular Functions, eds. by V. Blazek, U. Schultz-Ehrenburg (VDI Verlag Düsseldorf, 1998), pp. 69–76. ISBN 3-18-330020-6 51. H.J. Schmitt, V. Blazek, M. Hülsbusch, Assessment and visualisation of different rhythmical phenomena in the skin perfusion obtained by optical sensing (PPG and PPGI), in Computeraided noninvasive vascular diagnostics proceeding of 11th International Symposium CNVD eds. by V. Blazek, U. Schultz-Ehrenburg (Mainz Wissenschaftsverlag Aachen, 2005), pp. 71– 78. ISBN 3-89653-942-6 52. M. Hülsbusch, V. Blazek, M. Herzog, S. Vogel, T. Wartzek, D. Starke, T. Hennig, Development of a miniaturized in-ear pulse oximeter for long term monitoring of risk patients, in IFMBE Proceedings 25/IV, eds. by O. Dössel, W.C. Schlegel (Springer Verlag, 2009). pp. 779–781. ISBN 978-3-642-03897-6 53. M. Hülsbusch, V. Blazek, S. Leonhardt, S. Vogel, D. Römhild, T. Henning, F. Panish, Sensoreinheit, insbesondere für nichtinvasives Langzeitmonitoring der Hämodynamik. German Patent Application (2008). No. 10 2008 002 742.1 54. S. Vogel, M. Hülsbusch, D. Starke, S. Leonhardt, A system for assessing motion artifacts in the signal of a micro-optic in-ear vital signs sensor, in Proceeding of 30th Annual International Conference of the IEEE EMBS, (Vancouver, Canada, 2008), pp. 510–513 55. S. Vogel, M. Hülsbusch, T. Henning, V. Blazek, S. Leonhardt, In-ear vital sings monitoring using a novel micro-optic reflective sensor. IEEE Trans. Inf Technol. Biomed. 13(6), 882–889 (2009) 56. B. Venema, N. Blanik, V. Blazek, H. Gehring, A. Opp, S. Leonhardt, Advances in reflective oxygen Saturation monitoring with a novel in-ear sensor system: results of a human hypoxia study. IEEE Trans. Biomed. Eng. 59(7), 2003–2010 (2012) 57. B. Venema, H. Gehring, I. Michelsen, N. Blanik, V. Blazek, S. Leonhardt, Robustness, specifity and reliability of a novel In-ear pulse oximetric sensor in surgical patients. IEEE J. Biomed. Health Inf. 6(2013). 10.1109/JBHI.2013.2292118 58. B. Venema, J. Schiefer, V. Blazek, N. Blanik, S. Leonhardt, Evaluating Innovative In-ear pulse oximetry for unobstructive cardiovascular homecare monitoring during sleep. IEEE J. Trans. Eng. Health Med. 1, 8(2013). 10.1109/JTEHM. 2013.2277870 59. S. Mallat, A Wavelet Tour of Signal Processing, 2nd edn. (Academic Press, USA, 2001) 60. B. Venema, N. Blanik, V. Perlitz, S. Laffar, H.G. Ortlepp, S. Borik, J.-P. Jansen, M. Koeny, V. Blazek, S. Leonhardt, Biofeedback—smart modality fusion for clinical, home and outdoor health monitoring. Biomed. Technik 59(S1), 132–135 (2014) 61. B. Venema, V. Blazek, S. Leonhardt, In-ear photoplethysmography for mobile cardiorespiratory monitoring and alarming. in Proceeding of BSN IEEE Conference (2015), 10.1109/BSN.2015.7299367 62. J. Havlik, L. Lhotska, J. Parak, J. Dvorak, Z. Horcik, M. Pokorny, A Modular system for rapid development of telemedical devices. J. Univ. Comput. Sci. 19(9), 1242–1256 (2013) 63. J. Allen, Photoplethysmography and its application in clinical physiological measurement. Physiol. Meas. 28, R1–R39 (2007) 64. D.H. Kim et al., Epidermal sensors. Science 233, 838–843 (2011). 10.1126/science.1206157 65. I. Oliva, K. Roztocil, Pulse wave analysis in obliterating atherosclerosis. Angiology 34, 610– 619 (1983) 66. V. Blazek, A. Fronek, H.J. Schmitt, R. Hütter, G. Holst, Ein neues optoelektronisches ScreeningMesssystem für die Langzeit-Diagnostik peripherer arterieller Durchblutungsstörungen. VASA, Suppl. 32, 533–536 (1991) 67. M. Koeny, N. Blanik, X. Yu, M. Czaplik, M. Walter, R. Rossaint, V. Blazek, S. Leonhardt, Using photoplethysmography imaging for objective contactless pain assessment. Acta Polytech. 54(4), 275–280 (2014)

32

V. Blazek

68. M. Koeny, S. Bertling, M. Czaplik, V. Blazek, S. Leonhardt, Respiratory rate estimation in postoperative care—state of the art and perspectives. Clin. Technol./Lékaˇr a Technika 44(2), 25–29 (2014) 69. K. Navratil, J. Halek, P. Havranek, S. Binder, Pulse wave analysis in objective evaluation of pain—a preliminary communication. Cesk Slov. Neurol. N 71(104), 303–308 (2008) 70. D. Korpas, J. Halek, L. Dolezal, Parameters describing the pulse wave. Physiol. Res. 58, 473–479 (2009) 71. V. Blazek, New insights in the phenomenon of distributed dermal perfusion rhythmicity using computer aided optoelectronic sensor measuring strategies. Acta Mech Slovaca 14(2), 42–50 (2010) 72. O. Such, S. Acker, V. Blazek, Mapped hemodynamic data acquisition by near infrared CCD imaging. in IEEE/EMBC Proceeding, 637–639 (1997). 0-7803-4252–3/97 73. M. Hülsbusch, A functional imaging technique for optoelectronic assessment of skin perfusion. Ph.D. Thesis, RWTH Aachen University (2008) 74. N. Blanik, A.B. Abbas, B. Venema, V. Blazek, S. Leonhardt, Hybrid optical imaging technology for long-term remote monitoring of skin perfusion and temperature behavior. J. Biomed. Optics 19(1), 16 (2014). https://doi.org/10.1117/1.JBO.19.1.016012 75. N. Blanik, C. Pereira, M. Czaplik, V. Blazek, S. Leonhardt, Remote photoplethysmographic imaging of dermal perfusion in a porcine animal model. IFMBE Proc. 43, 92–95 (2014) 76. N. Blanik, B. Venema, V. Blazek, S. Leonhardt, Remote pulse oximetry imaging—fundamentals and applications. Clinician Technol/Lékaˇr a Technika 44(3), 5–11 (2014) 77. C.S. Pilz, J. Krajewski, V. Blazek, On the diffusion process for heart rate estimation from face videos under realistic conditions, in Pattern Recognition, ed. by V. Roth, T. Vetter (GCPR 2017, Springer International Publishing AG, 2017), pp. 361–373. https://doi.org/10.1007/978-3-31966709-6_29 78. C.S. Pilz, S. Zaunseder, J. Krajewski, V. Blazek, Local group invariance for heart rate estimation from face videos in the Wild, in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (2018), pp. 1367–1375. https://doi.org/10.1109/CVPRW.2018.00172 79. C.S. Pilz, I.B. Makhlouf, V. Blazek, S. Leonhardt, On the vector space in photoplethysmography imaging, in Proceedings of CVPR 2019 (Computer Vision for Physiological Measurement) (2019). CoRR abs/1906.04431 80. S. Borik, S. Lyra, M. Paul, Ch. Hoog Antink, S. Leonhardt, V. Blazek, Photoplethysmography imaging: camera performance evaluation by means of an optoelectronic skin perfusion phantom. Physiol. Meas. 41(5) (2020) 81. X. Yu, T. Laurentius, C. Bollheimer, S. Leonhardt, Ch. Hoog Antink, Noncontact monitoring of heart rate and heart rate variability in geriatric patients using photoplethysmography imaging. IEEE J. Biomed. Health Inform 1–1

Chapter 2

Influence of Controlled Breathing (Pranayama) on Dermal Perfusion Mandavilli Mukunda Rao

Abstract Yoga, a well-known method that combines meditation and physical exercise, originated in India. There are several claims that yoga is not just an exercise, but an wholistic process that cures the bodily ailments and purifies the brain. However, such claims are not substantiated with scientific evidence, though several anecdotal evidences exist. In this chapter experiments were conducted to identify whether one of the yoga exercises, namely, Pranayama (Rhythmic breathing exercise) leads to quantifiable positive changes in the body functions. The method of photoplethysmography is utilized to ascertain perceptible changes in the cardiovascular functions.

2.1 Introduction It is a well-acknowledged fact that man knows more about outer space than the functioning of his own brain. That is why the 1990s were declared as the decade for brain research with many R&D institutions and Universities/academic institutions taking part in this research activity around the world. For scientists who study the human brain, even the simplest act of perception is an astonishing event of intricacy. Using Positron Emission Tomography (PET) researchers found evidence suggesting that the human brain focuses attention on parts of the body where stimulation is expected by suppressing ‘competing’ information from other areas of the body. In a recent issue of the British Journal ‘Nature’, researchers reported decreased blood flow in parts of the brain’s somatosensory cortex in response to anticipated stimulation in various parts of the body. With the new technologies like functional Magnetic Resonance Imaging (fMRI) and other sophisticated instrumentation, researchers catch brains in the very act of cognition, feeling or remembering. However, the use of such tools for brain research has been limited because of astronomical costs of these instruments and the complexity in installation/operation. M. Mukunda Rao (B) Indian Institute of Technology Madras, Chennai, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_2

33

34

M. Mukunda Rao

More than 85 years ago in 1927, Hans Berger (1873–1941), a German doctor and scientist has succeeded in recording Electro-Encephalo-Gram (EEG)—the pattern of minute changes in the electrical potential of the scalp of a human subject (his own son!). He then hoped that it might be possible to establish a definite relationship between these potentials and the performance of the brain. Subsequent studies showed that the EEG reflects only general functional states of the brain, such as different states of wakefulness and sleep or metabolic disturbances, which, however, only emphasizes the importance of EEG as a Non-Invasive Diagnostic Tool in gathering wealth of neurological information on brain-related disorders like: epilepsy, early detection and localization of brain tumours, coma assessment in intensive care units (ICU) and in the definition/assessment of sleep stages/disorders. [1]. In recent decades, the advent of computers and signal processing techniques has enhanced the capabilities of the present-day EEG machines which have emerged as a relatively low-cost non-invasive diagnostic tool for neurological disorders. Consequently, they are still in use worldwide both for diagnostics as well as for research. In recent decades, medical technology has made breath-taking advances. Optical sensors are increasingly being used for non-invasive diagnostics in biomedical applications. However, there is far very scanty information in the literature about their possible application for investigating brain-related events. This could possibly due to the fact that brain is optically opaque due to the surrounding skull and hair. However at the temples, the tissue is relatively soft and by positioning the optical sensors at this point, one can monitor the transcutaneous blood volumetric changes which are related to the brain activity. Among the biomedical optical sensors, photoplethysmography (PPG) has a unique position. They operate in the near infrared region (around 940 nm) where the skin is relatively transparent and thereby could register the blood volume changes in the near skin microcirculatory blood vessels [2]. Using the PPG realtime detection of brain events have been reported and the left & right brain activities could be monitored thus opening the possibility of studying the brain asymmetry [3]. From the real-time data it is possible to obtain the frequency related information by using the Fast Fourier Transform (FFT) techniques. Thus, it is possible to extract the breathing-related information from the PPG signals. Further studies in this direction have led to the understanding of the influence of breathing on brain using these PPG sensors in conjunction with breathing sensors [4]. This has opened the gates for the scientific studies of classical Indian breathing techniques called Pranayama. These studies have revealed that deep breathing techniques like Pranayama caused what might enhance the coupling and operation between the heart and the lungs which might lead to an optimization of exchange of gases which is vital significance. In the process, the emergence of the low-frequency rhythms in the range of 0.12–0.15 Hz has been observed which are known to be linked to the relaxation of the human body [5].

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

35

2.2 Breathing The normal rate of breathing in a human being is on the average fifteen times a minute and 21,600 times in every twenty-four hours. However, this number varies according to one’s way of life, health and emotional state. The lungs are directly concerned with the absorption of oxygen in the arterial blood and the disposal of carbon dioxide in the venous blood. As food is necessary to sustain the body, breathing is essential for maintaining the life force. It is said that man may survive without food for several days but without breathing he may not survive for more than three minutes. Thus, if breathing stops so does life. The depth and rate of breathing is so regulated as to meet the supply of fresh oxygen which is constantly needed by the cells and to discharge the carbon dioxide accumulated in them. The breathing cycle consists of three parts: inhalation, exhalation and suspension. Inhalation is an active expansion of the chest by which the lungs are filled with fresh air. Exhalation is normal and passive recoil of the elastic chest wall by means of which the stale air from the lungs is emptied. Suspension is a pause at the end of each inhalation and exhalation. These three form one cycle of breathing. The breathing affects the heart rate as well as the quality of blood pumped by the heart to the various regions of the body including the brain [6]. Breathingas the source of oxygen uptake is the most important function of the human body which is fundamental for its very survival. This maintains the oxygen level in the blood stream to the required amounts, which, when transported through the arteries, will fuel metabolism of the multitude of functions within the human body. The most important consumer of oxygen in the body is the brain which accounts for almost 25% of all oxygen uptakes through the breathing process. The quality of the breathing process plays an important role for maintaining good health [7]. This refers to the manner in which oxygen is inspired and carbon dioxide is expired. In this respect, the external sections of the nose serve to gather air and accelerate its flow, forming a rapid jet that enters the cavity within the face, the internal sections of the nose. The internal nose is strategically connected to the brain through the olfactory bulbs which are responsible for the sense of smell [8].

2.3 Control of Breathing Normally, the air enters and leaves the lungs at a rate of 14 to 16 times per minute without one being aware of it. Breathing is one of the few bodily functions which, within limits, can be controlled both consciously and unconsciously.

36

M. Mukunda Rao

2.3.1 Conscious Control Conscious control of breathing is common in many forms of meditation, specifically forms of yoga for example pranayama unlike anapana which is only awareness of breath. In swimming, cardio fitness, speech or vocal training, one learns to discipline one’s breathing, initially consciously but later sub-consciously, for purposes other than life support. Human speech is also dependent on conscious breath control.

2.3.2 Unconscious Control Unconsciously, breathing is controlled by specialized centres in the brainstem, which automatically regulate the rate and depth of breathing depending on the body’s needs at any time. When carbon dioxide levels increase in the blood, it reacts with the water in blood, producing carbonic acid. Lactic acid produced by anaerobic exercise also lowers pH. The drop in the blood’s pH stimulates chemoreceptors in the carotid and aortic bodies in the blood system to send nerve impulses to the respiration centre in the medulla oblongata and puns in the brain. These, in turn, send nerve impulses through the phrenic and thoracic nerves to the diaphragm and the intercostal muscles, increasing the rate of breathing. Even a slight difference in the blood’s normal pH 7.4, could cause death, so this is an important process. This automatic control of respiration can be impaired in premature babies, the elderly or by drugs and disease (7).

2.3.3 Breathing Asymmetry It is interesting to note that in a healthy subject, there is right-left asymmetry of breath flow, first described by Kayser as early as in 1895 [9]. Breathing is predominant either through the right nostril or through the left nostril. If not interfered with, the nasal air flow will exibit rhythmic functioning as the predominance of breathing through one nostril lasts for 1–2 h after which it shifts to the other nostril. The flow increases in one side until it reaches a peak, and then it begins to decrease. Finally, most of the air starts flowing through the opposite nostril. Though this is a natural biological rhythm, it can be interfered voluntarily or by factors like e.g. shifts in emotional arousal [10].

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

37

2.3.4 Introduction to Pranayama and Yogic Breathing Breathing is so simple and so obvious we often take it for granted; ignoring the power it has to affect body, mind and spirit. With each inhale we bring oxygen into the body and spark the transformation of nutrients into fuel. Each exhale purges the body of carbon dioxide, a toxic waste. Breathing also affects our state of mind. It can make us excited or calm, tense or relaxed. It can make our thinking confused or clear. What’s more, in the yogic tradition, air is the primary source of prana or life force, a psycho-physio-spiritual force that permeates the universe. Pranayama is loosely translated as prana or breathe control [11]. The ancient yogis developed many breathing techniques to maximize the benefits of prana. Pranayama is used in yoga as a separate practice to help clear and cleanse the body and mind. It is also used in preparation for meditation, and in asana, the practice of postures, to help maximize the benefits of the practice, and focus the mind.

2.4 How to Do It Hold your right hand up and curl your index and middle fingers toward your palm. Place your thumb next to your right nostril and your ring finger and pinky by your left. Close the left nostril by pressing gently against it with your ring finger and pinky, and inhale through the right nostril. The breath should be slow, steady and full. Now close the right nostril by pressing gently against it with your thumb, and open your left nostril by relaxing your ring finger and pinky and exhale fully with a slow and steady breath. Inhale through the left nostril, close it, and then exhale through the right nostril. That’s one complete round of Nadi Shodhana: • • • •

Inhale through the right nostril, Exhale through the left, Inhale through the left, Exhale through the right.

Begin with 5–10 rounds and add more as you feel ready. Remember to keep your breathing slow, easy and full [1]. Various postures of Pranayama are given in Fig. 2.1.

2.5 Experimental Details The results reported in the present study are based on the optical sensors (PPG) mounted on the temples for monitoring the brain activity in terms of the underlying

38

M. Mukunda Rao

Fig. 2.1 a The finger position, b Pranayama exercise

blood volumetric changes. For monitoring the breathing activity of the subject, the LM 45 series precision integrated circuit temperature sensors manufactured by the National Semiconductor Corporation are used. The output voltage from these sensors is linearly proportional to the Celsius (Centigrade) ambient temperature. The LM 45 s low output impedance; linear output and precise inherent calibration make interfacing to readout or control circuitry especially easy. Since it draws only 120 micro-amps from its power supply, it has very low self-heating. The long-term stability of this sensor is +0.12 C/1000 h. The time constant of this device makes it possible to record the periodic temperature changes following the breath. Continuous heavy breathing will cause a steady increase in the average ambient temperature, thus shifting the DC base line. The small size of these temperature sensors (1.2 mm × 2.8 mm × 0.8 mm) makes them ideally suitable to mount them inside the nostrils without much discomfort to the subject. The complete experimental station with the necessary recording equipment is shown in Fig. 2.2. The typical signal patterns from the PPG and breathing sensors is shown in Fig. 2.3 below. It can be seen from this figure that the breathing from one of the nostrils is shallower than the other. Besides, it can also be observed that the rate/frequency of the PPG pattern, which follows the heartbeat, is approximately four to five times that of the breathing rate/frequency. The signals obtained from the two optical sensors—OS 1 and OS 2—mounted on the temples of the subject as shown in the picture given above and the signals obtained from the two nostrils—BR 1 & BR 2—are shown in Fig. 2.3. The PPG sensors were properly placed on the left and right temples of the subject with the transmitting diode (emitter) and the receiving diode (receiver) in line with the underlying arterial blood vessel, after thoroughly cleaning the surface of the skin with medical alcohol. The subject was also requested to avoid unnecessary movements during the measurement in order to obtain a recording of superior quality free from motion artefacts. Thermistors with dimensions of the order of few microns,

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

39

Fig. 2.2 The subject with the sensors mounted and the measuring equipment in the background

Fig. 2.3 Typical signal patterns from the PPG sensors & breathing sensors

as already mentioned earlier, were introduced into the left and right nostrils of the subject without causing any inconvenience/irritation to the subject. This will facilitate recording of the breathing patterns for the left and right nostrils which are independent of each other. As seen from the above figure, the PPG sensors follow the heart rate which is typically four to five times that of the breathing rate. Finally the ECG electrodes were placed on the body of the subject appropriately as per the norms. The sensor placements, amplification settings, filter settings, ambient light in the room and all other necessary connections were thoroughly crosschecked by two independent investigators prior to starting the experiment. Any variation in the optical coupling between the sensor head and the subject, or physiological changes which dynamically alter the transmitted light give rise to what is commonly termed as motion artefacts. In fact, even a simple movement

40

M. Mukunda Rao

may produce a complex motion artefact, just as ambient light can also cause artefacts by coupling to probe receiver, either directly or by transmission through the tissue. Whilst it is theoretically possible by straightforward means to remove ambient artefact, practical limitations mean that sufficiently bright or high frequency artificial light sources can still cause artefacts. Thus it is imperative that the investigations/measurements should be conducted under controlled conditions. A normal PPG signal free from artefacts is shown in Fig. 2.4. Figures 2.5 and 2.6 show the breathing signal and ECG signal recorded under normal breathing conditions. A special equipment which was designed and built for this study called Universes II by Mr. K. H. Strotmann, Germany, for simultaneous recording of all the above parameters namely PPG, Breathing & ECG was used for the present investigation.

Fig. 2.4 Typical PPG signal free of any artefact

Fig. 2.5 Typical breathing signal free of any artefact

Fig. 2.6 Typical ECG signal free from any artefact

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

41

The signals were recorded at a sampling rate of 250 Hz and the measurement system has a built-in notch filter for removing the 50 Hz noise. This system also has an inbuilt variable filter confined to each channel. This was used as a 25 Hz low pass filter for removing other noises while making the measurements. The Diadem 7.0 software (National Instruments) was used for the filtering of raw data, the application of the Fast Fourier Transform (FFT)/Power Spectral Density (PSD) and for plotting of graphs. This equipment was generously donated by Alexander von Humboldt (AvH) of Germany to the author who was an AvH Fellow and is presently located in Sri Ramachandra University where these investigations were carried out. Deep breathing (Protocol followed in making the measurements): The subject is first informed about the protocol to be followed during deep breathing. Breathing is done consciously and continuously such that the stomach and the chest of the subject remain fully stretched after one full inspiration. During expiration, the subject is requested to expire as much as possible continuously without break and then start inspiration again. Essential care is taken to ensure that inspiration and expiration are done with the same force. While under normal breathing conditions, tidal volume of an average adult person is about 500 ml, it amounts in the case of deep breathing to roughly 2000–3000 ml. The measurements were carried out using the above set-up under two different breathing conditions viz. Normal breathing and Deep breathing. Measurements were also made in subjects asked to relax with the help of autogenic training (AT), a relaxation technique to compare results. AT is usually performed by internally visualizing limbs and projection of physiological sensations onto them, and on an enhanced level extending such sensations to internal organs, e.g. heart pace and respiration, feeling that they are relaxed. This kind of relaxation was done for 15 min before the measurement was started. AT suggestion was given by Sanysi Krishna Yogam, Chennai and the measurements were made on subjects trained by him. Seven male subjects and three female subjects with a mean age of 22 years with no breathing disorders were taken for the present study. As the FFT of these signals contain numerous peaks, PSD which is nothing but the square of FFT is used to avoid any confusion in interpreting. However, only FFT is suitable for analyzing the breathing data. The Figs. 2.7 and 2.8 show the PSD of the ECG signal and the FFT of the breathing signal recorded along (simultaneously) with the PPG signal. Figure 2.9 displays the spectral analysis (PSD) of a PPG signal Fig. 2.7 Spectral analysis of ECG data before performing autogenic training (AT)

42

M. Mukunda Rao

Fig. 2.8 Spectral analysis of breathing data before performing autogenic training (AT)

Fig. 2.9 Spectral analysis of PPG data before performing autogenic training (AT)

from the temples of a subject during a naïve state prior to performing the AT. Figure 2.9 displays the spectral analysis (PSD) of a PPG signal from the temples of a subject during a naïve state prior to performing AT. The Figs. 2.7 and 2.8 show the PSD of the ECG signal and the FFT of the breathing signal recorded along with the PPG signal. The most dominant peak in the PPG spectrum at 0.9 Hz shown in Fig. 2.9 corresponds to the peak at 0.9 Hz in the ECG which represents the heart rate. The same PPG spectrum also shows a barely detectable peak at 0.3 Hz which corresponds with the peak at 0.3 Hz in frequency spectrum (FFT) of the breathing signal. Other than these two peaks, another peak may be detected/observed varying over a range at 0.14 Hz and below, depending on the mental and physical condition of the subject under study. Though several explanations are given for the appearance of this low-frequency component, the origin is still debatable. Figures 2.10, 2.11 and 2.12 show a similar spectral analysis of similar data Fig. 2.10 Spectral analysis of ECG data after performing autogenic training

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

43

Fig. 2.11 Spectral analysis of breathing data after performing autogenic relaxation

Fig. 2.12 Spectral analysis of PPG data after performing autogenic training

recorded on the same subject after performing the autogenic relaxation technique described earlier. By carefully comparing the several spectral components discussed above in Figs. 2.7, 2.8, 2.9, 2.10, 2.11 and 2.12, several interesting features be observed. Firstly, a much stronger and sharper (narrow width) breathing peak may be clearly noticed in Fig. 2.11 as compared to Fig. 2.8. Apart from this an important low-frequency peak at 0.12 Hz can be observed in the PSD of the PPG signal in Fig. 2.12 not present in the PPG spectrum of the signal recorded before performing AT (Fig. 2.9). This periodic low-frequency component (low frequency rhythm) around 0.12–0.14 Hz may disappear and reoccur repeatedly depending on the physical and mental condition of the subject [12]. In the second phase of the experiment, PPG, Breathing and ECG data were recorded under two different breathing conditions. Figure 2.13 displays the spectral analysis (PSD) of PPG signal under normal breathing conditions. The same figure also shows the FFT of the breathing signal. Here too, the frequency spectrum of the PPG signal shows a dominant peak at 1 Hz corresponding to the heart rate and a relatively weaker peak at 0.3 Hz which corresponds to the breathing rate. The PSD of the left and right PPG signal (Fig. 2.13—upper panel) also provides evidence on the lateralization of subcutaneous blood perfusion dynamics since the 0.3 Hz peak is clearly more prominent in the PSD of the left PPG signal. This temporal difference in the functional properties of nostrils can be clearly observed in the FFT spectrum of the breathing signal recorded from the left and right nostrils as shown in Fig. 2.12. In this case, the strength of the breathing activity through the right nostril is roughly three times that of the right nostril. Figure 2.14 shows a similar spectral analysis of similar data recorded on the same subject while he was performing the protocol for deep breathing as elaborated earlier.

44

M. Mukunda Rao

Fig. 2.13 Spectral analysis (PSD) of breathing and PPG data recorded during normal breathing

Comparing the PPG spectra shown in Figs. 2.13 and 2.14, it can be noticed that under these conditions the breathing peak in the spectrum is much sharper and has a slightly lower frequency at 0.25 Hz. More importantly, a prominent peak can be noticed at a frequency marginally less than 0.12 Hz which has no harmonic relation with the respiratory activity or the cardiac activity. This is similar to for the lowfrequency peak obtained in the PPG spectrum of the data recorded after AT. It can also be noted that the respiratory peak and the cardiac peak are multiples of one frequency. In other words, the ratio of heart rate to breathing rate is an integer under conditions of deep breathing. It can also be noticed that the harmonics of the cardiac and respiratory peaks are strengthened during deep breathing. The impact of respiration on cutaneous on blood flow in the upper limb have been investigated by Weyman [13]. Their study indicates that during sleep, the respiration seems to be dropping from 0.35 to 0.15 Hz and simultaneously the low-frequency rhythms in the frequency range of 0.075 Hz seems to be surfacing in the transcutaneous blood volumetric changes as monitored by r-PPG sensors. At this point of time, it is not clear whether is causing these rhythms or vice-versa but the implications of this in breathing therapies for cardio-vascular diseases has already been discussed. The present comparative study of the PPG signal recorded during regular and deep breathing clearly shows that deep breathing strengthens both the low-frequency rhythm, or relaxation rhythm and the respiratory component. Following animal studies, it has been established that the source of this low-frequency rhythm is in

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

45

Fig. 2.14 Spectral analysis of breathing and PPG data recoded during deep breathing

reticular neurons of the reticular formation of the lower brain stem [14]. It has also been convincingly demonstrated in invasive recordings in as many as 500 neurons that a rhythm at ca. 0.15 Hz emerged in response to a decline in the level of activation, e.g., triggered by administration of anaesthetics. These mechanisms have recently been thoroughly compared in acine and man [15]. This too helped in identifying the decline in the level of psychological, mental or physical activation as the controlling parameter needed to spark the emergence of the reticular rhythm. It appears most possible that those mechanisms also apply to the emergence of the 0.12 Hz rhythm observed under the conditions of deep breathing as described above. This assumption is corroborated by our observation of the phenomenon of harmonic resonation. It is also noticed that deep breathing caused what might enhance the coupling and coordination between the heart and the lungs which might lead to optimization of exchange of gases which is of vital significance. Therefore the practice of deep breathing will improve physical and mental relaxation and in turn improve human well-being and correct various respiratory problems.

46

M. Mukunda Rao

2.6 Studies on the Effects of Controlled Breathing or Pranayama We all breathe but we do not all derive the proper benefit from our breathing. This was because we do not know to regulate our breath. We breathe automatically, involuntarily, unconsciously; we must learn to breathe consciously, properly and rhythmically. Controlling the life-force by conscious control of breath was known as Pranayama. When we understand how to do this, we can fill ourselves with Prana-vital energy-and thus eliminate all our impurities.” The procedure for doing Pranayama is already described earlier. The study comprised of 40 subjects who were made to sit in an erect normal position. The PPG signal from the normal subjects sitting in a straight posture was recorded from the left and right ear lobes. The duration of the recording in both the normal and Pranayama subjects was for 30 min. The outputs of the two PPG sensors were digitally sampled at a sampling rate of 40 Hz and stored for further analysis. In order to eliminate noise, the data was filtered using a Butterworth low pass filter of order 8 (cutoff—7 Hz) to remove any high frequency component that might be present in the signal. The very low-frequency components contained in a signal are sometimes an artifact caused either by the instruments or the movement of the subject, which shifts the PPG signal up or down. The PPG sensor is also very sensitive to these shifts. These low-frequency components smear the power spectrum of the PPG signal and can affect the results. The PPG signal consists of a quasi DC signal that corresponds to changes in the venous pressure. This quasi DC signal was removed by using a 100 point moving average detrending algorithm before being subjected to analysis. The PSD estimate gives a distribution of energy in the frequency domain and yields a smoother curve to the conventional FFT frequency domain analysis. We observe these distinct components given in Fig. 2.15. This figure. Fig. 2.15 The power spectral density of a normal subject before he carries out Pranayama

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

47

Fig. 2.16 The power spectral density of PPG signal of a subject after he concludes Pranayama procedure

• • • •

(1) is the low-frequency component existing at 0.08–0.1 Hz; (2) is the breathing component with a frequency of 0.2–0.25 Hz; (3) is the cardiac component with a frequency of 1 Hz; (3a & 3b) are the sidebands, displaced on either side by 0.2 Hz from the cardiac component.

The spectral analysis (PSD) of the PPG signals recorded on a subject who has carried out Pranayama as per the norms shows the following features as can be seen from Fig. 2.16. The cardiac component is much broader occurring at a higher frequency. The low-frequency component around 0.1 Hz is much stronger than the cardiac component. The ratio of low-frequency component to heart component is 6:1 compared to 1:2 for the normal subject. Absence of sinus arrhythmia and transfer of energy to lower frequencies indicating relaxation. The Heart Rate Variability (HRV) spectrum comprises of two peaks, one corresponding to the sympathetic component of the heart rate and the other corresponding to the parasympathetic component. Usually, it is measured in ECG using R peak detection. Similar technique can be used for PPG signal and it is easier to extract HRV from it. The typical HRV spectrum on a normal subject is given in Fig. 2.17. The HRV spectrum of the PPG signal of a subject after he performs Pranayama is given in Fig. 2.18. A comparison of Figs. 2.17 and 2.18 reveals the following features: • Normal subject has two prominent peaks—one centered at 0.1 Hz and the other centered at 0.3 Hz of nearly equal amplitudes. • For the subject doing Pranayama the sympathetic component shifts to much lower frequency and the relative magnitude compared to the parasympathetic component is nearly 3:1. • Thus, it can be concluded that Pranayama breathing exercises as taught in ancient yoga practices bring about certain benign changes in the human physiology.

48

M. Mukunda Rao

Fig. 2.17 HRV spectrum of a normal subject indicating the sympathetic and parasympathetic components

Fig. 2.18 The HRV spectrum of PPG signal from a subject who has completed Pranayama

However, the exact time in of the onset of the sympathetic activity in Pranayama subjects are not known.

2.7 Summary At rest the heart rate increases on inspiration and decreases on expiration. This variation in beat-to-beat interval, which occurs during a respiratory cycle, has been of interest to cardiopulmonary physiologists since the middle of last century. Although this phenomenon is called the respiratory sinus arrhythmia (RSA), rhythmic breathing movements are not required for its appearance.

2 Influence of Controlled Breathing (Pranayama) on Dermal Perfusion

49

Fluctuations in the heart-rate are attributed to modulation of the basic heart-rate by a number of physiological mechanisms such as blood pressure control, thermoregulation, and respiration. The effect of respiration on heart-rate, respiratory sinus arrhythmia (RSA), has long been recognized and many explanations have been proposed. It has been considered as a lung inflation reflex where stimulation of stretch receptors during inflation caused cardiac acceleration. Others have suggested the origin of RSA is within the central nervous system and is due to interaction between the respiratory centers and the autonomic control of heart-rate. Respiratory sinus arrhythmia (RSA) is mainly due to respiratory modulation of the baroreceptor reflex. The origin of the respiratory modulation remains unsettled. It may arise peripherally from intrathoracic pressure changes causing increased filling of the heart and stimulation of low-pressure receptors in the atria. Alternatively, a direct inhibition arising from the central respiratory rhythm maybe the cause. Indeed both mechanisms might play a part depending on the experimental circumstances thus resolving some of the experimental contradictions. Many Yogic methods emphasis control and suspension of breath as important components on the path to transcendence. However, the Pranayamas are varied and their role is to both improve the physical aspects of breathing and for calming the mind, the latter being very important in the management of many psychosomatic disorders. Different types of Pranayama seem to influence the brain functioning in specific ways. Since the breath seems to link the body and the mind, it is possible to study this link by studying the effect of Pranayama on some brain functioning.

References 1. https://www.yogasite.com/pranayama.htm 2. J.J. Vidal, Real-time detection of brain events in EEG. Proc. IEEE 65(5), 633–641 (1977) 3. V. Blazek, H.J. Schmitt, Quantitative photoplethysmographie—eine nicht-invasive screeningMeßmethode für die funktionelle Gefäßdiagnostik. Min. Inv. Med. 5(Bd. 3), 123–128 (1994) 4. M.M. Rao, V. Blazek, H.J. Schmitt, real-time detection of brain events using optical sensors: preliminary results. in Advances in vascular imaging and photoplethysmography eds. by V. Blazek, U. Schultz-Ehrenburg. (VDI Verlag Köln, 1966), pp. 111–118. ISBN 3-18-322120-9 5. M.M. Rao, V. Blazek, H.J. Schmitt, Investigations on the influence of breathing on brain activity using optical sensors. Proc. SPIE (USA) 2982, 53–64 (1997) 6. R.R. Ram, V. Perlitz, M.M. Rao, Prominence of low frequency rhythm in the human body during deep breathing as monitored by photoplethysmography, in CNVD—Computer-aided Noninvasive Vasculasr Diagnostics, eds. by V. Blazek, U. Schultz-Ehrenburg. Vol. 3, (Mainz Verlag Aachen, 2005), pp. 57–70. ISBN 3-89653-942-6 7. B.K.S. Iyengar, Light on Pranayama. (Unwin Paperbacks, London, 1983). ISBN 0-04-1490606 8. Breathing. From Wikipedia, the free encyclopedia 9. M.M. Rao, N. Srinivasan, S. Rajagopal, S. Ramamoorthy, Influence of Pranayama on Microcirculation as monitored by optical sensors. Presented at the International conference on Biomedical Engineering—BIOVISION, Dec. 21-24, 2001, IISc/Bangalore, India, Published in its proceedings (2001) 10. R. Kayser, Die exakte Messung der Luftdurchgängigkeit der Nase. Arch. Laryngol. Rhinol. 3, 101–120 (1985)

50

M. Mukunda Rao

11. J. Keuning, On the nostril cycle. J. Inst. Rhinol. 6, 99–136 (1968) 12. H.J. Schmitt, V. Blazek, M. Hülsbusch, Assessment and visualization of different rhythmical phenomena in the skin perfusion obtained by optical sensing (PPG and PPGI), in CNVD— Computer-aided Noninvasive Vascular Diagnostics, eds. by V. Blazek, U. Schulz-Ehrenburg. Vol. 3, (Mainz Verlag Aachen, 2005), pp. 71–78. ISBN 3-89653-942-6 13. M. Weymann, R. Buche, D. Hager, P. Joraschky, Rhythmical changes of the cutaneous blood flow during various relaxation techniques, in Innovations in Computer-Aided Noninvasive Vascular Diagnostics eds. by U. Schultz-Ehrenburg, V. Blazek. (VDE Verlag Düsseldorf, 1988), pp. 69–76. ISBN 3-18-330020-6 14. M. Lambertz, R. Vandenhouten, R. Grebe, P. Langhorst, Phase transitions in the common brainstem and related systems investigated by Nonstationary time series analysis. J. Auton. Nerv. Syst. 78, 141–157 (2000) 15. V. Perlitz, M. Lambertz, B. Cotuk, R. Grebe, R. Vandenhouten, G. Flatten, E.R. Petzold, H. Schmid-Schönbein, P. Langhorst, Cardiovascular rhythms in the 0.15 Hz band: common origin of identical phenomena in man and canine in the reticular formation of the brain stem? Pflügers Archiv. Eur. J. Physiol. 448(6), 579–592 (2004)

Chapter 3

Pulse Oximetry for the Measurement of Oxygen Saturation in Arterial Blood Jagadeesh Kumar V. and K. Ashoka Reddy

Abstract The method of photoplethysmography (PPG) detailed in Chapters 1 and 2 gained enormous prominence due to the development of pulse oximetry. In pulse oximetry, the fact that hemoglobin bound with oxygen (called oxyhemoglobin) and hemoglobin without oxygen (deoxy-hemoglobin or reduced hemoglobin) absorb/reflect light differently is exploited in ascertaining, noninvasively, oxygen saturation in arterial blood. Most pulse oximeters that are in existence today use a couple of PPGs obtained using red and infrared wavelength light sources and calculate oxygen saturation in arterial blood using the red and IR PPGs and an empirical equation. This chapter details the development of pulse oximetry. It describes in detail a couple of novel methods of oxygen saturation calculation using the red and IR PPGs. The methods presented here do not need any calibration to be performed.

3.1 Physiological Signals for Diagnostics Dynamic and static measurements on the physical, electrical, chemical and acoustic signals of a human body will help ascertain its health [1–3]. Whenever a person is affected by a disease or gets injured, one or more of these physical, electrical, chemical and acoustic signals change. However, these physical, electrical, chemical and acoustic signals also change day to day due to natural variations in the life cycle. Moreover, the changes in these signals are a complex combination of different parameters. Hence it is very difficult to delineate the functioning or malfunctioning of the underlying biological parts or processes directly from these signals. The four traditional vital signs, namely, the pulse rate, the respiratory rhythm (and sound), the body temperature and the blood pressure are normally used by medical practitioners all over the globe to assess a patient’s state of health. In the modern times, a fifth vital signal, namely, the oxygen saturation in blood has also gained importance. Today any patient with coronary or pulmonary problems must be evaluated for the oxygen J. Kumar V. (B) · K. A. Reddy Indian Institute of Technology Madras, Chennai, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_3

51

52

J. Kumar V. and K. A. Reddy

saturation in blood. Patients undergoing surgery or recovering after surgery must be monitored for the oxygen saturation in blood.

3.2 Blood and Its Composition Nutrients and oxygen for all the cells of the body are carried by blood. Blood also transfers waste from all the cells of the body to waste disposal centres of the body. Blood is composed of plasma (55%), red blood cells (RBCs in short form and known as erythrocytes, 43%), white blood cells (termed as leukocytes, 1.5%), and platelets (called thrombocytes, 0.5%). Nearly 90% of plasma is made of water with some proteins and other chemicals dissolved in it. RBCs are mainly made of haemoglobin molecules that responsible for the transport of oxygen to various other cells of the body [4]. Typically, each mm3 of blood contains approximately six million RBCs and each RBC is made of about 280 million haemoglobin molecules. Haemoglobin concentration in whole blood lies in the range: 134 gL−1 and 173 gL−1 [5]. The absorption spectrum of the natural chromophore contained in the haemoglobin depends on the number of oxygen molecules attached to a molecule of haemoglobin. Haemoglobin without any oxygen molecule (Hb) absorbs most of the visible light except in the dark blue region, and hence blood in the veins that has carbon dioxide and lacks oxygen has the colour ‘dark blue’. If the haemoglobin carries oxygen (HbO) then the natural chromophore absorbs light at all other wavelengths in the visible spectrum except the red wavelength. That gives the bright red colour to the arterial blood which is rich with oxygen. Blood is circulated throughout the body by the systemic and pulmonary circulation system [6].

3.3 Systemic and Pulmonary Circulation The flow of arterial blood, the replenishment of oxygen and exhaling of carbon dioxide are controlled by the cardio-pulmonary system, comprising: (i) the heart, (ii) the lungs and (iii) the blood vessels (arteries, veins and capillaries). Arteries (except the pulmonary artery, which carries oxygen-depleted blood to the lungs for replenishment of oxygen) carry oxygenated blood from the heart to all parts of the body. Arteries terminate into capillaries and the blood in the capillaries provides oxygen and nutrients to the cells and picks up the waste including carbon dioxide from the cells. The capillaries terminate to small veins and the small veins lead to bigger veins and the blood returns to the right atrium of the heart. The returned venous blood then passes to right ventricle and gets pushed through the pulmonary artery to the lungs. In the lung capillaries, the exhale of carbon dioxide and infusing of oxygen takes place and the oxygen-rich blood from the lungs returns through the pulmonary vein to the heart, thus completing one cycle.

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

53

3.4 Mechanism of Oxygen Exchange Air inhaled during breathing enters the lungs. Oxygen in the air is trapped by the haemoglobin molecules and carbon dioxide from the blood is released to the air that gets exhaled [7]. A heterogeneous collection of gas exchange units called alveoli in the lungs surrounded by large pulmonary capillary beds aid this gas exchange through diffusion. Diffusion of a gas requires differential partial pressure. The inhaled air (at near sea level) is at a total pressure of 101 kPa and is made up of nearly 21% of oxygen, 78% nitrogen (N2 ) and small quantities of carbon dioxide, argon and helium. The partial pressures exerted by the two main gases added together nearly equal the atmospheric pressure. The partial pressure of oxygen (PO2 ) of dry air at sea level is therefore approximately 21 kPa but by the time air passes through the trachea and reaches the alveoli, the PO2 falls to about 13 kPa. Venous blood returning to the lungs has a PO2 of 5 kPa. A thin wall (about 0.5 μm thick) between the pulmonary capillaries and the alveoli permits diffusion of gases. Since PO2 of air in the alveoli is 13 kPa and in the pulmonary capillaries is 5 kPa, oxygen diffuses from alveoli to the blood in the pulmonary capillaries. On the other hand, the reverse partial pressure gradient for carbon dioxide ensures the diffusion of carbon dioxide from the blood to the air trapped in the alveoli. Normally partial pressure of nitrogen in the blood and the alveolar air is about the same hence very little nitrogen is diffused in either direction. Thus, the blood returning to the left side of the heart to be pumped into the systemic circulation is replenished with oxygen. If this process is normal, then PO2 of pulmonary venous blood would be equal to the PO2 in the alveoli. Any malfunction would render the pulmonary vein PO2 to be less than the PO2 in alveoli, resulting in reduced amount of oxygen in the arterial blood. In fact, the amount of oxygen bound to the haemoglobin at any time is dictated by the PO2 to which the haemoglobin is exposed. When the arterial blood enters body cells through capillaries, wherein the PO2 is lower than the arterial PO2 , oxygen is detached from the haemoglobin and enters the cell. The total quantity of oxygen bound to haemoglobin in normal arterial blood is approximately 19 mL per 100 mL of blood at a PO2 of 13 kPa. On passing through tissue capillaries this amount is reduced to 14 mL per 100 mL of blood at a PO2 of 5 kPa. Thus, under normal conditions, about 5 mL of oxygen is consumed by tissues from each 100 mL of blood that passes through tissue capillaries during each cycle. When blood returns to the lungs, approximately 5 mL of oxygen diffuses from alveoli into each 100 mL of blood, bringing back the oxygen dissolved in the blood to normal condition. Nearly 98% of the diffused oxygen gets bounded to haemoglobin molecules and the remaining 2% gets dissolved in plasma. Each haemoglobin molecule is made up of four “heme” (the iron-containing portion of haemoglobin) groups and a protein group, known as “globin” (amino acid chains that form a protein). Each heme unit can carry one oxygen molecule and hence one molecule of haemoglobin can carry up to four molecules of oxygen. When one haemoglobin molecule binds four oxygen molecules, it becomes fully saturated

54

J. Kumar V. and K. A. Reddy

(100%) with oxygen. Each gram of fully saturated haemoglobin contains 1.3 mL of oxygen. However, not all haemoglobin molecules participate in oxygen transport.

3.5 Functional and Dysfunctional Haemoglobin Blood contains several forms of haemoglobin, of which some are useful in oxygen transport and some are not. Functional haemoglobins are those that can carry oxygen and include haemoglobin bounded with oxygen molecules, called oxyhaemoglobin (oxygenated haemoglobin, HbO). Haemoglobin not bounded with any other molecule is called reduced haemoglobin (deoxy-haemoglobin, Hb). Haemoglobin which is incapable of carrying oxygen is called dysfunctional haemoglobin (dyshaemoglobin). These are haemoglobin bounded with molecule(s) other than oxygen. They include carboxyhaemoglobin (COHb) and methaemoglobin (MetHb). COHb is formed when carbon monoxide (CO) bond to haemoglobin. COHb exists in varying degrees because of smoking and urban pollution. The level of COHb may become as high as 45% as a result of smoke inhalation. MetHb is oxidized haemoglobin, and in normal healthy persons will be less than 1% of the total haemoglobin. COHb and MetHb are not capable of binding oxygen and hence cannot aid in oxygen transport. Under normal conditions, HbO and Hb amount to 99% of the total haemoglobin present in the blood.

3.6 Oxygen Saturation Whether a person is sleeping, resting or active, every part of that person’s body requires oxygen. The amount of oxygen required for a part of the body depends on the degree of activity of that part but is never zero. While parts of the body can tolerate deprivation of oxygen for limited periods of time, vital organs cannot withstand reduction in oxygen, even for a very short period. These organs may become irreversibly damaged with reduced oxygen supply. Of these vital organs, the brain is by far the most sensitive to reduction in blood oxygen level and can become dysfunctional if oxygen deficit occurs even for a short period. Hence it is necessary that the amount of oxygen carried by the arterial blood is measured to estimate the level of functioning of various parts of the cardio-pulmonary system. The direct method of measurement of oxygen content in arterial blood is to perform complete “gas analysis” and ascertain the various concentrations of gasses in arterial blood. Such an analysis would require drawing of blood directly from an artery and hence requires the services of a competent surgeon. Alternate methods for the determination of the gas contents of arterial blood without the need for puncturing and drawing blood from an artery have been proposed [8] wherein the amount of oxygen in arterial blood is indirectly measured in terms of oxygen saturation in arterial blood [9, 10].

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

55

The functional oxygen saturation in arterial blood is a measure of how much oxygen the arterial blood is carrying as a percentage of the maximum it could carry. If N Hb is the concentration (in molL−1 ) of haemoglobin without oxygen (reduced haemoglobin, Hb) and N Hb represents the concentration of oxygenated haemoglobin (oxyhaemoglobin, HbO) in arterial blood, then % SaO2 =

N H bO 100 % N H b + N H bO

(3.1)

Functional oxygen saturation (% SaO2 ) in Eq. (3.1) is derived assuming the concentrations of dysfunctional haemoglobins, namely, carboxy-haemoglobin and methaemoglobin in arterial blood are negligible [11]. If both functional and dysfunctional haemoglobin are present in the arterial blood, then oxygen saturation is expressed as a fractional oxygen saturation and is given by. % Fractional SaO2 =

N H bO 100 % N H b + N H bO + N M H b + NC O H b

(3.2)

Here NC O H b and N M H b are the concentrations of carboxyhaemoglobin and methaemoglobin. Rearranging Eq. (3.1) results in % SaO2 =

N H bO /N H b Q 100 100 = 1 + N H bO /N H b 1+ Q

(3.3)

Where Q = N H bO /N H b , the ratio of oxyhaemoglobin to reduced haemoglobin. Under normal physiological conditions, a healthy, non-smoking person should have arterial oxygen saturation (% SaO2 ) between 94 and 100%. If % SaO2 goes below 90% life-threatening complications may arise. Saturations lower than 90% may be caused by chronic obstructive pulmonary disease (COPD), excessive bleeding, smoking and malfunctioning blood vessels, especially capillaries. Once oxygen is supplied to various parts of the body, the returning blood in the veins will be depleted of oxygen. The functional oxygen saturation of venous blood (SvO2 ) is about 75% [12].

3.6.1 Measurement of Oxygen Saturation (Oximetry) Three methods of oximetry are now in clinical use [13]: (i) Invasive (in vitro) Co-Oximetry and arterial blood gas (ABG) analysis. (ii) Invasive fibre optic based oximetry to determine oxygen saturation in arterial flow, mixed arterial-venous flow or intra cardiac flow. (iii) Non-invasive pulse oximetry to monitor arterial oxygen saturation at any part of the body.

56

J. Kumar V. and K. A. Reddy

3.6.2 Arterial Blood Gas Analysis and Co-Oximetry Prior to the widespread use of the present-day non-invasive pulse oximeter, the arterial blood gas (ABG) analysis and Co-Oximetry were the main methods employed for the measurement of arterial oxygen saturation. As its name implies, the ABG test is conducted by taking a blood sample from an artery and performing complete gas analysis. For this purpose, either the radial artery at the wrist or the brachial artery at the elbow would have to be punctured. The common practice was to draw the samples at regular intervals and analyze using in vitro blood gas analyzer. The sampling can be several times a day or even several times an hour. In addition to the SaO2 , the ABG analyzer provides the pH value, PO2 , PC O2 and the bicarbonate concentration. Co-Oximeter or haemoximeter calculates the actual concentrations of the Hb, HbO, COHb and MetHb but again requires a sample of blood drawn from an artery. It works on the spectrometric principle using four different wavelengths of light and is also capable of measuring the fractional SaO2 . The most accurate method available for measuring the four clinically relevant haemoglobin species is the Co-Oximeter. Due to this fact, it is accepted as the “gold standard” against which other methods of oxygen measurement in blood are compared [14]. However, both these methods are invasive and risky as puncturing an artery may result in spasm, excessive bleeding, vessel obstruction and infection [15]. Because both these methods are time-consuming, invasive and provide SaO2 readings only at the times at which the samples are drawn, non-invasive methods that provide continuous measurements were needed. Pulse oximetry is such a non-invasive method that can measure arterial oxygen saturation continuously without extracting blood. Pulse oximetry has now become the most popular method employed for the determination of SaO2 . When SaO2 is determined using the principle of pulse oximetry, it is customary to indicate it as SpO2 . In the method of pulse oximetry SpO2 is computed using a couple of photoplethysmographs [9].

3.6.3 Photoplethysmography Hertzman pioneered Photoplethysmography, a non-invasive electro optic method, that provides information on the blood volume changes at a test site on the body [16]. A Photoplethysmogram (PPG) is the signal extracted from either the reflected or transmitted light, obtained by illuminating a part of the body of interest. Reflective type PPG can be obtained close to the skin on any part of the body. Transmission type PPG can be obtained only from an accessible extremity such as earlobe or finger. Obtaining a PPG employing the reflected light is already dealt with in Chaps. 1 and 2. To obtain a PPG through the transmitted light, a light source of wavelength λ having a constant intensity IINλ is placed on one side of an extremity, say fingertip or earlobe, and the transmitted light I oλ through the finger or earlobe is detected by a

3 Pulse Oximetry for the Measurement of Oxygen Saturation … Fig. 3.1 Sensor for obtaining a PPG signal utilizing the transmitted light through finger

57

IIN

Io

suitable photo detector placed on the side opposite to that of the source as indicated in Fig. 3.1. A typical PPG signal (either reflection type or transmission type), shown in Fig. 3.2, is made of a large DC component. The DC part of a PPG arises out of light from the source interacting, only with skin-muscle-bone and completely missing contact with blood vessels at all and reaching the photo detector. As seen in Fig. 3.2, a PPG also contains a very low frequency component due to light from the source passing through the venous blood, apart from skin-muscle-bone. A third and much smaller component of a PPG is due to light from the source passing through arterial blood vessels apart from skin-muscle-bone. This component will be at the frequency of the heartbeat. Blood volume increases in the arteries just after the systole resulting in the reduction of the received light intensity. On the other hand, blood volume in the arteries decreases during diastole, that results in increase in the received light. Thus, the part of detected signal due to the arterial blood appears pulsatile in nature at the heart rate, as shown in Fig. 3.2. In a typical PPG, about 90% of the detected light comes from skin-tissue-bone (DC). While nearly 9.5% of light travels through venous blood, only about 0.5% of the detected light is from arterial blood volume. Since the pulsatile portion arises due to the light passing through arterial blood and hence the pulsatile signal of a Fig. 3.2 Components of a typical PPG signal

58

J. Kumar V. and K. A. Reddy

PPG has information on the arterial blood flow such as: heart rate, heart rate variability, respiration and blood pressure [17]. In Chaps. 1 and 2, it has been shown that utilizing the slow varying component (due to venous blood) of a single wavelength PPG, diagnosis and monitoring of peripheral vascular haemodynamic and venous dysfunction (thrombosis) can be achieved [18, 19]. In pulse oximetry, two PPG signals, one obtained using a light source in the red wavelength region and the other in the infrared region are employed to determine the level of oxygen saturation in arterial blood non-invasively.

3.6.4 History of Pulse Oximetry In 1935, Karl Matthes showed that using photoplethysmographs obtained at two wavelengths, it is possible to track oxygen saturation [20]. He is now regarded as the father of oximetry. He built the first device to continuously track blood oxygen saturation in vivo by trans-illuminating the tissue. In this method, two wavelengths of light, one in the red region and the other in the green region were used. Later he switched to red and infrared light. Although this method was useful in following the trends in oxygen saturation, calibrating the device to reduce errors in actual measurements on oxygen saturation in arterial blood was very difficult. In 1942, Millikan devised an instrument and coined the term “oximeter” to measure arterial oxygen saturation from the ear of a pilot [21]. During World War II, his oximeter was utilized to regulate the oxygen delivery system to help pilots flying at high altitudes in pressurized cockpits. The ear oximeter proposed by Millikan could not be calibrated, and one had to guess the normal saturation level for each person. Most important subsequent works were performed by Goldie [22], Wood and Geraci [23], that resulted in the improvement of Millikan’s ear oximeter. In 1949, Brinkman and Zijlstra [24] were first to describe the monitoring of SaO2 based upon skin reflectance spectroscopy from the forehead, first in vitro, and then in-vivo. Their innovative idea to use light reflection instead of tissue trans-illumination resulted in monitoring of SaO2 from virtually any part of the body. This was followed by a photoelectric method proposed by Sekelj et al. [25] for SaO2 determination. In 1960, Polanyi and Hehir [26] developed the fibre optic catheter oximeter which is the basis for the modern invasive oximeter. In 1964, a surgeon, Robert Shaw built a self-calibrating ear oximeter, using eight wavelengths between 650 and 1050 nm, to identify and separate Hb species including COHb and MetHb. In early 1970s, Hewlett-Packard improved his method and released the first commercial eight-wavelength ear oximeter (HP 47201A). This oximeter delivered light via a fibre optic cable to a sensor mounted on the ear and employed a heating element to keep the tissue locally perfused with blood. Meanwhile, Cohen and Wardsworth added significant advancements in non-invasive reflectance oximetry [27]. In 1972, Takuo Aoyagi, an engineer working with Nihon Kohden Corporation in Tokyo, Japan, invented the present day two-wavelength pulse oximetry [28]. Aoyagi while trying to develop a non-invasive method to determine cardiac output using cardiogreen dye, measured light passing through the earlobe,

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

59

noted that the light transmitted through the earlobe exhibited pulsatile variations. While attempting to eliminate these variations, he discovered that the ratio of pulsatile signals measured at two different wavelengths could be related to the oxygen levels of the arterial blood. Aoyagi’s method also eliminated the requirement to know the intensity of light that was entering the tissue-under-test. In March 1974 Aoyagi and his team announced the first pulse oximeter model OLV-5100. This was a significant development as it reduced the number of wavelengths necessary down to two for measurement of SpO2 from the eight used in the Hewlett-Packard instrument [29]. The model OLV-5100 employed a tungsten light source and two narrow band filters to obtain red and IR monochromatic lights. Unfortunately, these filters blocked majority of the light intensity from the source, resulting in very low levels of light available for the measurement of oxygen saturation. All these early instruments suffered from one or more of the following drawbacks [30]: (a) Lack of adequate calibration procedures. (b) Difficulty in differentiating tissue, arterial blood and venous blood. (c) Error introduced due to unknown optical path length. In the late 1970s, several groups began developmental work in pulse oximetry using Aoyagi’s idea and fingertip probes were introduced. With subsequent developments in semiconductor technology, leading to the invention of the solid-state devices such as LEDs, photodiodes and microprocessors steered the current era of modern pulse oximetry. LEDs generated required narrowband light with controlled wavelengths, exactly the type of light required to vastly improve the signal quality of oximeters. In 1981, Nellcor and Ohmeda (now GE) introduced commercial pulse oximeters utilizing small LEDs and photodiode mounted directly on the sensor probe applied to the patients. Today, there are many manufacturers producing pulse oximeters with elevated levels of confidence in the readings of oxygen saturation [31]. In course of time, pulse oximetry has revolutionized the concept of clinical monitoring of blood oxygen saturation since electrocardiography. The American Society of Anaesthesiologists (ASA) made the pulse oximeter to be a standard for intraoperative monitoring in 1990 [32]. Since then, pulse oximetry has become the standard technique for monitoring oxygenation during procedural sedation, anaesthesia, post anaesthesia care unit, intensive care unit (including neonatal intensive care unit), and recovery from anaesthesia.

3.7 Principle of Operation of a Pulse Oximeter Pulse oximeters derive their name since they operate on the pulsatile portions of red and IR PPG signals to estimate the oxygen saturation in arterial blood. It is seen from Eq. (3.1), to compute SpO2 , the concentrations of oxy and deoxy haemoglobin, N H bO and N H b , must be known. To extract these concentrations from PPG signals, all the present-day pulse oximeters utilize Beer-Lambert’s law. Beer-Lambert’s law states

60

J. Kumar V. and K. A. Reddy

that the concentration of an absorbing substance in a solution can be determined from the attenuation of light through that solution at a specific wavelength. If the input intensity of light is I IN , then the received light intensity I 0 , after transmission through the light absorbing medium is: Io = I I N e−λ cl

(3.4)

In Eq. (3.4), ελ is the wavelength dependent extinction coefficient (normally expressed in L mmol−1 cm−1 ), c is the concentration of the absorbing solution (mmol L−1 ) and l is the optical path length (cm). The light absorbed while passing through the solution is given by:  A = ln

II N Io

 =∈λ cl

(3.5)

where A is called the absorbance of the medium. It is also referred as the optical density (OD). OD is a dimensionless quantity. If multiple absorbers are present in the path of light, then each absorber contributes its part and the resulting total absorbance AT can be expressed as AT =

k 

(∈λi ci li ).

(3.6)

j=1

Where k represents the number of independent absorbers. Since arterial blood flow is pulsatile, the absorbance due to it will also be a pulsatile signal. The time period of each pulse is dictated by the heartbeat and its amplitude by the concentration of various constituent parts of arterial blood and path length of light travelling through the arteries. In human blood, the main light absorbers are the haemoglobin. Previous research had indicated that oxy and deoxy-haemoglobin have different optical attenuation characteristics [33–35] as given in Fig. 3.3. It is evident from Fig. 3.3 that the most appropriate window of wavelength operation for a pulse oximeter is between 600 and 1000 nm.

3.7.1 Traditional Method of Computation of SpO2 Most of the commercial pulse oximeters employ two LEDs, one emitting red (near 660 nm) light and the other infrared (near 900 nm) light as the sources. Either the transmitted light through an extremity such as fingertip or earlobe or the reflected light at these wavelengths are detected to obtain two PPG signals, say PPGR and PPGIR . The amplitudes of the cardiac synchronous pulsatile portions AC R and AC IR in the red and IR PPG signals (PPGR and PPGIR ) respectively are extracted. Similarly, the

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

61

Fig. 3.3 Absorption spectra of Hb and HbO2

Fig. 3.4 AC and DC components of a typical PPG signal

DC values DC R and DC IR of the red and IR PPG signals are extracted (vide Fig. 3.4). Then a normalized red to IR absorption ratio R is obtained as: R=

AC R /DC R AC I R /DC I R

(3.7)

For obtaining a calibration curve, a group of healthy non-smoking young volunteers are made to breathe hypoxic gas mixtures to regulate their arterial oxygen saturation between 80 and 100% [36, 37]. Samples of their arterial blood are drawn at regular intervals and the oxygen saturation (SaO2 ) values are measured using laboratory Co-Oximeter. At the same time, normalized ratios (R) as given in Eq. (3.8) are calculated from the red and IR PPG signals obtained from the volunteers. The values obtained by Co-Oximeter and ratio of ratios “R” are plotted to realize a calibration curve. A typical calibration curve is shown in Fig. 3.5. Then the relationship between R and % SpO2 is obtained by linear interpolation as: %SpO2 = (K 1 + K 2 R)%

(3.8)

In Eq. (3.8), K 1 and K 2 are calibration constants extracted from the calibration curve through linear interpolation. Hence K 1 and K 2 are dependent on the composition; such as age, gender and ethnicity of volunteers chosen for obtaining the calibration curve of Fig. 3.5 [9]. As a result of this process, majority of commercial

62

J. Kumar V. and K. A. Reddy

% SpO2 = 110 – 25R

Fig. 3.5 A typical pulse oximeter calibration curve showing empirical relationship between actual SaO2 and normalized ratio R

pulse oximeters rely on a fixed calibration curve that was extracted from a group of young volunteers of a given population to compute SpO2 . Erroneous readings in SpO2 can occur when such oximeters are employed for a different set of population [37, 38]. A typical calibration curve used by Ohmeda pulse oximeter [39] is shown in Fig. 3.5. An empirical linear approximation to the calibration curve shown in Fig. 3.5 [9, 39] is given as: SpO2 = (110 − 25R)%

(3.9)

A pulse oximeter manufacturer will use a calibration curve(s) obtained from a group of volunteers. However, it is seen that in most cases that R = 1 results in the oxygen saturation to be ~85%. In a standard pulse oximeter algorithm, once R is calculated from the two PPG signals, SpO2 values are determined utilizing R and a look-up table that has entries corresponding to the empirically derived calibration curve. Normally, a microcontroller sitting inside the pulse oximeter calculates of R from the acquired PPG signals and displays corresponding SpO2 from a look-up table. Consequently, use of different calibration curves extracted from data obtained from different volunteer groups was proposed. Adaptive calibration and signal processing methods for SpO2 estimation have also been proposed to reduce the errors [40]. Attempts were also made to curve-fit the data using logarithm. In a patented algorithm, instead of R, an instantaneous ratio R’ is found from the ratio of derivatives of red and IR PPGs. R’ is again plotted against, SpO2 and from the plot, an empirical formula is arrived at to compute the average SpO2 value. Recently, a method based on two wavelengths close to each other has also been reported. This method too utilizes a slightly modified computation of ratio from red and IR PPGs and calibration constants. In all the patents and the information available in the literature, the ratio of ratios R and an empirical calibration equation for SpO2 computation is used. A couple of novel calibration free methods of computation of SpO2 are explained next.

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

63

3.8 A Model-Based Calibration-Free Method for the Measurement of SpO2 The calibration-free methods of measurement of SpO2 described below are based on a model for the attenuation or reflection of light through or from the body and is applicable for both the transmission and reflectance type of PPG sensors [41, 42]. First the attenuation of light in a typical pulse oximeter sensor is modelled as shown in Fig. 3.6. A processing method is then devised to remove the patient and sensor dependency of the red and IR PPG signals and extract a computational method that uses the parts of the processed PPG that reflect the attenuation only by blood volume changes within the sensor. The expression for calculation of SpO2 utilizing the processed PPG signals provides SpO2 readings directly, dispensing with the necessity of calibration curves. Figure 3.2 portrays a typical PPG and its components. Most photons emitted by the red and IR sources of the sensor head of Fig. 3.1 pass through epidermis-tissue-soft bone-tissue-epidermis of the finger and reach the detector. These photons give rise to the DC component, because the attenuation in this path does not vary within the measurement time. Very small number of photons from the red and IR sources go through the path that includes veins resulting in a very low magnitude very low frequency component at the output of the detector. Few photons also go through arteries resulting in a detected pulsatile AC voltage output at the heart rate. The path of light from the source to the detector in Fig. 3.1 is assumed to be of cylindrical in shape as given in Fig. 3.6 having a cross-sectional area A and length T F , where T F is the thickness of the finger. A disc having thickness dl is considered in this cylinder. The incident light intensity on this disc is il and the attenuation across the disc is dil . The attenuation across this disc is dictated by attenuation of cells in the disc. Depending on the optical property of a cell, the cell absorbs some photons and scatters some. The attenuation due to a cell in the bone with an input intensity of light iiλ and an output light intensity i0λ is portrayed in Fig. 3.7a. The attenuation α Bλ (including scattering) at a wavelength λ is modelled as shown in Fig. 3.7b. Here, AB is the cross-sectional area of the cell. A fraction σ Bλ of the area of the cell AB is made completely opaque. The rest of the cross-sectional area of that cell (AB –σ Bλ ) is made

Fig. 3.6 A model showing possible light interactions with cells in trans-illuminated object (finger)

64

J. Kumar V. and K. A. Reddy

Fig. 3.7 a Attenuation of light due to a bone cell b Binary model of a

to be completely transparent as indicated in Fig. 3.7b. It is to be noted that σ Bλ is wavelength dependent. With this, if we have N B as the cell concentration of bone cells (cells per unit volume), then the total opaque area due to all bone cells in the disc is σBλ N B A dl. Similarly, other cells will contribute to opaque area of the chosen disc and hence the total opaque area Aop due to all cells in the disc is: Aop = −(σ Dλ N D + σT λ N T + σ Bλ N B )A dl

(3.10)

Here σ Dλ and σT λ are the opaque areas in the binary model and represent the attenuation at the wavelength λ of dermis and tissue cells respectively. N D and N T are the dermis and tissue cell concentrations. Then the optical attenuation dil across the disc of thickness dl is: dil =

Total Opaque area il Total Area

Hence, dil =

(σ Dλ N D + σT λ N T + σ Bλ N B )Adl il A

(3.11)

Rearranging and integrating Eq. (3.11) with limits l = 0 to l = T F , we get: l=T F

dil = il

l=0

l=T F

(σ Dλ N D + σT λ N T + σ Bλ N B )dl l=0

Evaluation of the integral results in l=TF F ln(il )|l=T l=0 = (σ Dλ N D + σT λ N T + σ Bλ N B )|l=0

(3.12)

At l = 0, il = I Sλ and at l = T F , il = Ioλ . With these relations, Eq. (3.12) can be simplified as: 

Ioλ ln I Sλ

 = −(σ Dλ N D + σT λ N T + σ Bλ N B )TF

(3.13)

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

65

Here Ioλ is the intensity of the light received at the detector and I Sλ is the intensity of the source emitting light at a wavelength λ. Applying natural logarithm to Eq. (3.13) results in: Ioλ = I Sλ e−(σ Dλ N D +σT λ NT +σ Bλ N B )TF

(3.14)

The output light intensity Ioλ as given in Eq. (3.14) is detected by the photo detector and result in the DC component Vλ = vdλ |dc as: Vλ = K Dλ Ioλ = K Dλ I Sλ e−(σ Dλ N D +σT λ NT +σ Bλ N B )TF

(3.15)

Here K Dλ is the sensitivity of the photo detector. To express the attenuation due to cells in terms of an attenuation coefficient ε (L mol−1 cm−1 ) and the concentrations as mol L−1 , Eq. (3.15) becomes Vλ = K Dλ I Sλ e−(ε Dλ D+εT λ T +ε Bλ B)TF

(3.16)

Here ε Dλ , εT λ and ε Bλ are the extinction coefficients and D, T  and Bare the molar concentrations of epidermis, tissue and bone cells respectively. As brought out earlier, some photons travel through arteries carrying blood. The constitution of blood is: 55% plasma, 43% red blood cells (made of haemoglobin, without oxygen Hb and with oxygen HbO), 1.5% white blood cells (leukocytes) and 0.5% platelets. Plasma has nearly zero optical attenuation at the red and IR wavelengths utilized in pulse oximetry and hence does not contribute to the attenuation by blood. Since the amount of white blood cells and platelets are negligible attenuation of blood is largely dictated by red blood cells alone. In Fig. 3.6, all the individual arteries that are in the path of light are combined as a single equivalent artery. The equivalent artery shown in Fig. 3.6 has height of interaction with light H and has a width x. As blood is pumped by heart, the arteries enlarge and contract at the rate set by the heart. Hence width x is a time-varying quantity. When blood flowing in the artery attains a maximum the value of x attains maximum, say, x . With this, the attenuation di bλ due to arterial blood can be derived as: 

di bλ =

(ε H bλ H b + ε H bOλ H bO)x H dx xH

(3.17)



The intensity of light that passes through equivalent artery of Fig. 3.6 is ibλ . In H b and H bO are molar concentrations of haemoglobin without oxygen and with oxygen respectively. Extinction coefficients of haemoglobin and oxyhaemoglobin are ε H bλ and ε H bOλ [32–34]. Rearranging Eq. (3.17) and evaluating the integral over the range x = 0 to x = x , we get: 

 ln

Ibλ β I oλ



 (ε H bλ H b + ε H bOλ H bO)x  =−  2x x=0



2 x= x



(3.18)

66

J. Kumar V. and K. A. Reddy

In Eq. (3.18) a small fraction (β ≈ 0.001) of light I Sλ from the source, namely β I oλ , falls on the equivalent artery and is attenuated by the arterial blood. From Eq. (3.18) we can derive the pulsatile component vdλ |ac of the output of the detector as: 

vdλ |ac = voλ = K Dλ β Ioλ e

(ε H bλ H b+ε H bOλ H bO)x 2

−



2

x

x= x   

(3.19)

x=0

Applying natural logarithm to Eq. (3.19), we get: x=x (ε H bλ H b + ε H bOλ H bO)x 2  ln(voλ ) = ln(K Dλ β Ioλ )−  2x x=0





(3.20)

We see in Eq. (3.20) that the patient (colour of skin and thickness of finger) and the sensor parameters (intensity of source and sensitivity of the detector) dependent component is a DC value ln(K Dλ β Ioλ ) and can be easily removed. Removing the DC part from Eq. (3.20) we get the pulsatile portion of ln(voλ ) as: ln(voλ )| Pulse

 (ε H bλ H b + ε H bOλ H bO)x  = vλ =  2x



2 x= x



(3.21)

x=0

The peak value V pλ of vλ is: 2V . 2 V pλ = (∈ H bλ H b+ ∈ H bOλ H bO)xˆ and x = ( H bλ H b+ H pλ bOλ H bO) Substituting x in Eq. (3.21) and taking Q = H bO/H b results in: 



vλ =

[(ε H bλ + ε H bOλ Q)]2 H b2 x 2 4V pλ

(3.22)

A typical processed PPG after applying natural logarithm and extracting only the pulsatile component is shown in Fig. 3.8. It is seen that during the climb to systolic (in the period (t 1 to t 2 ), vλ varies linearly with time. This portion is expanded and shown in Fig. 3.8b. Between t 1 to t 2 , vλ can be expressed as: vλ = m λ t|tt21 Q)] where m λ is the slope vλ = [(ε H bλ +ε H bOλ 4V pλ oxygen saturation is given below.

2

H b2 x 2

(3.23) . The procedure to compute the

(1) First natural logarithm is applied to the red and IR PPG signals voR and voIR to get ln(v O R ) and ln(v O I R ). (2) The pulsatile portions vR and vIR respectively are delineated from ln(v O R ) and ln(v O I R ). (3) The peak to peak values VpR and VpIR of vR and vIR are then computed.

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

t1

t2

t1

Time t

67

Time t

t2

Fig. 3.8 a Pulsatile part of the PPG b Linear part from a

(4) The linear portions of vR and vIR are identified and the slopes mR and mIR respectively are computed therefrom. Here,

mR =

[(ε H b R + ε H bO R Q)]2 H b2 x 2 and Vp R

(3.24)

[(ε H bI R + ε H bO I R Q)]2 H b2 x 2 Vp I R

(3.25)

mI R =

In Eqs. (3.24) and (3.25) and are the extinction coefficients of Hb and HbO at red wavelength. Similarly, and are the extinction confidents of Hb and HbO at IR wavelength. The value of Q is then computed by dividing Eq. (3.24) by (3.25) as:   ε H b R m I R V p I R − ε H bI R m R V p R   Q= ε H bO R m R V p R − ε H bO R m I R V p I R

(3.26)

Substituting the value of Q from Eq. (3.26) in Eq. (3.3) results in SpO2% = 

  ε H b R m I R V p I R − ε H bI R m R V p R  100. m I R V p I R (ε H b R − ε H bO I R ) − m R V p R (ε H bI R − ε H bO R ) (3.27)

Equation (3.27) contains neither the patient dependent parameters (skin colour and finger thickness) nor the instrumentation (intensity of the light source, sensitivity of the photo detector and gain of the amplifier). Thus the computation of SpO2 as per Eq. (3.27) is independent of patient dependent variables as well as the characteristics of the instrumentation.

68

J. Kumar V. and K. A. Reddy

3.9 Alternate Method for Computation of SpO2 The pulsatile component v0λ of the PPG signal after applying natural logarithm to a PPG obtained from equation is: v0λ = ln(v0 ACλ )| pulse

 (ε H bλ H b + ε H bOλ H bO2 )x 2  x=xˆ =−  x= 0 2 xˆ

(3.28)

Applying Eq. (3.28) for the outputs of the red and IR photo detectors, we get the pulsatile portions v0R and v0I R as  (ε H b R H b + ε H bO R H bO2 )x 2  x=xˆ v0R = −  x= 0 2 xˆ  (ε H bI R H b + ε H bO I R H bO2 )x 2  x=xˆ v0I R = −  x= 0 2 xˆ The peak-to-peak values V p R and V p I R of v0R and v0I R respectively are: Vp R = Vp I R =

(ε H b R H b + ε H bO R H bO2 )xˆ 2

(3.29)

(ε H bI R H b + ε H bO I R H bO2 )xˆ 2

(3.30)

Dividing Eq. (3.29) by (3.30) results in. Vp R (ε H b R + ε H bO R Q) (ε H b R H b + ε H bO R H bO2 ) = = Vp I R (ε H bI R H b + ε H bO I R H bO2 ) (ε H bI R + ε H bO I R Q) , where Q =

H bO2  H b .

Rearranging the above equation, we get:  Q=

V p I R ε H b R − V p R ε H bI R V p R ε H bO I R − V p I R ε H bO R

 (3.31)

Substituting Q from Eq. (3.31) in Eq. (3.3) we get:  SpO2 =

(V p R ε H bO I R

V p I R ε H b R − V p R ε H bI R − V p I R ε H bO R ) + (V p I R ε H b R − V p R ε H bI R )

 100 % (3.32)

Once again it is easily seen that Eq. (3.32) is devoid of not only patient dependent parameters but also independent of the red and IR source intensities and detector

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

69

sensitivity. Equation (3.32) is simpler than Eq. (3.27) as computation of slopes are not required. A comparison of Eq. (3.9) employed in traditional pulse oximeters and Eq. (3.27) or Eq. (3.32) employed in the present method for computation of SpO2 reveals the following: (1) Here SpO2 is computed without using the ratio of ratios R as well as an empirical equation obtained through calibration. (2) Since the methods described above do not require calibration constants K 1 and K 2 , the methods are ‘calibration-free’. (3) Since calibration is not required, the SpO2 values obtained are not dependent on the composition of volunteers engaged in obtaining constants K 1 and K 2 . (4) To compute SpO2 , only the slope and peak-to-peak values of the AC part of the red and IR PPGs after applying natural logarithm alone are used. The DC part is not at all used and hence the SpO2 values obtained are not influenced by patient and sensor dependent parameters. The details of experiments conducted to validate Eqs. (3.27) and (3.32) and the practicality of the method are given in [41, 42].

3.10 Problems Associated with Pulse Oximetry Using the red and IR PPGs, not only oxygen saturation in arterial blood but also heart rate and heart rate variability in a wide variety of clinical situations can be determined [43]. Pulse oximeters enable quick, continuous and reliable measurement of oxygen saturation at the same time avoid discomfort and risks of an arterial puncture. However, present-day commercial pulse oximeters have shortcomings and research is being carried out all over the globe in removing these shortcomings [44, 45]. The factors that adversely affect the accuracy of pulse oximeter output are: (1) (2) (3) (4) (5) (6) (7) (8) (9)

Probe placement [38], Peripheral vasoconstriction (poor perfusion) [39], Dyshaemoglobins [43], intravascular dyes [44], ambient light [46] nail polish [47, 48], skin pigmentation [49], and reliability problems due to movement of patient [50, 51]. Design of pulse oximeter circuitry and the principle of measurement employed.

As the pulse oximeters use two wavelengths only, they can measure only two haemoglobins and thus, estimate functional saturation, not fractional saturation. Other compounds that absorb light at the same wavelengths will thus introduce errors. Important limitation of pulse oximetry arises due to the presence of abnormal haemoglobins. The most common and potentially most serious is

70

J. Kumar V. and K. A. Reddy

carboxyhaemoglobin, which the oximeter detects as oxyhaemoglobin and thus overestimates the true concentration of oxyhaemoglobin. Other dyshaemoglobins such as methaemoglobin and acraboxyhaemoglobin also interfere [52, 53]. Only Co-Oximetry should be employed if the presence of abnormal haemoglobin is expected. Other parameters such as dyes/pigments also affect pulse oximetry measurements. Methylene blue, which is recommended as a therapy for methemoglobinemia, absorbs light at 660 nm, like the absorption of Hb. If methylene blue is present in blood, then the readings obtained on a pulse oximeter will be erroneous [54]. Similar problems can occur with other intravenous dyes such as fluorescein, indocyanine green, and indigo carmine that are used for therapeutic or diagnostic purposes [55]. Since pulse oximeters estimate SpO2 level using red and IR PPGs, clear noise and artefact free red and IR PPGs are essential. To obtain proper red and IR PPG signals an adequate arterial pulsation is required. Shock, severe hypotension, vasoconstriction or hypothermia can lead to significant decrease in peripheral vascular pulsation and thus can produce unreliable oximeter readings [56]. Another variable that affects pulse oximetry readings can be due to nail polish on a finger or toe probe. It is ascertained that oximeter readings of oxygen saturation are affected by black, blue, and green nail polish. Hence it is better to remove fingernail polish or artificial nails before placing the pulse oximetry probe. Another technique by which the effect due to nail polish could be eliminated is by placing the finger probe side to side instead of top to bottom. For error-free SpO2 estimation in a pulse oximeter, clean noise-free and artefact-free red and IR PPG signals with distinct DC and AC parts are necessary. Errors in the measured values of SpO2 can also arise in situations that produce low signal-to-noise ratio on the detected red and IR PPG signals. Light from sources such as surgical lamps, infrared, xenon and fluorescent lamps can interfere with oximetry accuracy [57]. If excessive ambient light interference is present, then both the red and IR PPG signals will be corrupted. However, a recent study revealed that moderate ambient light has no significant effect on pulse oximetry readings [58]. Even had the differences been statistically significant, the magnitude of the differences was small and thus clinically unimportant. Interference with highintensity ambient light can be easily avoided by placing an opaque material (darkcoloured cloth) to surround the probe. Studies on the effect of skin colour on the pulse oximetry readings revealed that dark skin colour may influence the performance and concluded that accuracy was slightly less for subjects with dark skin colour than those with lighter skin colour [49]. Effects of absorption by platelets and plasma indicate that these have negligible effects on the SpO2 measurement [59]. Movement of a patient introduces artefacts in the red and IR PPG signals and thus will cause inaccurate readings in a pulse oximeter [51]. Since the pulsatile (ac part) component of a PPG is ≈0.1% of total signal amplitude, even a slightest movement of the patient will disturb the PPG. Additional random pulsations will be added to the pulsations due to heartbeat. Such additional random pulsations in a PPG due to motion artefact will result in inaccurate estimation of SpO2 . A straightforward solution is to altogether avoid motion artefacts by securing the probe housing the sources and detectors rigidly to the skin of the patient at a monitoring site such that

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

71

the relative motion between the probe and the patient is arrested even when the patient moves. By suitable processing of the PPG signals, the effect of motion artefact can be reduced on the pulse oximetry readings. A simple solution to reduce error due to motion artefact is to find and display the average of several SpO2 readings.

3.11 Motion Artifact Reduction in PPG Signals In pulse oximetry, computation of SpO2 requires red and IR PPG signals. To obtain a PPG signal, a sensor with two sources (one in the red wavelength and the other in the IR wavelength) and a photo detector is placed on the body. If the PPG sensor is of the transmission type, then the sources are kept on one side of an extremity such as finger or toe and the detector on the opposite side (the finger or toe is sandwiched between the sources and the detector). On the other hand, in reflective type PPG sensors, both the sources and the detector are kept on the same plane. In both cases the magnitude of the PPG obtained is dictated by the amount of light from the source that is coupled to the body and the amount of light (transmitted or reflected) emanating from the body, coupled to the detector. In order to ensure maximum light is coupled to the body part, both the sources and the detector must be in close contact with the body. To ensure close contact, the sources and the detector need to be pressed against the body. However, applying pressure on the body causes: (i) an increase in the temperature of the region underneath the sensor resulting in pain and sweating and hence discomfort to the patient. (ii) Pressure constricts the blood vessels, leading to reduced blood perfusion, resulting in a reduction in the PPG signal. Hence a PPG sensor in any pulse oximeter is designed to exert just bare minimum pressure required to make a contact between the body and the sensor. Any movement by a patient connected to a pulse oximeter results in variations in the contact between the sensor and the patient’s body. Variations in the contact corrupt the red and IR PPG signals obtained during such periods of movement with motion artefacts. SpO2 computed using the motion artefact corrupted PPG signals would be erroneous. Most pulse oximeters manufacturers solve this problem by simply suppressing the output (with or without indication on the front panel of the oximeter) during such periods. Some pulse oximeters simply display the last valid SpO2 reading available during such periods of movements. Even this simple technique of suppression during periods, where the computation of SpO2 is not possible, requires that the oximeters are made capable of recognizing such periods. In other words, most of the processing in present day pulse oximeters is geared towards artefact recognition rather than artefact reduction. In order to obtain uninterrupted readings from a pulse oximeter, the red and IR PPG signals must be processed with a view to remove or reduce motion artefacts, if any, present in the PPG signals. Several methods have been proposed to reduce the influence of motion artefacts from corrupted PPG signals. A popular and common technique employed for the reduction of the effect of the motion artefact in

72

J. Kumar V. and K. A. Reddy

a PPG is the moving average method [9, 39]. The moving average method works well only for a limited range of artefacts [60]. When the spectra of motion artefact and that of the PPG signal overlap significantly, motion artefact becomes in-band noise [61]. In-band noise can be reduced to a large extant with the use of adaptive filters [62–67]. However, to apply adaptive filter technique, a reference signal that is required. The reference signal must be strongly correlated with the signal but uncorrelated with artefact. It is also possible to remove artefact with a reference signal that is strongly correlated with artefact but uncorrelated with the signal. It is possible to obtain a reference signal correlated with motion artefact by employing additional motionsensing hardware. Synthetic reference signal estimated from the artefact-free part of a PPG signal [63] can be used for reducing motion artefact. The Masimo SET® [67] uses the fact that any motion artefact affects both the arterial pulsations and venous blood volume change in a similar manner. Utilizing this fact, a reference signal is extracted from the venous component of a PPG. Such a technique avoids the necessity of extra hardware to sense motion artefact. A signal processing technique that uses multi-rate filter bank and a matched filter provides better performance compared to the moving average and adaptive filtering approaches [67]. Biological signals such as the PPG are generally non-stationary and their properties change substantially over time, mostly due to their dynamic nature. Thus use of time-frequency methods like the wavelet transforms [68, 69] and smoothed pseudo-Wigner-Ville distribution [70] that are best suited for processing non-stationary signals have been applied to process PPG signals to obtain significant improvements compared to traditional approaches. An artefact reduction methodology that uses a physical artefact model coupled with an inversion technique (nonlinear optical receiver) has been proposed [71]. The model-based approach for reduction of motion artefacts requires an additional source-detector pair, resulting in the three-wavelength probe [72]. Exploiting the independence between a PPG signal and motion artefact signal, it is possible to reduce motion artefact using the Independent component analysis (ICA) technique. It has been shown that a third-order ICA applied on the time-derivative of a PPG signal provides artefact suppression for pulse oximetry [73–75]. Compared to ICA alone, ICA in conjunction with block interleaving and low pass filtering provides better performance [73]. But a study showed that motion artefacts, arterial and venous components of a PPG signal are not statistically independent [74]. It was also shown that wavelet transform and adaptive filtering techniques introduce phase shifts in the processed PPG signals and hence these methods have limited application in restoring motion artefact corrupted PPG signals, especially when PPG signals are used for estimation of heart rate (HR) and pulse transit time (PTT) [75]. A motion artefact reduction method based on singular value decomposition (SVD), that extracts clean artefact-free PPG signals from artefact riddled PPG signals preserving all the essential morphological features required is described next [76].

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

73

3.11.1 SVD for Motion Artefact Reduction In this method, K samples of a PPG signal corresponding to few cycles of heart rate are arranged in matrix form, say matrix X with m rows and n columns. The matrix thus obtained s subjected to singular value decomposition.

3.11.1.1

Singular Value Decomposition

The matrix manipulation tool of linear algebra called singular value decomposition (SVD) was proposed in 1870. In SVD, a matrix X with m rows and n columns containing real-valued data is decomposed into three sub-matrices, U, S and V. Here it is assumed that m ≥ n and hence rank (p) of X ≤ n. Applying SVD to X we get [76]: X = U SV T

(3.33)

In Eq. (3.33), U is an m x n matrix, S and V are n x n matrices. While S is a diagonal matrix with (S = diag(s1 ,…,sn )), U and V are unitary matrices. That is U T U = I and V T V = I, where I is an identity matrix. Here, (s1 , …, sn ) are called the singular values (SV). SV represents the positive square roots of the eigen values of the matrix X T X. Also sq > 0 for 1 ≤ q ≤ p and S q = 0 for (p + 1) ≤ q ≤ n. The columns of U are called left singular vectors of X, while the columns of V are called right singular vectors of X. An important observation is that as the singular values decay rapidly, with, S 1 ≥ S 2 ≥ … ≥ S n ≥ 0, we can expect that there will be a good lower rank approximation ( X ) to X by setting the small singular values to zero. The lower rank approximation X is given by. X=

j 

Ui Si viT

(3.34)

i=1

where S i assumed to be zero for i > j; ui and vi are the jth columns of U and V respectively. The singular values of a given data matrix also contains information about the noise level in the data, energy and rank of the matrix. This fact is exploited for signal processing such as data compression, noise reduction and pattern matching and extraction). This feature can also be exploited to remove motion artefacts from corrupted PPG signals.

3.11.1.2

Principal Component Extraction Using SVD

Let X be an m by n matrix formed with data (samples) of a perfectly periodic waveform x(k), k = 1,2,…,mn having a period of n samples. If we place data such that

74

J. Kumar V. and K. A. Reddy

each row of X is made of n (period of the input) samples as given below, ⎡ ⎢ ⎢ ⎢ ⎣

x(1) x(n + 1) .. .

x(2) x(n + 2) .. .

⎤ . . . x(n) . . . x(2n) ⎥ ⎥ ⎥ .. ⎦ .

(3.35)

x((m − 1)n + 1) x((m − 1)n + 2) . . . x(mn)

Applying SVD to X we get the diagonal matrix S to be a rank one matrix with the first entry alone nonzero and all other entries (S 2 to S n ) will be zero. Rephrasing, we can say that if samples of a periodic signal are formed into a matrix X such that the rows contain samples of an integral number of periods of the signal, then the dominant first singular valueS 1 obtained after applying SVD to matrix X indicates a strong periodic component in the rows of the data matrix. It can also be shown that the periodic signal can be recovered by computing s1 u2 vT2 . Here u1 and v1 are column matrix extracted from the first columns of the corresponding U and V (called left singular vector and right singular vector). If noise is present in the signal or if the signal is quasi-periodic, then after applying SVD we will get nonzero values for to S 2 to S n . However, if we obtain a matrix X r with a row length r containing samples of integral number of periods of the signal or the dominant frequency (principal component) of a quasiperiodic signal, then the ratio of first two singular values S 1r /S 2r for X r will be a maximum. Using this artefact-free PPG can be extracted from a motion artefact corrupted PPG. The steps involved are: (1) Take M samples of the data, say x(k), k = 1…M of the PPG signal to be processed. (2) Prepare different matrices X 1 , X 2 …X r …X n , of different row lengths using x(k) such that the row lengths are in the range 0.8 Hz to 2 Hz (the dominant frequency PPG will be in this range) (3) Perform SVD on each of these matrices X 1 , X 2 …X r …X n (4) For each SVD, compute the singular value ratio (SVR) of the first two singular values. Here SVR = .s1 /s2 (5) Plot the SVR against the row length r to obtain a graph called the SVR spectrum of the signal. (6) From the SVR spectrum, determine the value of row length, say r, for which the SVR (s1r /s2r ) is maximum as seen in a typical SVR spectrum shown in Fig. 3.9. (7) From the SVD of the corresponding data matrix X r reconstruct the PPG signal by taking the average of all rows of s1 u1 v1 t . It should be noted here that the averaging process eliminates not only artefact present in the PPG but also eliminates any additive noise in the PPG. Once one cycle of a PPG is recovered, a kind of roll over procedure can be implemented by removing the first r samples from and appending r new samples at the end. For example, if the initial samples are x(k), k = 1…M then, after extracting the first PPG cycle, x(k) is modified as: x(k), k = (r + 1) … (M + r). Once again steps 2

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

75

16

7

x 10

6.5

SVR

6 5.5 5 4.5 4 120

140

160

180

220 200 Row length

240

260

280

300

Fig. 3.9 Typical SVR spectrum obtained from a PPG

to 7 are performed to get the second cycle of the PPG. This process can be continued endlessly to obtain an artefact and noise-free PPG, cycle after cycle. Figure 3.10 shows a typical example of applying SVD for removing motion artefact from a PPG.

Fig. 3.10 a PPG corrupted with artefact b PPG recovered using SVD method c Dotted line shows the SpO2 computed from corrupt red and IR PPG signals, solid line indicates the SpO2 computed using recovered PPG signals

76

J. Kumar V. and K. A. Reddy

References 1. E.N. Bruce, Biomedical Signal Processing and Signal Modeling (John Wiley & Sons, NY, 2001) 2. R.M. Rangayyan, Biomedical Signal Analysis: A Case Study Approach, IEEE Press Series on Biomedical Engineering (John Wiley & Sons, Singapore, 2002) 3. L.S. Sornmo, P. Laguna, Bioelectrical Signal Processing in Cardiac and Neurological Applications (Elsevier academic press, USA, 2005) 4. W.F. Ganong, Review of Medical Physiology, 16th edn. (Appleton & Lange, Norwalk, CT, 1993) 5. D.L. Kasper, Harrison’s principles of internal medicine, 16th ed. (McGraw-Hill, NY, 2005) 6. K. John, J. Li, Dynamics of vascular system, Series on bioengineering & biomedical engineering, vol 1 (World scientific publishing Co. Pte. Ltd, Singapore, 2004) 7. R.D. Miller, Miller’s anesthesia, 6th edn. (Elsevier Churchill Livingstone, Philadelphia, 2005) 8. J.W. Severinghaus, P.B. Astrup, History of blood gas analysis-VI. Oximetry. J Clin Monit 2(4), 270–288 (1986) 9. J.G. Webster, Design of Pulse Oximeters (Taylor & Francis Group, NY, 1999) 10. J.A. Dorsch, S.E. Dorsch, Understanding Anaesthesia Equipment (Williams & Wilkins, Baltimore, 1999) 11. G.D. Baura, System Theory and Practical Applications of Biomedical Signals, IEEE Press Series on Biomedical Engineering (John Wiley & Sons, NJ, 2002) 12. T. Ahrens, K. Rutherford, Essentials of Oxygenation (Jones & Barlett, Boston, 1993) 13. Y. Pole, Evolution of the pulse oximeter. Int Congress Series 1242, 137–142 (2002) 14. C. Secker, P. Spiers, Accuracy of pulse oximetry in patients with low systematic vascular resistance. Anesthesia 52(2), 127–130 (1998) 15. A.J. Williams, ABC of oxygen: Assessing and interpreting arterial blood gases and acid-base balance. BMJ 317, 1213–1216 (1998) 16. A.B. Hertzman, The blood supply of various skin areas as estimated by the photoelectric plethysmograph. Am J Physiol 124, 328–340 (1938) 17. K. Nakajima, T. Tamura, H. Miike, Monitoring of heart and respiratory rates by Photoplethysmography using a digital filtering technique. Med Eng Phys 18(5), 365–372 (1966) 18. V. Blazek, U. Schultz-Ehrenburg, Quantitative Photoplethysmography: Basic Facts and Examination Tests for Evaluating Peripheral Vascular Functions (VDI Verlag, Düsseldorf, 1996) 19. S. Sarin, D.A. Shields, J.H. Scurr, P.D. Smith, Photoplethysmography: a valuable noninvasive tool in the assessment of venous dysfunction ? J Vasc Surg 16(2), 154–162 (1992) 20. K. Matthes, Untersuchungen über die Sauerstoffsättigung des menschlichen Arterienblutes— Studies on the Oxygen Saturation of Arterial Human Blood. Naunyn-Schmiedeberg’s Archives of Pharmacology (in German) 179(6), 698–711 (1935) 21. G.A. Millikan, The oximeter, an instrument for measuring continuously the arterial saturation of arterial blood in man. Rev Sci Instrum 13, 434–444 (1942) 22. E.A.G. Goldie, Device for continuous indication of oxygen saturation of circulating blood in man”, J Sci Instrum 19, 19–23 (1942) 23. E.H. Wood, J.E. Geraci, Photoelectric determination of arterial oxygen saturation in man. J Lab Clin Med 34, 387–401 (1949) 24. R. Brinkman, W.G. Zijlstra, Determination and continuous registration of the percentage oxygen saturation in clinical conditions. Arch Chir Neerl 1, 177–183 (1949) 25. P. Sekelj, A.L. Johnson, H.E. Hoff, P.M. Scherch, A photoelectric method for the determination of arterial oxygen saturation in man. Amer Heart J 42, 826–848 (1951) 26. M.L. Polanyi, R.M. Hehir, New reflection oximeter. Rev Sci Instru 31(4), 401–403 (1960) 27. A. Cohen, N.A. Wardsworth, Light emitting diode skin reflectance oximeter. Med Biol Eng 10, 385–391 (1974)

3 Pulse Oximetry for the Measurement of Oxygen Saturation …

77

28. T. Aoyagi, M. Kishi, K. Yamaguchi, S. Watanabe, Improvement of an ear-piece oximeter, in Proc. Abstracts 13th Annu. Japanese Soc Med Electro, Biol Eng (JSMEBE, Osaka, 1994), pp. 90–91 29. E.B. Merrick, T.J. Hayes, Continuous, non-invasive measurements of arterial blood oxygen levels. Hewlett-Packard J 28(2), 2 – 9 (1976) 30. S. Takatani, J. Ling, Optical oximetry sensors for whole blood and tissue. IEEE Eng Med Biol Mag 347–357 (1994) 31. R.J. Falconer, B.J. Robinson, Comparison of pulse oximeters: accuracy at low arterial pressure in volunteers. BJA 65(4), 552–557 (1990) 32. F.W. Cheney, The ASA closed claims study after the pulse oximeter. ASA Newsletter, vol. 54 (1990) 33. B.L. Horecker, The absorption spectra of hemoglobin and its derivatives in the visible and near infra-red regions. J Biol Chem 148(1), 173–183 (1943) 34. W.G. Zijlstra, Buursma A and Meeuwsen-van der Roest WP, “Absorption spectra of human fetal and adult oxyhemoglobin, de-oxyhemoglobin, carboxyhemoglobin and methemoglobin.” Clin Chem 37(9), 1633–1638 (1991) 35. J.G. Kim, M. Xia, H. Liu, Extinction coefficients of hemoglobin for near-infrared spectroscopy of tissue. IEEE Eng Med Biol Mag 24(2), 118–121 (2005) 36. Y. Mendelson, Pulse oximetry: theory and applications for noninvasive monitoring. Clin Chem 38(9), 1601–1607 (1992) 37. J.E. Sinex, Pulse oximetry: Principles and limitations. Am J Emerg Med 17(1), 59–68 (1999) 38. J.P. Decock, L. Tarassenko, Pulse oximetry: Theoretical and experimental models. Med Biol Eng Comput 31, 291–300 (1993) 39. M.W. Wukitsch, M.T. Petterson, D.R. Tobler, J.A. Pologe, Pulse oximetry: Analysis of theory, technology, and practice. J Clin Monit 4(4), 290–301 (1988) 40. T.L. Rusch, R. Sankar, J.E. Scharf, Signal processing methods for pulse oximetry. Comput Biol Med 26(2), 143–159 (1996) 41. K.A. Reddy, B. George, N.M. Mohan, V.J. Kumar, A novel method of measurement of oxygen saturation in arterial blood, in Proc. I2MTC (Victoria, BC, Canada, 2008), pp. 1627–1630 42. K.A. Reddy, B. George, N.M. Mohan, V.J. Kumar, A novel calibration-free method of measurement of oxygen saturation in arterial blood. IEEE Trans Meas and Instrum. 58(15), 1699–1705 (2009) 43. M. Brown, J.S. Vender, Noninvasive oxygen monitoring. Crit Care Clin 4(3), 493–509 (1998) 44. W.A. Bowes, B.C. Corke, J. Hulka, Pulse oximetry: a review of the theory, accuracy, and clinical applications. Obstet Gynecol 74(3 Pt 2), 541–546 (1989) 45. A. Huch, R. Huch, V. Konig, M.R. Neuman, D. Parker, J. Yount Lubbers, Limitations of pulse oximetry. Lancet 1(8581), 357–358 (1988) 46. D. Amar, J. Neidzwski, A. Wald, A.D. Finck, Fluorescent light interferes with pulse oximetry. J Clin Monit 5(2), 135–136 (1989) 47. C.J. Cote, E.A. Goldstein, W.H. Buschman, D.C. Hoaglin, The effect of nail polish on pulse oximetry. Anesth Analg 67(7), 683–686 (1988) 48. M.M. Chan, M.M. Chan, E.D. Chan, What is the effect of fingernail polish on pulse oximetry? Chest 123(6), 2163–2164 (2003) 49. A.L. Ries, L.M. Prewitt, J.J. Johnson, Skin color and ear oximetry. Chest 96(2), 287–290 (1989) 50. J.A. Langton, C.D. Hanning, Effect of motion artefact on pulse oximeters: evaluation of four instruments and finger probes. Br J Anaesth 65, 564–570 (1990) 51. J.L. Plummer, A.Z. Zakaria, A.H. Ilsley, R.R.L. Fronsko, H. Owen, Evaluation of the influence of movement on saturation readings from pulse oximeters. Anesthesia 50, 423–426 (1966) 52. S.J. Barker, K.K. Tremper, J. Hyatt, Effects of methemoglobinemia on pulse oximetry and mixed venous oximetry. Anesthesiology 70(1), 112–117 (1989) 53. W.P. Bozeman, R.A. Myers, R.A. Barish, Confirmation of pulse oximetry gap in carbonmonoxide poisoning. Ann Emerg Med 30, 608–611 (1987) 54. A. Sidi, D.A. Paulus, W. Rush, N. Gravenstein, F.R. Davis, Methylene blue and indocyanine green artifactually lower pulse oximetry readings of oxygen saturation: Studies in dogs. J Clin Monit 3(4), 249–256 (1987)

78

J. Kumar V. and K. A. Reddy

55. M.S. Scheller, Unger RJ and Kelner, Effects of intravenously administered dyes on pulse oximetry readings. Anesthesiology 65(5), 550–552 (1986) 56. L.M. Schnapp, N.H. Cohen, Pulse oximetry: uses and abuses. Chest 98(5), 1244–1250 (1990) 57. N.S. Trivedi, A.F. AGhouri, N.K. Shah, E. Lai, S.J. Barker, Effects of motion, ambient light and hypoperfusion on pulse oximeter function. J Clin Anesth 9(3), 179–183 (1997) 58. R.R. Fluck Jr., C. Schroeder, G. Frani, B. Kropf, B. Engbretson, Does ambient light affect the accuracy of pulse oximetry? Resp Care 48(7), 677–680 (2003) 59. M. Meinke, G. Muller, J. Helfmann, M. Friebel, Optical properties of platelets and blood plasma and their influence on the optical behavior of whole blood in the visible to near infrared wavelength range. J Biomed Opt 12(1), 014024 (2007) 60. https://sine.ni.com/nips/cds/view/p/lang/en/nid/14040. 61. S.J. Barker, N.K. Shah, The effects of motion on the performance of pulse oximeter in volunteers. Anesthesiology 85(4), 101–108 (1996) 62. A.B. Barreto, L.M. Vicente, I.K. Persad, Adaptive cancellation of motion artifact in photoplethysmographic blood volume pulse measurements for exercise evaluation. Proc IEEEEMBC and CMBEC 2, 983–984 (1995) 63. A.R. Relente, L.G. Sison, Characterization and Adaptive Filtering of Motion Artifacts in Pulse Oximetry Using Accelerometers (Houston, TX, Proc. Conf. EMGS/BMES, 2002), pp. 1769– 1770 64. K.W. Can, Y.T. Zhang, Adaptive reduction of motion artifact from photoplethysmographic recordings using a variable step-size LMS filter. Sensors 2, 1343–1346 (2002) 65. M. Goetzee, Z. Elghazzawi, Noise-resistant pulse oximetry using a synthetic reference signal. IEEE Trans Biomed Engg 47(8), 1018–1026 (2000) 66. J.M. Goldman, M.T. Petterson, R.J. Kopotic, J.S. Barker, Masimo signal extraction pulse oximetry. J Clin Monit Comput 16(7), 475–483 (2000) 67. J. Lee, W. Jung, I. Kang, Y. Kim, G. Lee G, Design of filter to reject motion artifact of pulse oximetry. Comput Stand Interf 26(3), 241–249 (2004) 68. C.M. Lee, Y.T. Zhang, Reduction of motion artifacts from photoplethysmographic recordings using a wavelet denoising approach. Proc. IEEE EMBS Asian-Pacific Conf on Biomed Engg, 194–195 (2003) 69. Y.J.A. Foo, Comparison of wavelet transformation and adaptive filtering in restoring artifactinduced time-related measurement. Biomed Signal Proc Control 1, 93–98 (2006) 70. Y.S. Yan, C.C. Poon, Y.T. Zhang, Reduction of motion artifact in pulse oximetry by smoothed pseudo Wigner Ville distribution. J Neuro engineering Rehabil 2:3 (2005). Available from https://www.jneuroengrehab.com/content/2/1/3. 71. M.J. Hayes, P.R. Smith, Artifact reduction in photoplethysmography. Appl Opt 37(31), 7437– 7446 (1998) 72. M.J. Hayes, P.R. Smith, A new method for pulse oximetry possessing inherent insensitivity to artifact. IEEE Trans Biomed Engg 48(4), 452–461 (2001) 73. B.S. Kim, S.K. Yoo, Motion artifact reduction in photoplethysmography using independent component analysis. IEEE Trans Biomed Engg 53(3), 566–568 (2006) 74. P.F. Stetson, Independent component analysis of pulse oximetry signals, in Proc 26th Annual Intern Conf IEEE EMBS (San Francisco, CA, 2004), pp. 231–234 75. J. Yao, S. Warren, A short study to assess the potential of independent component analysis for motion artifact separation in wearable pulse oximeter signals, in Proc 27th Annual Conf IEEE Engg Med and Biol (Shanghai, China, 2005), pp. 3585–3588. 76. K.A. Reddy, V.J. Kumar, Motion artifact reduction in photoplethysmographic signals using singular value decomposition, Proc. 24th Int. Conf. IEEE (IMTC-2007, Warsaw, Poland, 2007), pp. 1–4

Chapter 4

Reflective Arterial Pulse Oximetry for New Measuring Sites and Long-Term Assessment of Dermal Perfusion Boudewijn Venema

Abstract Until 2010, photoplethysmography (PPG) was almost exclusively used commercially as a measurement method for vital signs in medical diagnostics. Most of the conventional sensing systems were placed on the finger, for the noninvasive recording of cardiologically synchronous blood volume fluctuations and arterial oxygen saturation. However, as the new decade entered, a clear trend became apparent. New applications of PPG were developed and marketed for the first time in medical diagnostics as well as in other areas of life. In addition to medical diagnostics, the basic technology of PPG can now be found in health care, life science and other areas. It is clear that vital signs will become an integral part of the information age and that PPG, as a non-invasive technology, will always play a key role in data collection. The research activities presented in this chapter on the topic of new application sites of the PPG, conducted by the Chair of Medical Information Technology at RWTH Aachen University, are in the context of this evolutionary process. The described in-ear PPG system not only brings significant physiological advantages, but also opens up new possibilities outside of clinical routine. The validation through various studies carried out on the dynamics of skin perfusion and arterial oxygen saturation in the external auditory canal show fundamental findings and principles with regard to the in-ear PPG. The chapter thus places the PPG in the context of new fields of application: the research and development activities began with an initial vision and, in retrospect, are a piece of the puzzle in the evolution of photoplethysmography as it exists today. In addition, due to the general validity of the study results, this chapter can be an inspiration and basis for future developments and trends of PPG.

B. Venema (B) Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_4

79

80

B. Venema

4.1 Exploring New Application Sites for PPG Conventional finger-based photoplethysmography (PPG) systems, like finger pulse oximeters, are widely used. They belong to the standard devices in almost all areas of clinical facilities, from examination rooms to operating theaters. Based on decades of research and development, they provide accurate measurement of heart rate and arterial oxygen saturation (SaO2 ), are inexpensive, easy to use, and fully accepted by medical staff. Therefore, it is noteworthy that PGG has not yet extended beyond the clinical realm to other areas of application where visualization of vital signs has long been of considerable interest. One typical application scenario is that of ‘home care’. This type of health care is based on the principle ‘outpatient before inpatient’ and aims to provide home-related health care and therapy. Due to an increasingly aging society, rising healthcare costs have become a major challenge in many countries. Therefore, the concept of home care is becoming more important in medical technology and will probably play an important role in the future. It is expected that the outsourcing of healthcare from clinical facilities to home-related home care will lead to a significant reduction in costs [1]. Another application scenario is covered by the term ‘personal health care’ (PHC). PHC signifies accessing, monitoring, and/or recording of vital parameters that are independent of the usual professional medical attendance. Although PHC is a relatively new term, many people have always (to some extent) been reliant on self-help, often due to lack of adequate medical care. Nevertheless, due to improved technical possibilities, PHC has become increasingly important for various health departments and users [1]. Whereas healthy users become more interested in their vital signs during sports activities (ranging, e.g., from jogging to high-altitude mountain expeditions), an ever-aging society hopes for some degree of medical independence based on ambient-assisted living. In addition, rural regions and developing countries will also benefit from robust health information infrastructure [2]. For this special purpose, these sensors must be unobtrusive and easy in use, since the correct application may sometimes be difficult to control. Although PPG has considerable potential for the monitoring of vital signs in PHC and home care, this has not yet been achieved. One reason for this is the limited interest in measurement of peripheral capillary oxygen saturation (SpO2 ) and the slow blood volume shifts in non-clinical environments. For heart rate monitoring the ECG is generally used (e.g. long-term ECG, sports application) even though this requires distributed electrodes and is, therefore, more complicated to use. The preference for ECG is mainly due to two problems associated with conventional PPG: (i) compared to the ECG which derives information on the heart based on the electrical activity of the heart, PPG measures sub-dermal blood volume changes that are highly susceptible to contact pressure variations caused by motion (i.e. motion artifacts), and (ii) PPG is usually applied at the periphery (e.g. a finger), which is always strongly affected by motion artifacts during daily activity. Therefore, it is useful to apply PPG to body parts that are unobtrusive and less affected by motion during daily activity. Until the late 1960s and early 1970s, the

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

81

measurement technique available only allowed transmissive PPG measurement that can be applied to body parts that are transparent for the measurement light, i.e., finger, earlobe, or tooth, as well as arms, legs, and feet in infants. Although it is known that reflective PPG (rPPG) is feasible because scattering predominates over absorption in human tissue, the pulsating arterial component of the measurement signal is about ten times smaller compared to the pulsating signal component measured in the transmissive measurement mode and, therefore, a more accurate measurement technique was required. Fortunately, technological progress substantially improved the analog and digital signal processing thereby making rPPG possible. This measurement method is no longer limited to a few body parts but can be applied on any skin surface that offers sufficient dermal perfusion [12]. A new wearable sensor concept can be developed with new possibilities for mobile monitoring and for routine clinical use with regard to unobtrusiveness, measurement robustness, and suitability for everyday application.

4.2 Motivation for PPG Measurement in the Inner/Outer Ear Channel For body sensors intended for use in mobile vital sign monitoring, the parameters ‘diagnostic reliability’ and aspects of ‘usability’ must be considered. Regarding inear PPG sensors, both requirements seem to be in balance. The ear channel provides both physiological and user-related advantages; both these aspects are discussed below [13].

4.2.1 In-Ear PPG from a Physiological Viewpoint From the physiological viewpoint, the ear channel is an ideal measurement side. Between the heart and the ear-channel cardiovascular diseases are very rare. In contrast, the peripheral system often suffers from insufficient perfusion. This can be due to the influence of negative temperature, or to cardiovascular diseases (e.g., as a result of diabetes), blood stasis, or centralization of blood flow. For PPG sensors that are attached to the periphery (e.g., the fingertip), these contraindications often lead to a disrupted readout, or may even inhibit measurement. However, sensors that are applied close to the body core (close to vital organs like heart, brain, etc.) perform valid measurements even in critical situations, since endogenous control of the human body aims to maintain a constant perfusion of vital organs, such as heart, lungs, brain, etc. Therefore, modern rPPG sensors focus on the forehead or the ear due to their proximity to the brain. In addition to constant perfusion, the ear channel also has the advantage of constant temperature conditions.

82

B. Venema

4.2.2 In-Ear PPG from the User’s Viewpoint Thus, wearable devices for the ear have a high level of consumer acceptance. In general, wearable ear-related devices do not represent a handicap for daily activity. In addition, devices like wireless or in-ear headphones have become everyday articles, mainly due to the extensive use of mobile telephones. Therefore, a high level of acceptance is also expected for ear-related vital parameter monitoring, e.g., for sports, or the life sciences. The combination of vital parameter assessment, headset, and headphones would be beneficial, since various useful systems can be combined without restriction or additional effort of the user. For the elderly or for patients at risk, a combination with a hearing aid could also be useful.

4.2.3 The MedIT in-Ear PPG Sensor System There is increasing interest in Personal Healthcare. Applications and wearable sensors have found their place in biomedical technology research. PPG still plays an important role in various sensor technologies. International research groups are working on an adaption of PPG to the ear channel. Details on in-ear PPG sensor concepts have already been published [3–7]. Besides conceptional design, and studying the feasibility and limitations of perfusion rhythm measurement, a few research groups have also demonstrated the feasibility of pulse oximetric measurement of SaO2 in the ear channel [3–6]. These results, based on experimental human trials including only a few volunteers or patients, show a correlation of the R-values with SaO2 during fluctuations at physiological blood oxygen saturation. However, no human studies have been performed to demonstrate the feasibility of SpO2 measurement over a wide SpO2 range. Figure 4.1 presents the MedIT In-ear rPPG system. The rPPG sensor is located in the outer ear channel and connected to a measurement device (Fig. 4.2) that

Fig. 4.1 The MedIT in-ear rPPG concept for assessment of outdoor long-term dermal perfusion

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

83

Fig. 4.2 Monitoring device for standard application behind the auricle: with a flexible band for sensor connection (at the top) and a micro USB charging port (below)

can be worn behind the auricle. The measurement device drives the bidirectional LEDs and converts the photocurrent into digital values. The digitized rPPG signals are preprocessed by a microcontroller and sent via Bluetooth to a computational interface device (i.e., PC, tablet, smartphone, etc.). The communication is based on the energy-efficient Bluetooth 4.0 standard, making the system compatible with widely-used communication devices. Figure 4.3 shows the reflective PPG sensor. The optoelectronic components are realized as a silicon chip with a silicon pin diode and two LEDs that are connected in anti-parallel. The spectral maximums of the two LEDs are 760 nm and 905 nm, respectively. These wavelengths provide an optimal balance regarding biological and technological characteristics, i.e. high sensitivity of the pin diode, high optical absorption coefficients of the blood cells, and a considerable difference between oxygen-saturated and reduced hemoglobin. A LED current of 30 mA leads to 2.74 mW (4.38 dBm) light intensity at red light and 3.69 mW (5.68 dBm) light intensity at infrared light. The sensor is applied to the outer ear channel at the tragus. For fixation, the sensor chip is sealed into an earmold that takes into account the individual shape of the auricle. Although this individualized sensor concept may seem to be too expensive for commercial purposes, it offers several advantages. When worn, the sensor has an

Fig. 4.3 Working principle and structure of the in-ear PPG sensor (left); image of the sensor chip (middle) and universal in-ear otoplastic sensor version (right)

84

B. Venema

Fig. 4.4 Operational diagram of the measurement system

exceptionally high level of comfort and a high level of reproducibility with regard to the sensor position and contact pressure. This concept is particularly suitable for individuals who use such a device on a frequent or daily basis. Moreover, the system might be useful for long-term monitoring of patients either within or outside clinical facilities, or for general health monitoring of occupational groups at increased risk, e.g. the military, police, security personnel, or firefighters. Figure 4.4 shows the main components of the interface electronics. The main internal communication is based on SPI and UART with an 8 MHz crystal oscillator and a MSP430F1612 main processor (Texas Instruments Inc., TX, USA). The bidirectional LED current is generated by means of a H-bridge current driver that contains two voltage-controlled current sources. The intensity and direction of the current can be controlled by the microprocessor. The photocurrent is converted to a proportional voltage by a trans-impedance converter and digitized with a 24-bit analog to digital converter without previous analog signal processing. In addition, a three-axis, 16-bit accelerometer is implemented for motion artifact detection. The digital information is sent by the Bluetooth 4.0 module Bluegiga BLE112.

4.3 Clinical Evaluation 4.3.1 Clinical Trial Setting Nowadays, pulse oximeters are tested in human hypoxia studies and calibrated according to the regression results [8]. This is in accordance with international standard specifications that date back to the document ‘Pulse Oximeter Premarket Notification Submissions’ of the US Food and Drugs Administration (FDA). During a

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

85

hypoxia study, the correlation between SpO2 and SaO2 is identified by applying the oximeter under test to healthy persons while reducing their blood oxygen saturation over a specific period of time. This controlled hypoxemia is achieved by decreasing the oxygen partial pressure in the breathing air and increasing the nitrogen partial pressure. Arterial blood gas analysis provides comparative values. A representative result requires a sufficient number of volunteers in the study. The regression analysis of SpO2 and SaO2 is done retrospectively. Based on these results, a general calibration function can usually be found that is often approximated by SpO2 = A R + B R 2 + C

(4.1)

In the development process, this calibration specification is permanently embedded in the pulse oximeter system [8]. Noise, external influences, or individual physiological differences can lead to limitations in oxygen measurement. European guidelines require a difference in effective value of ≤4.0% in the range of 70%–100% SaO2 . The difference in effective value is defined by  Armse =

n i=1

(SpO2i − SaO2i )2 n

(4.2)

where SpO2i are the measured values and SaO2i are the comparative values. The study was performed at the University Medical Center of Schleswig Holstein (Lübeck, Germany) [11]. Ten healthy volunteers (non-smokers, three women and seven men, age range 22 years–37 years) were included in the study in accordance with the recommendation of the FDA. The study was approved by the local Ethical Committee and written informed consent was obtained. The subjects were motionless in a beachchair position. Sensors were applied to the ear and the sensor cable was temporarily fixed (Fig. 4.5). SaO2 was influenced by changing the composition of the breathing air (nitrogen/oxygen relation). SaO2 was reduced stepwise from 70%–100% (Fig. 4.6).

4.3.2 Heart Rate The heartbeat is defined as the contraction of the heart muscle in combination with the resulting cardiac output. By means of the electrocardiogram (ECG), the R-peak of the QRS complex usually defines the time point of the heartbeat. The QRS complex is associated with the depolarization of the heart muscles. Therefore, it can also be seen as the initial time point of mechanical heart contraction [9]. After the heart contracts, a pulse wave propagates through the arteries to the periphery. Hence, a time delay exists between the heartbeat and the subdermal blood volume pulse at the periphery. Nevertheless, the pulse transit time is usually constant and therefore does not influence the heart rate measurement accuracy with PPG. In rPPG, heartbeat can

86

B. Venema

Fig. 4.5 Study setting reproduced from [10]

Fig. 4.6 Stepwise experimental reduction of arterial blood oxygenation (SaO2 ). At each step, five blood samples were taken and analyzed as a ‘gold standard’ with invasive blood gas analysis (BGA)

be detected by band-pass filtering and peak detection. An ECG was used as gold standard. Heartrate was correlated based on RR-intervals between the heartbeats. In this study, besides hypoxia, the subjects were not confronted with any form of physical stress. Therefore, the heart frequency was within a physiological range. The Bland-Altman plot (Fig. 4.7 right panel) illustrates the measurement results for all subjects in beats per minute (bpm). The standard deviation of all 16,813 heartbeats was 1.2 bpm and the regression coefficient was 0.9975 in the range from 48.3 bpm to116.4 bpm.

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

87

Fig. 4.7 Final heart rate regression analysis

4.3.3 Arterial Oxygen Saturation An R-value was calculated for every heartbeat and converted into a continuous signal by spline interpolation with the original sampling rate. In theory, SpO2 can be calculated by a single heartbeat. However, in practice, measurement fluctuations demand a 20-second moving average filter. By means of two regression results, Fig. 4.8 shows that SpO2 –SaO2 regression curves can be parallel shifted and are based on different calibration coefficients [10]. While the reason for this effect is not entirely clear, metrological aspects, like analog

Fig. 4.8 Regression curves of two measurements. Parallel shift inhibits global calibration

88

B. Venema

signal processing of different devices or bilateral differences between the right and left ear, could be excluded. In any case, the distance between the regression lines is almost constant over the entire SaO2 range. The mean steepness dSpO2 /dR of all regression lines between 70100% is almost constant (mean −99.78, standard deviation 12). Therefore, an absolute SpO2 measurement can be performed with a single-point calibration routine. For this, the current oxygen saturation must be identified at the beginning of measurement (i.e. with a conventional finger pulse oximeter). Roffset is defined as the distance of the R-value to 0.7 at normal saturation (98–100%) for each participant k: Roffset (k) = 0.7 − Rnorm (k)

(4.3)

By means of Roffset the systematic measurement difference can be compensated for the entire measurement range. Rcomp (i, k) = R(i, k) + Roffset (k)

(4.4)

The resulting parameter identification provides the global calibration instruction SpO2 = 176.64 · R 2 −122.64 · R + 14.78.

(4.5)

Figure 4.9 shows the final regression analysis between the SaO2 and SpO2 with a single-point calibration routine. The resulting regression coefficient is 0.96 and the mean measurement error according to formula [2] is 1.71%. Subdivided into different SaO2 sections the measurement error is • • • •

1.76% for normal oxygen saturation for 95%–100% SaO2 , 1.14% for 90%–95% SaO2 , 1.54% for 80%–90% SaO2 2.19% for 69.5%–80% SaO2 .

Fig. 4.9 Final SpO2 regression analysis

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

89

This measurement error is within the required measurement band.

4.3.4 Breathing Activity The combination of heart rate, its variability, and the breathing rate gives a comprehensive insight into the circulatory system. Usually, all physiological phenomena that lead to fluctuations in the blood pressure influence the PPG signal. Therefore, it is possible that respiration rate can also be obtained by PPG. A simple approach for the extraction of breathing cycle information is to analyze the signal amplitude fluctuations (SAV) in the respiratory frequency band between 0.1 Hz and 0.33 Hz (3s –10 s for the duration of the breathing cycles). However, the signal components that are associated with respiration are comparatively small. Hence, it might be difficult to distinguish between the respiratory-related frequencies and other slow oscillation phenomena, such as Traube–Hering–Mayer waves or thermal regulation [13]. In addition, the cardio-respiratory coupling (CC), also known as respiratory sinus arrhythmia, can be used to obtain breathing-related information from PPG signals by analyzing rhythmical fluctuations in heart rate. We applied a Naive Bayes’ classifier to combine both approaches for the estimation of the moment of inspiration and expiration. A time delay between SAV and CC was automatically compensated. The signals were divided into sections of 0.25 s each. Maximum, standard deviation, mean value, and slope were calculated for every section as classifier features. A total of 1,827 breathing cycles derived from diverse volunteers were used as training data. Table 4.1 presents the results of the binary classification for eight subjects. The thorax belt of the polysomnograph system was used as a reference. Good results were achieved for normal (resting) breathing rates with sensitivity was 0.81 and a specificity of 0.86. Table 4.1 Results of the binary classification of inspiration and expiration moment Subject no.

BF [min−1 ]

Cycles

Sensitivity (%)

Specificity (%)

1

13.5 ± 2.6

358

77.6

88.2

2

15.2 ± 4.3

464

77.1

89.9

3

14.8 ± 3.1

430

80.6

78.1

4

12.2 ± 5.9

350

81.8

89.3

5

15.5 ± 5.8

445

81.1

81.3

6

14.5 ± 3.4

465

86.3

89.7

7

15.8 ± 4.4

460

82.3

85.3

8

9.5 ± 5.8

243

84.4

86.2

81.4

86.0

Average

13.9

BF = breathing frequency; cycles = total number of breathing cycles

90

B. Venema

4.4 Perspectives This work presents results from human studies with a remission in-ear PPG sensor with the aim of continuous cardiovascular monitoring. Due to proximity to the brain, the system is expected to be independent of centralization and provide valid and reliable monitoring of vital parameters in clinical life-threatening situations. The hypoxia studies performed with 10 healthy volunteers show the feasibility of simultaneous measurement of heart rate and SpO2 with the ear sensor. According to heart rate, a very good measurement accuracy was achieved with a standard deviation of 1.2 bpm. In combination with a single-point calibration routine, the SpO2 measurement of our in-ear pulse oximeter meets the requirements specified for pulse oximetry [11]. Therefore, in-ear PPG seems to have good potential for SpO2 measurement. Even at this early stage of development, SpO2 can be derived with acceptable accuracy and with minimal variation. Despite the need for a single-point calibration, the physiological advantages of the new measuring position (the inner ear as ‘keyhole’ to the heart) seem to surpass the disadvantages [10].

References 1. B. Venema, J. Schiefer, V. Blazek, N. Blanik, S. Leonhardt, Evaluating innovative in-ear pulse oximetry for unobtrusive cardiovascular and pulmonary monitoring during sleep. IEEE J. Trans. Eng. Health Med. 1(8) (2013). https://doi.org/10.1109/JTEHM.2013.2277870 2. B. Venema, M.S. Wolke, V. Blazek, S. Leonhardt, A power consumption optimized reflective in-ear pulse oximeter for mobile health monitoring, in Proc. Biomed. Wirel. Technol. Networks Sens. Syst. IEEE Topical Conference IEEE, pp. 34–36 (2014) 3. J. Kreuzer, Alltagstauglich sensorik: kontinuierliches monitoring von körperkerntemperatur und sauerstoffsättigung. Ph.D Thesis (Technische Universität München, Germany, 2009) 4. K. Budidha, P. Kyriacou, The human ear canal: investigation of its suitability for monitoring photoplethysmographs and arterial oxygen saturation. Physiol. Meas. 35(2), 111 (2014) 5. K. Budidha, P. Kyriacou, Development of an optical probe to investigate the suitability of measuring photoplethysmographs and blood oxygen saturation from the human auditory canal, in Proc. Eng. Med. Biol. Soc. (EMBC), 35th Annual International Conference of the IEEE 2013, pp. 1736–1739 (2013) 6. M. El-Khoury, J. Sola, V. Neuman, J. Krauss, Portable SpO2 monitor: a fast response approach, in Proc. IEEE Int. Conf. Portable Infor. Devices, pp. 1–5 (2007) 7. J. Patterson, G. Douglas, G. Yang, A flexible, low noise reflective PPG sensor platform for ear-worn heart rate monitoring, in Proc. Wearable and Implantable Body Sensor Networks BSN 2009, Berkeley, CA, pp. 286–291 (2009) 8. J.G. Webster, Design of pulse oximeters. CRC Press (2002) 9. R. Klinke, H.C. Pape, A. Kurtz, S. Silbernagl, Physiologie. Georg Thieme Verlag (2009) 10. B. Venema, J. Schiefer, V. Blazek, N. Blanik, S. Leonhardt, Advances in reflective oxygen saturation monitoring with a novel in-ear sensor system: results of a human hypoxia study. IEEE Trans. Biomed. Eng. 59(7), 2003–2010 (2012) 11. B. Venema, H. Gehring, I. Michelsen, N. Blanik, V. Blazek, S. Leonhardt, Robustness, specificity, and reliability of an in-ear pulse oximetric sensor in surgical patients. IEEE Biomed. Health Inform. J. 18(4), 1178–1185 (2014)

4 Reflective Arterial Pulse Oximetry for New Measuring Sites …

91

12. B. Venema, N. Blanik, V. Blazek, J. Schiefer, S. Leonhardt, A feasibility study evaluating innovative in-ear pulse oximetry for unobtrusive cardiovascular homecare monitoring during sleep, in Proc. IEEE Point-of-Care Healthcare Technologies (PHT), pp. 124–127 (2013) 13. B. Venema, V. Blazek, S. Leonhardt, In-ear photoplethysmography for mobile cardiorespiratory monitoring and alarming, in Proc. IEEE Wearable and Implantable Body Sensor Networks (BSN), 5 (2015). https://doi.org/10.1109/bsn.2015.7299367

Chapter 5

Peripheral Venous Dynamics, Venous Oxygen Saturation and Local Oxygen Consumption Measured with an Extended Photoplethysmograpic Muscle Pump Test Vladimir Blazek and Claudia Blazek Abstract This chapter describes a novel photoplethysmographic test strategy to measure venous and arterial oxygen saturation noninvasively, employing an extended venous muscle pump test. During this test, the patient performs dorsal extensions of the ankle in a sitting position and pumps venous blood from the periphery back to the heart by contraction of different leg muscles. Subsequently, the ‘venous blood volume pulse’ is measured to determine the lower leg’s functional venous oxygen saturation level at the test site. We present this method as a unique and, for the first time, complementary to the arterial saturation level estimated from the arterial blood volume pulse. An advantage of our approach is that it is possible to calculate local oxygen consumption too and, hence, to estimate the new parameter related to metabolic activity of the tissue surrounding the test site. Therefore, this method is considered as a pioneering one. Based on our preliminary results, peripheral venous saturation under physiological conditions is significantly lower (around 75%) in comparison with arterial oxygen saturation about 98%.

5.1 Need for Venous Oxygen Saturation Measurement We have sufficient energy reserves in our bodies to survive several days without food intake. Unfortunately, we do not have an extensive oxygen in our bodies; hence we are in imminent danger once we stop breathing. Therefore, it is important that the oxygen content in blood is measured, especially for patients suffering from cardio pulmonary ailments. The pulse oximeters described in chaps. 3 and 4 accomplish this V. Blazek (B) Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany e-mail: [email protected] C. Blazek The Private Clinic of Dermatology, Haut im Zentrum, Zurich, Switzerland © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_5

93

94

V. Blazek and C. Blazek

task. Apart from the arterial oxygen saturation, a knowledge of oxygen saturation in the veins will aid in analyzing the physiological functioning of the human cardio pulmonary system in the maintenance of cellular oxygen usage [1]. Such a task would be very beneficial, especially to patients suffering from cardio-pulmonary disorders, if carried out noninvasively.

5.2 Some Physiological and Experimental Remarks One of the substances transported with our blood is oxygen. On its way from the right to the left heart, namely the pulmonary circulation, the deoxygenated blood flows through oxygen-filled alveoli in the lung. The active red blood cells (with hemoglobin as a very important component) absorb the oxygen and transport it to the farthest tissue structure in the body. The oxygen is mainly carried by hemoglobin to which it is chemically bound and exploited by different cells of the body. This process leads to the metabolic process. Since the blood returning to the heart through veins it is depleted of oxygen, the venous oxygen saturation is always less than the oxygen saturation in the arterial branch, as portrayed in Fig. 5.1. Therefore, the diagnosis and evidence-based treatment of peripheral tissue hypoxia plays a crucial role in many medical disciplines. The well-known and clinically accepted non-invasive method of measuring arterial oxygen saturation through pulse oximetry utilizing photoplethysmographic (PPG) signals was first presented by Takuo Aoyagi in 1972. Takuo Aoyagi found

Fig. 5.1 Schematic representation of the oxygen transport in our body, modified after [2]. Under physiological conditions, the arterial oxygen saturation is about 98%. The amount of oxygen delivered from the arterial inflow to the body tissue is normally about 25% so that the oxygen saturation in the venous circulation section will be about 73%. The schematically shown oxygen transport system thus ensures adequate oxygen supply to the tissue. Therefore, through oxygen supply and oxygen consumption in the body, a balance is maintained. This is a very important factor for the occurrence or non-occurrence of organ disorders, not only in ICU patients Note When arterial oxyhemoglobin saturation is measured by an arterial blood gas analyzer, it is referred to as SaO2 . When arterial oxy-hemoglobin saturation is measured non-invasively by pulse oximetry, it is referred to as SpO2 ; venous oxygen saturation is indicated as SvO2

5 Peripheral Venous Dynamics, Venous Oxygen Saturation …

95

that arterial oxygen saturation could be measured noninvasively by quantifying the frequency spectra of the light emissions coming through the tissue [3], which made heating the tissue (used in former measuring setups) [4] obsolete. This device was the precursor for all the modern pulse oximeters [5–8]. In 2015, Aoyagi was recognized with the IEEE Medal for innovations in healthcare technology for his “… pioneering contributions to pulse oximetry that have had a profound impact on healthcare” [9]. In contrast to the advances in pulse oximetry, the monitoring of the venous oxygen saturation (SvO2 ) measurement continues to be conducted invasively. Of the available methods, the invasive in vitro measurement of the extracted venous blood is the gold standard in SvO2 measurement. The in vivo monitoring of blood gas through an intravascular catheter (e.g. in a. pulmonaris) is another technique that is clinically accepted [10]. A photoplethysmographic based solution is described in the literature, which uses a small finger cuff [11]. Using the cuff, external pressure is applied to modulate the venous flow in the vascular segment downstream of the cuff. From the pulsatile venous blood flow produced by this occlusion maneuver, SvO2 can be calculated. One of the major impediments in this method is that using a cuff, only very small blood volume changes in the venous blood flow are possible and therefore, the ability to detect venous oxygen saturation by this method is very limited.

5.3 Photoplethysmographic Measurement of the Peripheral Venous Oxygen Saturation This chapter describes a novel non-invasive method for the determination of the peripheral, dermal venous oxygen saturation (SvO2 ) in human extremities. In its minimum configuration, this PPG based venous oxygen saturation measurement method consists of • a flat, flexible optoelectronic sensor that can operate in reflection or transmission mode, depending on the selected tissue area, with at least one photodetector and two light sources (preferably one in the red wavelength region and the other in IR wavelength region) whose light illuminates the assessed skin area, • a sensor control and evaluation unit, in which the output of the sensor signals is digitized so that the PPGs at the two different (red and IR) wavelengths are separated and made available for further processing, • a digital filter unit which detects the arterial DC and AC signal components of the PPG signals at the two (red and IR) wavelengths, • a second digital filter unit that extracts the venous DC and AC signal components from the PPG signals and • a processor that evaluates the venous and arterial oxygen saturation using the AC and DC signal components filtered by the two digital filters portrayed above. The proposed venous oxygen saturation (SvO2 ) and arterial oxygen saturation (SpO2 ) measurement method employ the well known venous muscle pump test

96

V. Blazek and C. Blazek

(VMPT). A VMPT is performed to measure the “venous refilling time“ (VRT) and “muscle pump capacity” (MPC). It is shown here that using the standardized leg exercise (muscle pump) performed during a VMPT, apart from determining the VRT and MPC, arterial oxygen saturation (SvO2 ) and venous saturation levels (SpO2 and SvO2 ) at the local test site also can easily be determined. An advantage of the method is that it is possible to calculate the oxygen intake from the SpO2 and SvO2 values and hence a measure on the metabolism activity of the tissues surrounding the test site. It should be noted here that such a measure on combined arterial and venous saturation and thus the determination of oxygen consumption (that provides insight into metabolism) is very difficult and thus, the present method is a pioneering one. On the other hand, in a venous muscle pump test, large variations in venous blood flow, especially in the peripheries, can be achieved. Thus, if such large blood volume changes that occur in veins during a VMPT are used for the measurement of SvO2 , clinically acceptable levels of accuracy can be reached [12]. Such a technique utilizing the absorption spectra of hemoglobin is described next.

5.4 Optical Absorption Spectra of Hemoglobin Hemoglobin acts as a transporter for oxygen within the blood. If the hemoglobin molecule is bound to oxygen then one has oxy-hemoglobin or HbO2 . If the hemoglobin molecule is bound to carbon monoxide then one has carboxyhemoglobin or HbCO [13]. If the hemoglobin molecule is bound to nothing then one has deoxy-hemoglobin or simply Hb. If the hemoglobin molecule has broken down then one has met-hemoglobin. The absorption spectra of Hb and HbO2 are shown in Fig. 5.2.

Fig. 5.2 Absorptions spectra of hemoglobin in visible and IR-A range of the spectrum

5 Peripheral Venous Dynamics, Venous Oxygen Saturation …

97

5.5 Standardized Venous Muscle Pump Test for Evaluating the Efficiency of the Calf Pump The main application of quantitative photoplethysmography is in the functional assessment of the blood transport properties of the vein system in the leg [14–17]. During sitting, standing or walking, the blood has to flow ‘uphill’ in the leg veins; in healthy individuals, this function is performed mostly by the calf muscle pump. Each muscle contraction compresses the leg veins, causing their contents to escape, proximal (to the heart). In persons with healthy veins, the venous valves prevent the blood that has been pumped upward from falling back again. The muscle pump test is performed with the patient in a sitting position; the patient executes eight dorsal extensions of his or her foot within 16 s. Figure 5.3 illustrates the test in a schematic form. To obtain a signal indicative of blood volume changes due to the calf muscle pumping action, a PPG sensor is placed about 10 cm above the inner ankle. Directly after exercise, the venous pressure in the legs drops, causing a decrease in bloodvessel filling. As a result, light reflection in the measurement area increases; this is accompanied by a rise in the measured PPG curve. During the resting phase at the end of the exercise, the leg veins are normally refilled by the arterial inflow and the PPG curve relatively slowly returns to near its baseline, as indicated in Fig. 5.3. In patients with incompetent venous valves, the pathologic reflux in the leg veins results in a smaller decrease in venous blood volume in the movement phase and a faster refilling in the subsequent resting phase. This corresponds to a decrease in the curve amplitude and a shortening of refilling time as shown in Fig. 5.4. The following parameters for assessing venous function can be taken from the PPG curve: • venous refilling time (To) in seconds and • venous pump power (Vo) expressed in percent of the optical signal (PPG% = percent change in skin reflection based on the calibrated baseline prior to the start of the exercise).

Fig. 5.3 Standardized venous muscle pump test. Left: Position of the PPG sensor on the leg and PPG device Vasoquant, ELCAT, Germany. Right: Scheme for the course of the test, nomenclature and the most important assessment parameters of the peripheral venous hemodynamics [17]

98

V. Blazek and C. Blazek

Fig. 5.4 Standardized PPG muscle pump test. Comparison of typical curves and functional degree graduation of the peripheral venous blood volume behaviour under active leg movement. Usually the venous refilling time and the venous pump power decreases with the increase in the hemodynamical severity of disorder [17]

Following previous procedure, the venous refilling time is considered to be the currently most important parameter. Depending on its shortening, for example as a result of pathological reflux in malfunctioning veins, the following severity classification was introduced for the active leg exercise procedure in sitting position: • • • •

Normal hemodynamics (healthy veins), Grade I (mild hemodynamic disorder), Grade II (moderately severe hemodynamic disorder), Grade III (severe hemodynamic disorder).

Statistical analysis of measurements made on various test subjects and patient groups indicate that the venous refilling time is able to early on distinguish between persons with healthy veins and those with a venous disease, whereas venous pump power mainly indicates if a varicosis requires treatment [18]. The oldest functional test for the disturbance of venous hemodynamics is the invasive measurement of venous pressure during exercise, which is—somewhat simplified—also referred to as phlebodynamometry. For many years this procedure has served as the ‘gold standard’ and all new methods had to be evaluated according to this standard. Comparison measurements made with dynamic photoplethysmography and phlebodynamometry display a good statistical correlation (r = 0.898 [19] or r = 0.96 [20]). The last-mentioned method, however, is painful for the patient and slightly difficult to carry out. Therefore, phlebodynamometry has not gained widespread acceptance as a routine diagnostic tool—all the more reason why PPG is an important alternative. Today, PPG is used—both in hospitals and private practices—on a large scale to determine the severity of disturbances of venous function. An equally important application is the measurement of functional improvements following varicose vein surgery or sclerotherapy and—in the event of a recurrence—especially for the assessment of the functional improvements, if any, during the follow-up period [16, 17, 21, 22].

5 Peripheral Venous Dynamics, Venous Oxygen Saturation …

99

5.6 Extended Multi-Wavelength Venous Muscle Pump Test for the Assessment of the Venous Blood Oxygen Saturation Level As already mentioned, the term “oxygen saturation“ means the ratio of oxygen present in the blood to the maximum oxygen carrying capacity of the blood. In the normal case, the arterial saturation is around 98% and the peripheral venous saturation is around 73% to 75%. Local disturbances in the human oxygen consumption chain are naturally extremely important because, without oxygen, the cells can not assimilate nutrients and hence, cannot survive. Thus, without our cells burning oxygen, there is no energy infused into the cells and thus no healthy metabolism. Usually, by puncturing a central artery and vein and performing invasive blood gas analysis on the samples collected from the vein and the artery, both arterial saturation and global (central) venous saturation can be measured. The non-invasive arterial pulse oximetry evaluates the local arterial oxygen saturation utilizing the heart synchronous pulsation in the arterial blood by analyzing dual wavelength PPG signal from that local site. However, a pulse oximeter does not provide information about the local hypoxia in terms of oxygen consumption, because the method of oxygen saturation estimation used in a pulse oximeter can not be used to detect the venous saturation and thus the local oxygen consumption (the difference between the arterial and venous oxygen saturations) can not be ascertained with a pulse oximeter. The difference between the arterial and venous oxygen saturations determines the metabolism “on-site” and thus diagnostically very important especially in patients with peripheral vascular disease. Such conditions are quite normal in patients with diabetes, patients suffering from peripheral arterial disease or patients with difficult to heal wounds (ulcers). In this case, the arterial oxygen saturation is determined from the arterial pulsation, and in addition, by applying an embossed leg exercise, the local venous oxygen saturation can be assessed (from the resulting venous pulsation). This is the first time that the local oxygen difference between arterial and venous saturation (oxygen consumption) is measurable. We applied the thin and flexible PPG sensor to the lower extremity to a sedentary subject/patient, (e.g. ankle area, toes, or on the longitudinal arches of the foot; Fig. 5.5 left). After an auto-calibration of the measuring system, the patient performs an extended leg exercise (15 dorsal extensions in 2 s intervals). It this time sequence, there are two recorded perfusion signals available, one at λ1 (preferably 940 nm) and another one at λ2 (preferably 660 nm); the combined arterial and venous blood volume variations in the measurement area at the extremity are recorded (Fig. 5.5 right). These signals are forwarded to two digital filter groups with different spectral signal characteristics. One of the filters or filter groups calculates DC and AC components of the arterial signal component, the other filter then calculates DC and AC signal components of the venous signal component (Fig. 5.6). The resulting signal

100

V. Blazek and C. Blazek

Fig. 5.5 Multi wavelength muscle pump test. Through innovative, flexible and flat PPG sensors (left) can—in addition to the venous refilling time (top right) and the peripheral arterial blood volume pulse - local peripheral (arterial and venous) oxygen saturation for example under the compression stocking (bottom right) be determined. The amplitude of the arterial blood volume pulse is approximately 100 times lower than the venous blood volume pulsation during the active leg exercises

components at the output of both filter groups are added to further signal processing to calculate the saturation values.

5.7 Preliminary Results and Conclusion To verify the presented measurement concept in a first study, a commercial pulse oximeter (tPPG mode) is used and attached to the second toe finger (volunteer in sitting position). The obtained SpO2 was 98% while the heart rate was at 84 bpm (Fig. 5.7 left). Then, locking the arterial inflow into the foot utilizing a wide tourni-

5 Peripheral Venous Dynamics, Venous Oxygen Saturation …

101

Fig. 5.6 Typical perfusion signals (with arterial and venous signal components), measured at two different wavelengths. Representation of the “arterial pulse (AP)” and “venous pulse (VP)” DC and AC signal components for muscle pump test signals from Fig. 5.5. The venous DC to AC difference namely during the active leg exercise; consequently, only in this leg exercise time sequence, the value of the venous oxygen saturation (SvO2 ) can be calculated. Rv (from venous pulse) and Ra (from arterial pulse) are the normalized red to infrared ratios for this purpose

Fig. 5.7 First experimental study with a conventional low-cost fingertip pulse oximeter. In the first measuring phase, the arterial oxygen saturation was determined by 98% (left), in the second measurement phase (after interrupting arterial perfusion of the leg) the peripheral venous oxygen saturation was determined as 82%

quet, the muscle pump test was performed as described above. Since the arterial pulsation was not available, even the commercial pulse oximeter has correctly identified the venous blood volume pulse generated from the exercise. The venous saturation of 82% and the venous pulse rate (resulting from by the 15 active leg movements one in every two seconds) of 29 bpm was detected (Fig. 5.7 right). Our method of providing

102

V. Blazek and C. Blazek

supplementary oxygen in venous blood can also quantify the oxygen uptake (peripheral arterio-venous oxygen consumption/oxygen metabolism) by tissue surrounding the site wherein the sensor is placed [23]. By consecutive measurements on one 67-year-old subject (two measurements per day in ten days in total, n = 20), the local mean arteriovenous oxygen consumption of 16.8 ± 1.83% was determined. This value is smaller than expected. Venous oxygen saturation measured invasively ranges from 53% to 80% [24, 25], whereas noninvasive measurement of venous oxygenation based of the pulsatility of the complex PPG signal clusters around 80% [26]. Although the final experimental system realization and clinical testing of the presented new PPG measurement approach is still to be conducted, our preliminary results indicate that the expected results provide clinically relevant differences in saturation between the local arterial and venous blood supply.

References 1. 2. 3. 4. 5. 6. 7.

8.

9. 10. 11. 12.

13. 14.

15.

16.

K. Golenhofen, Physiologie (Urban & Schwarzenberg, München, 1997). ISBN 3-541-16351-8 http://www.rudolf-deiml.homepage.t-online.de/Kapitel01.html T. Aoyagi, Discovery of pulse oximetry. Anesth. Analg. 105(6 Suppl), 1–4 (2007) G.A. Millikan, The oximeter: an instrument for measuring continuously oxygen-saturation of arterial blood in man. Rev. Sci. Instrum. 13(10), 434–444 (1942) J.A. Pologe, Pulse oximetry: technical aspects of machine design. Internat. Anesth. Clin. 25(3), 137–153 (1987) J.G. Webster, Design of Pulse Oximeters. Taylor & Francis, (1997), ISBN 0750304677 B. Venema, H.M. Gehring, I. Michelsen, N. Blanik, V. Blazek, S. Leonhardt, Robustness, specifity and reliability of a novel in-ear pulse oximetric sensor in surgical patients. IEEE J. Biomed. Health Inform. 6, https://doi.org/10.1109/jbhi.2013.2292118, (2013) J.J. Pretto, T. Roebuck, L. Beckert, G. Hamilton, Clinical use of pulse oximetry: official guidelines from the thoracic society of Australia and New Zealand. Respirology 19(1), 38–46 (2014) http://ethw.org/IEEE_Medal_for_Innovations_in_Healthcare_Technology P.J. Fahey, K. Harris, C. Vanderwarf, Clinical experience with continuous monitoring of mixed venous oxygen saturation in respiratory failure. Chest 86(5), 748–752 (1984) US 7,263,359 B2 (2008) United State patent: venous pulse oximetry DE 10 2011 122 700.1, German patent application: Bewegungskorreliertes Verfahren und opto-elektronische Vorrichtung zur nichtinvasiven Bestimmung der dermalvenösen Sauerstoffversorgung peripherer Beingebiete (2011) B.L. Horecker, The absorption spectra of hemoglobin and ist derivates in the visible and near infra-red regions. J. Biol. Chem. 148, 173–183 (1943) J. Kerner, U. Schultz-Ehrenburg, V. Blazek, Digitale photoplethysmographie (D-PPG): klinische eignung der neuen messmethode zur venösen funktionsdiagnostik. Phlebol Proktol 18, 98–103 (1989) J. Kerner, U. Schultz-Ehrenburg, W. Lechner, Quantitative photoplethysmographie bei gesunden erwachsenen, kindern und schwangeren und bei varizenpatienten. Phlebologie 21, 134–139 (1992) V. Blazek, U. Schultz-Ehrenburg, Quantitative Photoplethysmography. Basic facts and examination tests for evaluating peripheral vascular functions, (VDI Verlag Düsseldorf, 1996), ISBN 3-18-319220-9

5 Peripheral Venous Dynamics, Venous Oxygen Saturation …

103

17. V. Blazek, C. Blazek, Funktionelle Beinvenendiagnostik mit der quantitativen Photoplethysmographie (PPG)—bewährte und neue Untersuchungstests der peripheren venösen und arteriellen Diagnostik, ed. by K. Hübner, F.X. Breu, Praktische Sklerotherapie, (Viavital-Verlag Köln, 2013), ISBN 978-3-934371-49-1, pp. 64–76 18. U. Schultz-Ehrenburg, V. Blazek, New possibilities for photoplethysmography. Phlebology Digest 4, 5–11 (1993) 19. H.B. Abramowitz, L.A. Queral, W.R. Flinn, P.F. Nora Jr., L.K. Peterson, J.J. Bergan, J.S. Yao, The use of photoplethysmography in the assessment of venous insufficiency: a comparison to venous pressure measurement. Surgery 86, 434–441 (1979) 20. A.N. Nicolaides, C. Miles, Photoplethysmography in the assessment of venous insufficiency. J. Vasc. Surg. 5, 405–412 (1987) 21. A.N. Nicolaides, V. Blazek et al., Investigation of chronic venous insufficiency. A concensus statement. Circulation 102, e126–e163 (2000) 22. U. Schultz-Ehrenburg, V. Blazek, Value of Quantitative Photoplethysmography for Functional Vascular Diagnostics. Skin Pharmacol. Appl. Skin. Physiol 1416, 316–323 (2001) 23. V. Blazek, N. Blanik, C. R. Blazek, M. Paul, C. Pereira, M. Koeny, B. Venema, S. Leonhardt, Active and passive optical imaging modality for unobtrusive cardiorespiratory monitoring and facial expression assessment. Anesth. Analg. 124(1), 104–119 (2017) 24. A. Keys, The oxygen saturation of the venous blood in normal human subjects. Amer. J. Physiol. 124, 13–21 (1938) 25. K. Reinhart, F. Bloos, The value of venous oximetry. Curr. Opin. Crit. Care. 11, 259–263 (2005) 26. Z.D. Walton, P.A. Kyriacou, D.G. Silverman, K.H. Shelley, Measuring venous oxygenation using the photoplethysmograph waveform. J. Clin. Monit. Comp. 24, 295–303 (2010)

Chapter 6

Low-Frequency Blood Volume Rhythms in the Skin Perfusion Obtained by Optical Sensing Hans J. Schmitt

Abstract Optical sensing of physiological rhythms has been the subject of intensive research in past years. Photoplethysmography (PPG) is well established for optical reflection and/or transmission measurements of dermal blood volume changes phenomena. PPG recordings from the skin exhibit a rich spectrum including components around 1 Hz (and harmonics) due to heart pulse, breathing periodicity normally around 0.25 to 0.3 Hz, but varying over a wide range down to 0.1 Hz and less, depending on workload and wilful control, and a periodic low frequency component at around 0.12 to 0.14 Hz, which usually occurs for a number of minutes, then may disappear and possibly reoccur from time to time. The origin of this frequency component is still being debated. Its importance rests in the belief, that it is somehow related to a certain state of mind and awareness. This in turn could have therapeutic consequences if such oscillations and the corresponding mental state could be induced by outside control.

Various theories have been brought forward as to the origin of these low-f oscillations: as a sub-harmonic of breathing via parametric-, coupling- or hysteresis effects, either mechanically by thorax motion or by unspecified nervous effects [1], as a frequency modulation due to a slight heart arrhythmia known to occur with inhaling and exhaling [2], nonlinear coupling processes of blood flow and pumping between left and right sections of the heart [3], rhythmic exchange of brain fluid, such as palpable by osteopathy [4] or natural mechanical oscillations in the vascular system, various metabolic processes, or the action of a so far unknown ”internal clock”. This chapter was previously published in [5].

H. J. Schmitt—is Deceased. H. J. Schmitt (B) Institute for High Frequency Technology, RWTH Aachen University, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_6

105

106

H. J. Schmitt

6.1 Characteristics of the Low-F Response Through recent years hundreds if not thousands of recordings of blood flow variations have been made in various centres, on various positions of the body, and under different environmental and exercise conditions. Usually, perfusion is recorded in the time domain and subsequently Fourier-transformed for spectral representation. As a first feature it may be noted, that the recordings indicate primarily a superposition of the rapid (heart) pulse and the low frequency components, as indicated in Fig. 6.1. Contrarily, a modulation would give a different time response and would result in close sidebands around the pulse peak in the spectrum. Occasionally there may be additional weak AM or FM modulation effects exemplified in a slight broadening of the pulse peak. Secondly, previously speculated direct connection to breathing, i.e. by frequency doubling of the normal breathing rhythm could not be verified: by wilful change of the breathing rate from a normal 0.35 Hz to less than 0.1 Hz the frequency of the so-called ”relaxation rhythm”, if present, stayed more or less fixed around 0.13 Hz, which would definitely exclude a direct origin due to breathing, Fig. 6.2. Again, there are indications of some possible coupling phenomena if integer ratios of the individual frequencies are reached, but these would merely reflect the general inherent coupling of different biological processes. A third aspect concerns the fact, that the frequency of the slow fluctuation appears to be in a rather narrow range around 0.12 Hz–0.14 Hz for all subjects, and moreover, the same for measurements on the head, on extremities and thorax. This would indicate a central origin or at least central control somehow connected to fundamental

Fig. 6.1 a amplitude modulation, b superposition, c typical experimental PPG response, time domain and spectrum

6 Low-Frequency Blood Volume Rhythms in the Skin Perfusion …

107

Fig. 6.2 Spectra of breathing (left) and PPG on head (right): a normal breathing and b slow breathing, relaxation rhythm remains at about 0.14 Hz

human features, more or less excluding local mechanical vibrations, etc., which would tend to vary for different observations points due to differences in the local vascular geometry. The fourth observation under different external conditions shows that certain exercises like YOGA – by experienced people – have a significant influence on the rapidity of onset of the low-f rhythms and the duration of their measurable appearance, up to 15 min and more in a stretch. In general, YOGA-exercises enhance their occurrence, and in as much as these exercises are also related to a certain relaxation of the state of mind indicate an influence of the mental condition, enhanced by a sequence of stress/relaxation steps, Fig. 6.3. This is based on measurements of 13 persons (9 male, 4 female) with different YOGA-experiences, performed at the Krishnamacharya YOGA Mandiram in Madras [6]. Many other mental influences are known to affect the blood distribution, particularly on the head and ears: continuous mental stress leads to a nervously controlled redistribution with more blood in the skeleton muscles and less in the skin (”pale” face of thinkers!), while other, more sudden effects like being caught ”telling a lie” tend

108

H. J. Schmitt

Fig. 6.3 PPG left earlobe a breathing b in time low frequency spectra c and d during YOGA. Strong 0.12 Hz rhythm in PPG signal

to redden face and ears. These are all indications of a very effective and ubiquitous autoregulation system to adapt to different tasks and loads.

6.2 Concept of Potential Origin The most obvious feature concerns the rather fixed frequency of the observed ”relaxation rhythm”. Any explanation would have to call for phenomena with relevant time duration in the order of several seconds. One such phenomenon is the time delay incurred for blood flow in the capillary system, associated with metabolic processes,

6 Low-Frequency Blood Volume Rhythms in the Skin Perfusion …

109

in particular the O2 exchange from blood to the interstitium and the related CO2 balance. This would be particularly relevant for the brain, which reacts very sensitive to O2 deficiencies: for example, it is known that a complete blocking of blood flow to the brain leads to noticeable effects in 4 to 6 s and unconciousness after about 15 s. Normally, a deficiency due to whatever causes will be autoregulated by widening of the arterioles and enhanced blood flow. However, there are time constants involved in the control processes which may easily be in the order of magnitude of seconds. A naive estimate for capillary action in the brain based on literature values [7] for blood flow to the brain of 715 ml/min, ~ 11.9 ml/sec, and a blood volume of 300 ml in the head, with ~ 15% in the capillary system, i.e. ~ 45 ml would result in a crossing time of ~ 3.8 s. A similar value is obtained by estimate of the capillary length: with an overall cross-section of the capillary system of 3500 cm2 , a diameter of 0.0009 cm [7] together with a value of 5% of the total blood volume, would result in an average capillary length of around 0.086 cm and a flow velocity of v ∼ 0.026 cm/s. This is to be corrected for non-constant velocity profile across the capillary ”tubes”, as the first estimate with laminar flow and parabolic distribution a factor of π /4, hence v ∼ 0.021 cm/s. and a total time for crossing the capillary system t ~ 4.1 s. This in no way can describe the detailed and complicated blood flow in the capillaries, which partially may even be shut at times, but it is believed to give a reasonably correct order of magnitude of gross effects. In this purely speculative picture, the relaxation after some stress inducing enforced physical or mental activity would lead to a lowering of the blood pressure in the brain, particularly also at the hypothalamus, which controls a. o. the circulatory parameters. This would lead to a gradually increasing O2 deficiency due to reduced blood flow. By autoregulation, this in turn would lead gradually to a widening of arterioles via sympathic nerves tending to counteract the decrease in blood flow. However, until the original blood flow rate has been reached, there will still remain an overall O2 -deficiency, so blood perfusion would continue to rise for some time until stopped by counteracting nervous control, possibly even associated with some reactive hyperaemia. Thus, a rhythmic oscillation with a period of roughly 2 × t ~ 8.2 s could be set up, Fig. 6.4. The control process involves the action of both, the (slow) sympathic and the (fast) parasympathic nerves, which ”stray” into the entire circulatory system. In this picture, the low frequency rhythm appears as a centrally excited oscillation, which by nervous coupling can be observed in all different locations on the skin and also in impedance plethysmography [8] and in correlated small variations in the ECG, etc. [9]. In a sense, the model implies a slight dynamic oscillation around a ”rest state” which may be advantageous compared to a fixed, constant value. This

110

H. J. Schmitt

Fig. 6.4 (above) Perfusion and O2 transfer in capillary system. Period is in the order of 2t

is found often in nature, e.g. the rhythmic heart pumping, periodic breathing and muscular trembling under stress. In a very rough ”binary” approximation, an electrical equivalent circuit could be drawn, which would contain a delay line (delay t) with fast feedback via an amplifier with proper amplification and phase behaviour, Fig. 6.5.

6.3 Outlook Although clearly entirely speculative, this picture, leaning on electrotechnical considerations, correctly results in a superposition of rhythms and shows a low-f variation in the proper frequency range, which is not primarily correlated to breathing, and relates to the mental and physical state via the extreme O2 sensitivity of the brain. In order to substantiate or falsify this model experiments with artificial O2 excess and vice versa with O2 deficiency, e.g. in great heights could be performed. Furthermore, to support the picture of ”central” generation and nervous transmission the measurements of the relative phase or time shift of the low-f rhythms along portions of the body should be refined. The time constants of nervous propagation are reasonably well known and generally much shorter than the multi-seconds delay in the capillary flow. A further test of central or local origin could be made on denerved sections of the skin. It may be observed, that in this interpretation the origin of the 7 s–8 s. rhythm appears to be related or at least similar to the so called „primary respiration rhythm “in osteopathy. This is believed to be due to a pulsatile exchange of liquor cerebrospinalis at a comparable rate. It shows in the „motility“, a slowmotion of the structure. Even though the precise origin is still unknown, osteopaths

6 Low-Frequency Blood Volume Rhythms in the Skin Perfusion …

111

Fig. 6.5 (right) Simplified delay line model for capillary action and nervous feedback

claim that it is palpable on the cranium. Textbooks on physiology-while basically in agreement with the principal exchange of liquid-vary as to the amount of liquid involved. Conceivably we observe the same basic phenomena, a dynamic variation of amount and constituency of liquids in the brain area. To clarify relations it would be worthwhile to extend the optical investigations to small children, where osteopaths observe a slightly higher frequency for the „primary respiration rate“, deviating from the otherwise relatively fixed 0.12 Hz–0.14 Hz for adults. Rhythmic “jets” of liquor (CSF) have recently also been seen in MRI [10].

112

H. J. Schmitt

References 1. H.J. Schmitt et al., Note on the generation of breathing related subharmonics in arterial PPG rhythms, in Proc. 8th Int. Symp. CNVD 1998, VDI, Düsseldorf, ISBN 3-18-330020-6, pp. 41–48 (1999) 2. M.M. Rao et al., Influence of controlled breathing (PRANAYAMA) on dermal perfusion as monitored by optical sensors, in Proc. 8th Int. Symp. CNVD 1998, VDI, Düsseldorf, ISBN 3-18-330020-6, pp. 49–55 (1999) 3. D. Hoyer et al., Nonlinear dynamics of the cardio respiratory system. IEEE Eng. Med. and. Biol. 12, 16–61 (1998) 4. C. Newiger, Osteopathie. TRIAS-1999, Verb. der Osteopathie Deutschland, (1999) 5. H.J. Schmitt et al., Low frequency blood volume rhythms: possible origins and new measurements, in Proc. 9th Int. Symp. CNVD 2000, VDI, Düsseldorf, ISBN 3-89653-881-0, pp. 99– 104 (2001) 6. U. Fornefeld, Kreislaufrhythmik. Master Thesis 2442, Institute of High Frequency Technology, RWTH Aachen University (1999) 7. S. Silbernagl et al., Taschenatlas der Physiologie (Thieme, Stuttgart, 1979) 8. J. Halek, L. Dolezal, Local skin perfusion detection using impedance plethysmography, Proc. ICMDTP ’99, Madras, India (1999) 9. A. Oslová, K. Javorka et al., Spectral analysis of the heart rate variability in juvenile hypertonic. Cs Pediat 54(7), 340–343 (1999) 10. M.T. Vlaardingerbroek, J.A. den Boer, Magnetic Resonance Imaging (Springer, Berlin, 1996), pp. 327–328

Chapter 7

Synergetic Interpretation of Patterned Vasomotion Activity in Microvascular Perfusion: Application to Objective Recording of Subjective Responses to Pain Holger Schmid-Schönbein Abstract Using pragmatic recording and data compression tools developed by the Institute of High Frequency Technology, RWTH Aachen University, a method was developed allowing to portray the dynamics of periodically varying cardiovascular movement patterns of the blood content (depicted by transmission and reflecting plethysmography, PPG) and blood cell displacement in microvessels (depicted by Laser-Doppler anemometry technique, LDA) in the form of so-called frequency chromatograms. These are constructed subsequent to Fourier transformation of the aperiodically fluctuating data obtained by photometric or anemometric means, where characteristic frequencies are displayed as a function of the logarithm of the frequency. In addition, the compressed time series of the obtained data (recorded from the sites of interest and from the hyperemic forehead of the subjects and patients) were being displayed. The resulting records of computer-based data compression are being called “double plots”. These are a convenient means for displaying the fundamental fact emphasised by Haken and Koepchen, stating that all physiological activities controlled by central nervous systems are dominated by so-called quasi-attractors, i.e. transient cooperativities of ensembles of neurons and effectors.

Holger Schmid-Schönbein is deceased. H. Schmid-Schönbein (B) Department of Physiology, RWTH Aachen University Hospital, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_7

113

114

H. Schmid-Schönbein

7.1 Introduction: A Short Outline of the Problem of Uncovering “Order” in Apparently “Chaotic” Time Series as They are Recorded in PPG and LDA Measurements Fundamental rule of physiological cooperativity can be corroborated by the use of the “double plot technique”. While not allowing to simply calculating the true power of each spectral contribution to the overall energy content, the combined display of frequency chromatographs and the time series made it possible to identify distinct frequency and amplitude domains reflecting not only the cardiac activity and that of the neuronal vasoconstrictor effects. Furthermore, it clearly showed that there are periodically modulated cutaneous and mucosal microvascular activities passively following the fluctuations of arterial blood pressure on the one hand and by ventilation-related events (that require more detailed study). Thus, in applying both parts of the “double plots”, it has become possible to identify patients with severe inflammation of the gingival mucosa which could not be separated from healthy controls on the basis of the Laser-Doppler signal. Such pragmatic success prompted the development of more advanced data compression techniques shortly presented. This chapter was previously published in [1]. The modes of perfusion of the microvasculature, i.e. the instantaneous movement patterns and the RBC content of vessels vary considerably, being influenced by at least four independent processes, namely • the periodic cardiac activity influencing arterial driving pressure, • the periodic respiratory activity influencing the pressure in the venous capacitance vessels, • the periodically modulated activity of the efferent sympathetic nervous system and • the periodically varying “myogenic tone” of smooth muscle, known to be strongly influenced of the action of gravity on the transmural pressure, and thence the tensilation of the vascular walls as “stimulus” for myogenic contraction (v.i.). While occasionally, the overall “activity fluctuations” have been suspected to reflect “chaoticity”, more detailed studies employing rigorous criteria for critical dependency on initiating conditions, systematic divergence of trends (or the magnitude of Lyapunov exponents) and tests of multidimensionality have disproved this insinuation (see, for example, Coluantani and Intagliatta. Therefore, our group has approached this topic under a somewhat different logic, namely that developed in physiological synergetics. In the previous publication from our own group, performed in the attempt to comprehend the true nature of the well-known (yet undefined) “flux motion” in Laser-Doppler recordings, we combined frequency analysis of the movement patterns with the frequency analysis of the blood content patterns (obtained by simple photoplethysmography) and scrutinised the details of their “temporal order” (in short the complex qualitative patterns) rather than the means of pressure, flow or

7 Synergetic Interpretation of Patterned Vasomotion Activity …

115

blood content and their amplitudes as a measure of variance (i.e. trivially quantitative aspects of a transport process). Initially, our approach was based on simple pattern identification (v.i.) allowing to distinguish what we termed “actively modulated” from what we termed “passively modulated” fluctuation, where the former was considered to be the consequence of constrictor activity at the observation site proper, and the latter was considered to be the net effect of arterial pressure fluctuations, respiratory and resistive changes in the entire cardiovascular system. This approach has indeed proved to be pivotal in overcoming the ideology of pure phenomenological description as it is common in the so-called nonlinear sciences, but proved to be insufficient in light of the many interacting influences now known (v.i.). More recently, a sophisticated set of algorithms has been developed in the Department of Physiology in Aachen (see van den Houten and Grebe) and was applied to hypercomplex neurophysiological data sets obtained in animal experiments at the Free University of Berlin (Lambert et al.). The programme, meanwhile commercially available under the trade name SANTIS® , now permits to identify on strictly objective criteria the cause for the “multidimensionality” of temporally fluctuating records monitoring the instantaneous movement of red blood cells and the instantaneous red cell content of the skin (or oral mucosa) under investigation. The various segments of this communication are presented with the aim of delineating the “logic” behind a set of monitoring and depicting strategies developed in the last years in the institutions represented by the authors. The details of the underlying physiological determinants of fluctuating blood content and blood cell displacement cannot be presented in the current context, these mechanisms, however, are well known at the molecular and/or ionic level (Siegel).

7.2 System Analytical Background of Non-invasive Diagnostical Procedure and its Practical Consequences in Complex Systems Portrayable as “Quasi-attractors” In our attempts to uncover the fluctuating “power” of various influencing parameters, we were guided by the systematic work of the late Hans Peter Koepchen, an internationally known authority on the autonomous nervous system, known for his penetrating investigations on the interaction between respiration and cardiac activities (as, for example, it determines systemic arterial pressure in animals and human). In his cooperation with Hermann Haken, the founder of general synergetics (a transdisciplinary set of theories concerning non-equilibrium phase transitions in dynamically driven and attenuated systems), these two authors opened a highly fertile approach to the comprehension of “complexity” in the realm of the autonomic nervous system. In coining the term “quasi-attractor”, they paraphrased the consequence of a putatively universal performance characteristic of the advanced central nervous systems (somatic as well as autonomous). There is a simple rule that needs to be applied: throughout the world of animated biosystems, functionally linked subsystems are

116

H. Schmid-Schönbein

only cooperating in a transient fashion. Expressed in the words chosen in 1990 by Haken and Koepchen, when referring to neurodynamic systems, the topological term “attractor” thence should not be used, but the logic behind this metaphor should be applied, the transiency notwithstanding. They state in 1990: In biological systems a rather complicated network is established in which control parameters of one subsystem may order parameters of another system or vice versa. This lead to a phenomenon in which a system does not stay in… (the scope of) … specific attractor all the time but may change between … (the influence of) … one attractor. Instead of attractors one has rather to speak of quasi attractors which exist for a while but the disappear or are replaced by new attractors.

It is this straightforward statement (corrected by short insertions in normal font) which terminated the futile attempts of “chaos theoreticians” to detect “deterministic chaoticity” (in the strict sense of this unfortunate word paraphrasing “quasi-chaotic coordination) in central nervous systems and/or their efferent outflows in the form of macroscopically detectable, periodically modulated activity. The term “quasiattractor” thence means “transiently operational attractors” or transiently consensualised activities of many subensembles of functional determinants for a given fluctuating activity. There microscopic and mesoscopic causes have been clarified by reductionist research during the last decade (or for that matter during the entire twentieth century): neuronal activation in actual fact always means short-lived disinhibition and passive release of previously stored energy. This sweeping statement holds water for the simple reason that all neuronally controlled activities are tonically inhibited and are only transiently disinhibited: phasic activity thus reflects the sequentiality of disinhibition and reinhibition, where the latter is accompanied by a phase of refractoriness (Schmid-Schönbein). As detailed in the latter communication, this follows from a theory concerning the “Integrative Action of the Nervous Systems” , which was originated in 1906 by Sir Charles Sherrington: after almost a century, his basic assumptions are generally accepted (see any textbook of neurophysiology, especially the popular monograph on “Essentials of Neural Science and Behavior” edited by Eric R. Kandel, the Nobel laureate in physiology (2000). As detailed in the secular work by Sherrington who must be considered to be the founder of physiological synergetics, one must start from the straightforward assumption that neuronal cells as well as neuronal cell pools (“dynamic ensembles”) are indeed being permanently inhibited and only therefore only transiently can become disinhibited. All analysis of cardiovascular and respiratory activities in awake human subjects must, therefore take notice of the well-established “fact of life” that “Phasic MacroActivity”(PMA) is always reflecting the consequences of “self-limiting” evasion from tonic microinhibition (TMI evasion), which, expressed in the terms of contemporary systems analysis, reflect not only the truly microscopic events associated with transmembranal and/or neuronal currents, but the mesoscopic activity taking place in pools of neurons (which are sometimes paraphrased as “hyperneurons”). This abstract logic has highly significant practical consequences, which are quintessential for the correct interpretation of non-invasively obtained data from effectors of the autonomous nervous system. Exclusively under conditions of prolonged physical activity (which is a state rarely monitored in basic physiology), a prolonged, but

7 Synergetic Interpretation of Patterned Vasomotion Activity …

117

again “limited” macro-activity of neuronal systems is provoked, a situation then characterised by far less temporal fluctuation (v.i. Outlook; Fig. 7.10). Straightforward consequences for both the theory and the practice of the field of non-invasive cardiovascular research follows suit: superposition effects become manifest as various possible PMA-episodes happen to occur either simultaneous (giving rise to beat phenomena as one of the consequences) or consecutively (giving rise to variation in the dominance of subsystems taking turn in controlling the temporal patterns instantaneously observed). It goes without saying that in failing to apply fundamental neurophysiology, all previous system analytical attempts within the field of periodically fluctuation biological phenomena were bound to be erroneous. As in the general sciences “chaos description” has made room for attempts of “chaos control”, it is now obvious that all conceivable control mechanism that have evolved in higher forms of animal life can be and must be placed into the centre of efforts to improve non-invasive diagnostic procedures. These obviously have the advantage of employing recording techniques that do not interfere with the dynamics of the activities the physician or researcher attempts to analyse. They are “hampered”, as it were, by the highly complex nature of the phenomena responsible for the “vitality” of the records obtained in awake human subjects. Most importantly in the context of all attempts to comprehend the nature of fluctuations in dynamic records obtained from the circulatory systems of volunteers and of patients, we have now shown that the “conventional” definition of “physical rest” had been a priori mistaken. This erroneous assumption is primarily caused by the limitations of either sensing devices or the stratagems used to analyse data: “lack of physical activity” cannot be equated with “inactivity” of the autonomous nervous system. Since we have been able to record the strong and irregular activity in the cutaneous perfusion and/or the cutaneous blood content, we now know that any human “subject” (normal or patient) sitting in a chair or lying in supine position does not necessarily also reflect “rest” of the autonomous nervous system. Instead, his autonomous nervous system is engaged in fluctuations (or reiterating TMI-evasion scenarios) most likely subsequent to disinhibitory–reinhibitory sequences on the level of spinal neurons responsible for cutaneous vasoconstrictor activity. Therefore, we have to start from the tacid assumptions that with respect to the autonomous control of cutaneous perfusion, the opposite of rest marks the dynamic patterns prevailing in recumbent subjects under conditions of “thermoregulatory indifference temperature” (21 °C–24 °C). Contrary to our own expectations, and contrary to textbook wisdom, this situation is highlighted by very pronounced, highly irregularly fluctuating activities related to such physiological situations as “thermal reflexes”, “mental activities” and, for the case of respiratory influences, vocalisation or even silent mentation. Understandably, this so-called resting situation is strongly reminiscent of “chaos” for naive observers, but in actual fact it is reflecting the very situation Haken and Koepchen had referred to, i.e. a sequentiality of short-lived activities of subpools of autonomic neuronal ensembles in the spinal cord. When expressing this physiological activity in terms of topology, it can be said that the latter are transiently “attracting”, as it were, the limit cycle type of periodically modulated “actions”; in the latter, the precise course of events in aperiodically yet

118

H. Schmid-Schönbein

Fig. 7.1 Paradigmatic example of the ca 0.15 Hz band activity, obtained by transmission photoplethysmography of the strongly hyperemic ear lobe in a subject in recumbant position 30 min of psychomotoric relaxation in a subject. In this situation, the autonomous vasomotor activity in the ear lobe microcirculation is completely eliminated, as evidence by the subjective feeling of cardiac palpitation in the ear lobe

reiterating processes appears to be “strangely configurated”. In short, then, vasoconstrictor episodes can be said to be an episode of PMA reflecting the microvascular sequelae of a transiently released neuronal activity. The latter, of course, initially reflects a TMI-evasion, but which is obligatorily followed by replenishing transients leading firstly to transfer blockade (“refractorisation”) and secondly to replenishing of the neuronal energy content. Therefore, the blood content and the resistive control of blood transfer and the level of retensilised vascular smooth muscle show the aperiodic and asymmetrically configurated fluctuations we have earlier described (see also Fig. 7.4). These considerations are far from “esoteric” but are rather “pragmatically” helpful in the pursuit of computer-based non-invasive cardiovascular diagnostics. Having emancipated ourselves from a much too simple concept of “physical rest”, we were able to look out for a somewhat better definition of the “psychomotor activity state” of subjects undergoing “non-invasive tests” of their cutaneous (and/or mucosal) microcirculation. As shown in an earlier publication from our group, we were indeed able to identify from transmission photoplethysmographic measurements of the hyperemic ear (Fig. 7.1) or, more recently, from reflection photoplethysmographic measurements of the hyperemic forehead (Figs. 7.2 and 7.3), a characteristic activity characterised by what we now term “0.15 Hz band activity”, i.e. a fluctuation of the blood content of vasoparalised skin which has a characteristic basic rhythm (about 6-8/min) and a characteristic “beat phenomenon” (a superposition effect) producing a kind of “spindle pattern” with a characteristic duration in the range of 1 min–2 min. We stress that both these patterns in the LDA and the rPPG records (i.e. sites of microcirculatory exchange events) are primarily a “passive phenomenon”, i.e. follow from the fact that as the active control agents for the ear lobe and forehead resistance vessels has been eliminated, so that the blood content and/or the velocity of red cell displacement in all microvessels follows the temporal evolution of the arterio-venous pressure difference, i.e. a clearly macrovascular set of events. Without going into the physiological, topological and thence theoretical details of the mechanisms causing this characteristic mode of operation, we reiterate our earlier proposal concerning its pragmatic utility: in the future, a much better “standardisation” of non-invasive measurement of cardiovascular fluctuations can be obtained

7 Synergetic Interpretation of Patterned Vasomotion Activity …

119

Fig. 7.2 Typical example of the temporal evolution of the ca 0.15 Hz activity in a naively relaxing subject. Note the gradual appearance of the activity in the range between 0.1 Hz and 0.2 Hz (stippled peaks), associated with gradually emerging activity in the range between 0.01 Hz and 0.1 Hz (low frequency band activity not associated with sympathetic nerve activity). Note also the absence of activity in the range of respiratory activity (0.25 Hz–0.3 Hz)

Fig. 7.3 Various examples of the emergence of ca 0.15 Hz activity in the reflection photoplethysmographic records from the locally heated (29 °C–31 °C) in subjects where the local microvascular tone, but perhaps not the tone in the larger arteries supplying the forehead skin was eliminated

120

H. Schmid-Schönbein

under the provision that in the individual subject or patient the photoplethysmographic, electrocardiographic and anemometric records are only taken after the 0.15 Hz band activity has emerged. All experimental results to be described in Sect. 7.5 were obtained under these highly restrictive experimental conditions, and it was this situation, which allowed the physiological interpretation we are now in a position to propose.

7.3 Subjective and Objective Techniques for “Feature Extraction” from Non-invasively Obtained Time Series As an added dividend of the synergetic interpretation of fluctuating data obtained non-invasively from awake human subjects, the Haken–Koepchen “quasi-attractor” hypothesis proved to be highly successful in distinguishing (on the basis of both subjective and objective criteria) the highly variable influences of the determinants mentioned above. Obviously, neuronally mediated vasoconstrictor activity stems from periodically modulated activities in the neuronal pools (studied by the Gothenburg group) which by its typically “nonlinear” or “eruptive” kinematics which due to lack of regular periodicity can be clearly distinguished from that originating in the membranes of arteriolar smooth muscles. As shown by Siegel, the latter is characterised by sinusoidal modulated myogenic activity with a stable attractor and, most importantly, a “discharge pattern” reminiscent of a linear oscillator. To comprehend this latter claim, it must be taken into account that the direct measurements of sympathetic efferent activity in both animals and man have not only produced reliable criteria distinguishing between skeleto-muscular and cutaneous vasoconstrictor ensembles in the spinal cord, but have shown that—again by temporality criteria—are clearly separable from cardio-accelerator, sudomotor and pilomotor neurons. Exclusively on the basis of the temporal configuration of the discharge patterns (see Häbler for a review of this topic), a differentiation between various subpools of sympathetic efferent neurons has become possible. Since the now classical work of the Gothenburg group, it is positively established from detailed and synchronised neurographic and hemodynamic data, that in keeping with the quasi-attractor concept, the cutaneous microcirculation is indeed linked to discrete “volleys” of the efferent activity of the sympathetic neurons responsible for cutaneous vasoconstriction. There are, however, other influences that must be taken into account. After our group as well as that of Jepsen et al. using only Laser-Doppler measurement have proven the influence of gravitational effects (presumably by the effect on the transmural pressure and thence the myogenic tone of vascular smooth muscle), the Haken– Koepchen logic can be applied on the basis of experimentally established physiological knowledge. The configuration of “episodal neurogenic vasoconstriction” is identifiable on the basis of their “non-sinusoidal” configuration and asymmetry of rapid constriction and slow relaxation (Fig. 7.4), and their predominance can be clearly related to environmental conditions (ambient temperatures well below 27 °C) and to

7 Synergetic Interpretation of Patterned Vasomotion Activity …

121

Fig. 7.4 Typical example of the simultaneous recording of transmission PPG of the hyperemic ear, the reflection PPG and the LDA signal from the volar side of the second finger of the right hand. Note that emergence of the 0.15 Hz activity in the ear lobe, but coherency of episodal vasoconstrictor activity in the cutaneous microcirculation: simultaneous reduction in blood content and blood cell displacement with typical “asymmetry” of both records. (see also Fig. 7.5)

gravitational straining of microvessels by placing the hand well below the heart. Sinusoidal cutaneous fluctuations are clearly distinguishable by their temporal configuration; moreover, their emergence is related to ambient temperatures well above 27 °C. In other words, this temporal pattern occurs under conditions of eliminated thermoreflex activity, thus allowing a slower pacemaker to become the basis of an attractor for typical sinusoidally configurated fluctuation (in the range of 0.02 Hz–0.03 Hz, 1/min–3/min). The latter pattern could regularly be seen when the hand of healthy subjects was being placed well below the heart level (i.e. when the microvessels were exposed to the added load of the blood column subjected to gravitational acceleration). We propose to call these sinusoidally configurated fluctuations “myogenic tensilation nitric oxide relaxation“ (Myo-NOR-Activity): expressed in the terminology of physiological synergetics, it is likely to reflect the disinhibition due to stress-dependent Ca++ -channels in antagonistic cooperativity to shear-dependent release of nitric oxide from the endothelium acting as a powerful vasodilator mechanism due to smooth muscle relaxation. Based on the quantification of the number of vasoconstrictor episodes in unit time, and on the configuration as either neuronally mediated or sinusoidal fluctuations, we were able to propose a comprehensive “qualitative” classification scheme concerning attractors of periodically modulated cutaneous perfusion. The latter could soon be complemented by the identification of respiration-related activity, which we initially found under two highly different situations, namely in the LDA records in severe cases of decompensated peripheral arterial obliterative disease (a pattern referred to as “slow waves” in the original publication of Scheffler and Rieger, and in the mucosal

122

H. Schmid-Schönbein

photoplethysmographic activity of severe gingivitis (v.i.). While it is currently unsettled whether this rhythm reflects mechanical influences on the venous outflow or a neuronally mediated influences (Mück–Weymann), it is clear that respiration-related activity (RRA) has to be considered as a putatively powerful determinant of the fluctuations seen in the perfusion dynamics the superficial tissues amenable non-invasive techniques. These insights led to a strategy in which we strive to reduce the complexity of the actual measuring devices sensed by patients and volunteers, while the boundary conditions for measurements have to be closely standardised. This can be done by placing the subjects into thermocontrolled rooms (with ambient temperature ranging between 15 °C and 37 °C and by constructing a device allowing to place the hand at the heart level, well below and well above the heart level. Furthermore, a simple reflection photoplethysmograph was connected to a battery heated plate (3 cm in diameter), thus allowing to derive photoplethysmographic data from a vascular region devoid of vasomotor activity (heating above 31 °C leading to local vasorelaxation and thus to a situation where passive reaction to changes in the arteriovenous pressure gradient determine the blood content and thence the photoplethysmographic signal (and its temporal variation). As described in detail before the data obtained from the forehead reflection photoplethysmograph, those from the LDA device (we mostly used the Periflux Models 301 without filter) during measuring periods between 4.5 min (in the experiments on healthy and diseased gingiva) and 30 min (in measurements on the volar finger microcirculation. The data from the PPG of the glabella and the volar finger, as well as those obtained in the LDA systems used without filters were digitised on line and were fed into task adapted PC where a dual record was constructed by a software of own design (Blazek and Hülsbusch). Thus, from each measuring site, two complementary records were obtained, namely a “double plot” with compressed time series showing the gross appearance of the fluctuations in amplitudes and a Fourier transform. The information concerning frequency distribution was obtained by detection of discrete frequency bands based on the evaluation of the data for intervals between two minutes and the whole measuring interval (5 min–30 min). In light of the fact that the most important discrete activity bands were uncovered in the range between 0.25 Hz (respiration related, about 15/min) and 0.01 (various highly different activities in the range of 1/min–3/min), the data were initially displayed in such a fashion, that the peaks could be seen as a function of the logarithm of the frequency. As mentioned before, these plots not only proved to be far superior to power spectra for the work in cardiovascular physiology but opened an entirely new “venue” to the comprehension of complex time series. We propose to call these plots “frequency chromatograms”, and their application was later extended into the construction of “time frequency plots” (v.i. Fig. 7.10 in the section Outlook) which greatly aided in the comparison of various types of low frequency activities in the fluctuation of the effectors of the autonomous nervous and/or ventilatory system. As an added dividend, the strategem of “normalising” all frequency chromatograms to the power of the fluctuation in the ca 1 Hz band (cardiac activity in adult subjects during physical rest) allows to

7 Synergetic Interpretation of Patterned Vasomotion Activity …

123

compare the powers of distinctly different activities displayed in the range of “low frequency” (well below 0.1 Hz, v.i.). It is common knowledge that the techniques developed in the nonlinear sciences (“chaos theory”) depend on the identification of characteristic features of the rhythmically modulated output from a “black-box”, i.e. a system too complicated to scrutinise without destroying its basic modes of operation. In this situation, it is possible to refocus the attention in scientific inquiry: rather than quantifying the “magnitude” of an activity (reflected in the amplitude of a fluctuation or its power, namely the square of the amplitude), the temporality of a process can likewise be quantified objectively. We wish to stress that in last analysis, this logic was first introduced into the natural sciences in theoretical physics: we follow by extrapolating a discovery made and popularised by Ruelle and Talkin in their attempt to comprehend complexly timed hydrodynamic turbulence. A similar logic was later applied to the field of nonlinear optics in the early phases of constructing pulse-Lasers and led Haken to institute formal synergetics. However, as we are now portraying the temporality of dynamically vaxing and waning periodic activities in the form of “attractor portrays”, we no longer paraphrase as “strange”, but rather as “natural”. This functional logic makes necessary to develop problem appropriate techniques of data compression, which in taking up the potentialities of Fourier transformation extends their powerful utility to situations no longer allowing its use. There are many limitations which in a strict physical sense forbid the conventional technique of portraying the dynamics of non-stationary, nonlinear and noise afflicted time series as power spectra (see textbooks of theoretical physics and especially van den Houten. It proved to be helpful and depicts instationary activities subsequent to the transformation of amplitude modulated time series into the frequency domain. Consensualisation appears and disappears, i.e. the activity shows the “transiency” so typical for activities “driven” by central nervous systems. In such situations, conventional power spectra are of not just of questionable theoretical utility: pragmatically speaking, they blur, rather than disclose the information contained in the recorded activities. Using the Fourier transformed information about the dynamics of cutaneous perfusion (PPG and LDA activity), it can be pragmatically utilised by displayed it in the mode of a kind of “histogram” where the amplitude density of fluctuations in a given frequency band of closely related activities of clearly different origin are displayed in a non-conventional manner. Rather than constructing plots of powers as the function of the numerical value of a frequency range, plots were constructed that display amplitudes as the function of the logarithm of the respective medians (“frequency chromatograms”). Originally, these were plotted by averaging the information gathered within distinct observations periods, e.g. 5 min or 15 min. This “coarse grained” analysis had already greatly helped as a pragmatic strategy of information compression, but was, of course, insufficient in light of the concept of “quasi-attractors”. In order to cope with emerging and submerging cooperativities between subsystems (ventilatory, cardiac, arteriolar, neurodynamic) in a more fine-grained fashion, van den Houten developed a strategy to use “gliding” windows, i.e. to scrutinise temporal changes in the cooperation/superposition of subsystems.

124

H. Schmid-Schönbein

Using a number of additional techniques (such as wavelet-time-transient reconstruction, autoregressive moving average (ARMA-time series), estimates of instantaneous powers) the logic of “quasi-attractors” was extended by van den Houten with the aim of displaying “stochastic coherency” and “stochastic synchronisation” in such a fashion that both the appearance and disappearance of instantaneous frequency synchronisation and phase synchronisation can be intuitively comprehended by physicians without specific training in modern topology and nonlinear analysis. This goal has been achieved in the program package (SANTIS® ) which combines the clarity of the frequency chromatographs with the ability of simple time series to deal with non-stationarity as the most essential aspect of neurodynamic activities at large: here, the concept “quasi-attractors” is being visualised by “mountains of relevancy” separated by valleys of irrelevancy on the basis of intuitively comprehensible vaccilating pattern on the screen of computers (see Fig. 7.10). These novel instruments can replace a host of techniques from the late 1990s allowing to quantify the spatio-temporal patterns of highly complex, superimposed rhythmic time course as they have been developed in the last two decades: they all depend on the ability to extract “information concerning instationary temporality” as it is, for example, contained in the numerical value of Lyapunov exponents, or in the dimensionality of processes (which, of course, is “fractal” in the cases of natural quasi-attractors and, of course, is reflected in the “distance from equilibrium” (or neg-ergodicity indices responsible for the positioning of complex attractors in the so-called phase space). In depicting both the time–frequency plots and the original time series, easily comprehensible information concerning the “overall scope” (German “Spielraum”) of the complex system in question can be displayed. In simply displaying in the form of “double plots” both the original time series (with variable temporal resolution but reliable information concerning the “distance from equilibrium” and the time frequency diagrams (representing emergence and submergence of cooperation and thence of “quasi-attractors”), the entire information concerning the spatio-temporal patterns of cardiovascular events of any kind cannot only be retained at will, but can be displayed in a form in which the previous history as well as the response to defined stimuli can be monitored objectively. Records thus obtained in the form of both fine-grained and coarse-grained plots are easily intuitable and can be adapted to the problem under investigation. Last not least, the results can be stored and “replayed” at will and in a more or less comprehensive fashion. As will be shown in the closing Outlook section, under exceptional, but, of course, physiologically important conditions, stationarity of periodically fluctuation cardiovascular and respiratory parameters can in fact be induced, for example, when volunteers are made to exercise (easiest when driving a bicycle ergometer) in a fashion that seems “comfortable” to them. During the “steady state” of such an exercise, superimposition of periodically varying entities can be also seen, for example, in the data shown in Fig. 7.5: displaying a situation where a volunteer was asked to drive the ergometer into a periodicity convenient for him when loaded to 120 W. There is conspicuous consensualisation between respiratory periodicity (upper panel) and periodicity of arterial blood pressure, but also a strict coordination between ventilatory and skeletomotor activity, and therefore between cardiac and skeletomuscular

7 Synergetic Interpretation of Patterned Vasomotion Activity …

125

Fig. 7.5 Schematic representation of the characteristics of the “aperiodically” placed and “asymmetrically configured” records of LDA and PPG under the influence of neurogenic vasoconstriction (which needs to be clearly differentiated from myogenic vasomotion originating in the smooth muscle of the resistance vessels proper, see Siegel)

activity. This is a good example of “load-dependent ordering”, if the same occurs without external drive, one is confronted with “synergetic stereotypes”, i.e. highly pathological situations. In the case of prolonged drive, however, a unique combination of central nervous subsystems (circulatory reflex centres, respiration centres, higher centres for the spinal control of skeletomuscular activity) is found, under which physiologically important conditions one can indeed “identify” permanently dominating attractor domains (“permanent” meaning that they reflect a non-fatiguing and uninhibited “steady state” activity, where integer number coordination as a mode of central nervous “binding” is indeed frequently found. See Fig. 7.10 and the short Outlook to future techniques for displaying emerging and submerging “consensualisation” scenarios.

7.4 Outline of the Methods and Prototypical Examples for Complexity Reduction in Non-invasively Obtained Measurements Modern information technology as it is nowadays available in inexpensive laptop computer has made this new look a complexity in everyday clinical problems possible. In trying to display some of the pivotal results underlying the interpretation of the “double plots”, the current communication exemplifies the explanatory power of synergetic system analysis with the help of a program package now commercially available (van den Houten). After having established as the first step to “standardise” the measuring conditions for the long-term registration of the patterns found in cutaneous perfusion, regular “control” for the “physical” and “mental” rest in the individual subjects is now possible. We stress that contrary to a widely held misconception, the mere assumption of the supine position and the control of ambient

126

H. Schmid-Schönbein

temperature in the so-called thermoregulatory indifference range (21 °C–24 °C, see textbooks of physiology) is bound to produce most vivid fluctuations of the cutaneous vasoconstrictor activity in both the skin of either the ear lobe or the forehead (of course only unless they are treated with a vasoparalytic procedure). Therefore, initially the ear lobe and later the glabella region (mid of the forehead) were subjected to vasorelaxation (either by vasoparalytic ointment of the ear lobe or the localised heating of the skin, 29 °C–31 °C). This made it possible to identify a characteristic pattern of either LDA or PPG fluctuation in the frequency range of about 0.15 Hz, in most cases associated with a marked beat phenomenon. A typical example recorded in transmission photoplethysmography after pharmakological vasoparalysis of the ear lobe is shown in Fig. 7.1 the temporal evolution of this activity in what is now called “ca 0.15 Hz band” is shown in Fig. 7.2. Note that when measuring “naive subjects” during 60 min, the activity in this range is gradually augmented, In addition, there is a low power fluctuating activity in the range between 0.01 Hz and 0.1 Hz with variable peaks representing the variably pronounced superposition of the 0.15 Hz activity (v.i.). The ca 0.15 Hz band activity is also regularly found in the reflection photoplethysmographic records obtained in hyperemic forehead of subjects and patients in physical relaxation, but only occasionally in the “perfect expression” (Fig. 7.3A). In each volunteer or patient, it evolves in a different fashion (Fig. 7.3B–E), this notwithstanding, subsequently reported measurements were only taken after this “resting activity” had been established. Being thus in command of a novel interpretation of the physico-physiological “resting state”, it turned out to be easy to observe a universally valid pattern for the perfusion and the blood content and the blood cell displacement in the variable networks of microvessels in the subcutaneous vasculature. In each and every subjects, the typical pattern originally established by the group in Göthenburg with the help of joint measurement of neuronal activity and either reflection photoplethysmography or LDA activity in the region supplying by cutaneous vasoconstrictor neurons was found. As illustrated in Fig. 7.4, without exception, an “episodic perfusion pattern” was found, characterised by • Strict coherency between LDA and rPPG, • Lack of regular periodicities and • Marked asymmetry of both signals, with rapid decline and slow recovery of the perfusion by way of a decrescendo–crescendo sequence (as detailed in Schmid– Schönbein et al.). As detailed in the communication mentioned, we have clearly shown that this activity pattern governing the cutaneous perfusion under all conditions of normal perfusion in “normal” ambient temperature conditions does not reflect myogenic vasomation, but rather neurogenic vasoconstriction. Since 1994, the latter statement could be corroborated by the simple experimental procedure of eliminating neurogenic drive (in taking measurements at ambient temperatures above 27 °C) while enhancing myogenic activation (in placing the hand well below heart level, thus enhancing intravascular pressure by gravitational effects). In this case, a totally different pattern, albeit in the same frequency range can be regularly observed,

7 Synergetic Interpretation of Patterned Vasomotion Activity …

127

namely remarkably regular sinusoidal (“periodic”) and symmetric fluctuations in the LDA records, but “silent” rPPG records. More importantly, since it could be easily seen that a similar rhythm regularly prevails in patients with autonomic nervous system diseases (without nutritional deficits), for example, in patients afflicted by combined diabetic and alcoholic neuropathy also showing other signs of this disorder (see Fig. 7.5). Therefore, the assumption seems to be warranted that the sinusoidal activity we have attributed to myogenic vasomotion is detectable by the same combination of data: regularly timed sinusoidal fluctuations of the LDA record in the range of 0.02 Hz–0.03 Hz. As will be demonstrated elsewhere (Schmid-Schönbein, Blazek, Scheffler and Perlitz), we have now established that under the premis that there is true vasoparalysis in the cutaneous microvessels (by pharmacological intervention or by local heating the skin), there is a high degree of coherency in the patterns found in the arterial blood pressure, measured by means of FINAPRESS® or directly by intravascular manometry and the pattern of the perfusion of either the hyperemic ear lobe or the hyperemic glabella area (Fig. 7.6). From this, it is concluded that our original assumption concerning blood content fluctuation as an indirect method of detecting the patterns of higher-order arterial blood pressure waves (Hering-Mayer-type fluctuations) is indeed borne out by direct comparison. This has important practical significance: it means that the inexpensive and completely non-invasive rPPG assessment of the glabellar perfusion pattern provides sufficient information concerning discrete activity bands of microvascular perfusion that can be either related to neuronally controlled vasoconstriction or (under conditions where the neurogenic vasoconstriction is set to rest in warm environments) as

Fig. 7.6 “Double plot” of the reaction of the volar cutaneous microcirculation roared at the heart level and at an ambient temperature of 18 °C (periodic vasoconstriction in response to powerful sympathetic efferent drive presumably interrupted by episodes of vasodilatation). During the vasodilatation, a paraesthetic feeling appeared in the fingers, presumably caused by intermittent release of local vasodilator agents (“miniature hunting reaction” according to LEWIS)

128

H. Schmid-Schönbein

passive reflection of influences of the arterial pressure. Conversely, as the investigations are being carried out in uncomfortably cool environments, the periodicity of the fluctuations is highlighted by very powerful vasoconstrictor episodes, which, however, are regularly “interrupted” by a typical “escape” reaction, in which there is a short, self-limiting vasorelaxation interlude. These intermittencies of the perfusion are interestingly accompanied by a subjective feeling of “warming” of the hand, and it can be proposed that the vasodilator reactions constitute some local event reminiscent of the “hunting reaction” long known to occur when extremities are exposed to very cool environment (see textbooks of physiology). Note in Fig. 7.5 that an entirely different pattern is seen in the time course of the vasoconstriction/vasorelaxation sequences, and, most importantly, the normalised power of the activity in the range between 0.01 Hz and 0.1 Hz is very marked. Based on the results so far displayed, one can make the following preliminary statement. In using the “double plots” from two independent measuring sites and three different time physiological events, one can see that the ca 0.15 Hz band activity has a characteristic temporal pattern and is reflected in an activity in the 0.01 Hz to 0.1 Hz range which represents beat phenomena. The myogenic vasomotion with its sinusoidal activity also shows pronounced power in the 0.01– 0.1 Hz range, but an entirely different combination of patterns in the time series and the frequency chromatogram. Lastly, powerful vasoconstrictor activity with escape phenomena associated with high energy in the 0.01 Hz to 0.1 Hz band as found in the records in very cool environment represents still another type of reaction. These three different patterns, to which the ones obtained in the measurements in thermoregulatory indifference temperatures must be added, clearly show that at least four different physiological situations can be identified with marked activity in the so-called low frequency range (associated by most authors with the activity of the sympathetic nervous system). In short, then, only the conservation and utilisation of information in the amplitude and the frequency domain allows to identify the various different mechanisms determining the “quasi-attractors” identifiable in healthy human subjects and in patients. This straightforward logic can be extended by additionally monitoring the effects of venous pressure alterations in association with increase in venous return subsequent to deep inspiration (v.i.). As a preliminary conclusion, computer-based noninvasive cardiovascular diagnostics can start from the assumptions that we are now in command of reliable indicators for the “penetration” of effectors outside of the range of vasoconstriction/vasomotion in the cutaneous skin. Suffice it to say in the present context that in the domain of cutaneous perfusion, when psychophysical relaxation has been established, a more or less rapid evolution of a wide variety of attractors has become possible. There are many combinations of regularly configurated superpositions of the cardiac activity, of the respiratory activity (0.25 Hz–0.3 Hz at rest in adults subjects), of the activity in the 0.15 Hz range (characterised by beat), and of at least three periodic mechanisms in the range between 0.1 Hz and 0.01 Hz: they exist in combination and their activity can now be objectively documented. As the investigations were extended to cases of severe peripheral occlusive arterial disease with deficiency of epidermal perfusion, and to other tissues, a further elucidation of the

7 Synergetic Interpretation of Patterned Vasomotion Activity …

129

determinants of microvascular perfusion patterns. They were disclosed by results obtained under pathological conditions by way of comparative measurements of mucosal blood flow on the one hand, and by comparison between normal and pathologically altered perfusion in the lower extremity in patients with obliterative vascular disease. Interestingly, both experimental series disclosed an added “attractor” for the cutaneous perfusion, namely the arterial pressure on the one hand, and the influence of respiration on the other. Moreover, these measurements showed that independent (non-invasive) measurement of the ventilatory movements (by simple strain gauge or thermometric techniques) will have to be implemented in the future in order to identify the absence or presence of influences passively altering blood content and/or movement patterns of RBC in the microvasculature. A set of typical examples reflecting this important physiological fact is shown in Fig. 7.6, taken from the gingiva of healthy subjects. As shown in the compressed time series displayed in Fig. 7.6, in each cases we found a remarkable coherency of the perfusion pattern in the hyperemic glabella region and that in the gingiva: even when either an involuntary inspiration gasp occurred or when the subjects were asked to inspirate, the “gasp” activity, (see Mück-Weymann) was found in both cases. We stress that when such measurements were taken in a cold environment and without vasoparalysing the cutaneous vessels, a complete dissociation between cutaneous and mucosal activity patterns were found. Even more interesting were the data on the perfusion of the inflamed gingiva: while there is a first glance difference between the healthy (pale pink) mucosa and the inflamed one (deep red), the magnitude of the LDA signal was not significantly different, in some cases even lower than in the healthy controls. However, there was markedly enhanced flow pulsatility (intuitively visible in the compressed time series in the width of the fluctuations under identical recording conditions) and, most conspicuously, a clear respiration-related activity, i.e. spontaneously occurring fluctuations of mucosal perfusion in the range between 0.25 Hz and 0.3 Hz (Fig. 7.7). Taken in combination, these two qualitative “pattern”, objectively recordable by the measurement of the pulsatility index (difference of the peak and the minimum voltage of the LDA output) divided by the instantaneous mean) and the normalised power in the ca 0.3 Hz band range (power of the ca 1 Hz band taken as 1.0) allow to clearly separate between the healthy and the inflamed microcirculation of the gingival mucosa, see Fig. 7.8. There are several conclusions to be drawn from these mucosal measurements: • it is obvious that the normal gingiva seems to be influenced almost entirely by the activity patterns governing arterial blood pressure, • this fact can be easily detected non-invasively by simply taking simultaneous measurements of the perfusion of the hyperemic glabella area, and • gingivitis with marked macroscopically visible hyperaemie is not associated with accelerated blood cell displacement, but with a unique combination of enhanced flow pulsatility (obtainable by LDA measurement) and respiration-associated fluctuations of the blood content (measureable by PPG). A closely related finding was early recorded in severe cases of peripheral obliterative occlusive arterial disease (POAD), which had originally been called “small

130

H. Schmid-Schönbein

Fig. 7.7 Typical recordings of the glabellar PPG, the mucosal LDA and PPG signal from the healthy gingive of volunteers. Note that marked coherency of the signals in all three records

Fig. 7.8 Typical recordings of the glabellar rPPG and the mucosal LDA in cases of severe gingivities. Note the appearance of respiration-related activity in the inflammed gingiva

waves” (by Scheffler and Rieger). These early observations have since been studied more extensively by the combination of skin rPPG, skin LDA and glabellar rPPG which was developed for this purpose). In about 40% of POAD patients (with stage II FONTAINE, claudication), the characteristic respiration-related perfusion activity was detected, the power of which is inversely related to the so-called Doppler pressure, i.e. peak systolic poststenotic arterial pressure in the claudicants. Physiologically speaking it is important to note that not only the well-established primary

7 Synergetic Interpretation of Patterned Vasomotion Activity …

131

pulsations of perfusion patterns are a mere passive response, but slower activity bands easily identifiable by the comparative frequency chromatograms make it possible to distinguish them from other types of flow or blood content fluctuations. There is nothing “chaotic” in these measurements of obviously pathological microcirculation, instead there is a remarkable “order” (more appropriately referred to as “synergetic stereotypes”) which is associated with a situation where vasoconstrictor control mechanisms no longer prevails. This situation, of course, is most characteristic of perfusion patterns associated with degenerative neural diseases: physiological “attractors” disappear, pathological ones become prominent in the cutaneous microcirculation. In the latter disease states, we corroborated an earlier finding, namely a frequent dissociation between the LDA and the rPPG patterns, thereby providing a posteriori justification to institute the somewhat complex measurement scenario we have constructed in our group. A typical example is shown in Fig. 7.9, showing that in a patient with clinically established autonomous dysreflexia, there is a “silent” recording of blood content associated with a clearly varying recording of the LDA pattern. The double plot again allows clear diagnosis: absence of vasoconstrictor activity (silent rPPG, no aperiodic and asymmetric patterns in the LDA record), yet powerful myogenic vasomotion with 0.02–0.03 Hz activity in the LDA frequency chromatogram.

Fig. 7.9 Severe autonomic dysreflexia in a patient with combined alcoholic and diabetic neuropathy (Courtesy of Dr. Reinders, Department of Neurology, University of Düsseldorf). Typical recording of the ear lobe tPPG, the volar finger rPPG and LDA. Note the absence of PPG activity and the marked sinusoidally fluctuation activity in the finger of the patient

132

H. Schmid-Schönbein

7.5 Summarising Discussion—Physiological Analysis by Way of Identification of Temporal Patterns It is trivial to state that the measurements of the dynamics of cardiovascular reactions are hampered, as it were, by three simple “facts” of life: the data obtained • are notoriously noise afflicted, • are derived from reactions with very pronounced nonlinearity (i.e. unpredictable cause–effect relationship typical for vasomotor activities and • are naturally non-stationary. Each and every measurement reported in this review reflects the situation topologically dominated by Koepchen’s and Haken’s “quasi-attractors”, which is perhaps a triviality reflecting the obvious fact that “transiency is normalcy” in all systems driven by central nervous systems. It is also trivial to state that once the natural “nonlinearity” and, most importantly, the eruptivity of cardiovascular movement patterns have been accepted as “fact of life”, all attempts to use conventional data compression stratagems must be discontinued: the “calculation” of average and mean can be said to represent illegitimate destruction of information (“data murder”), and even the calculation of “power” in conventional Fourier spectra leads to serious information distortion (“involuntary data slaughter”). In the future, such procedures will no longer be acceptable: in the project of physiological synergetics, much more “system appropriated” data compression stratagems have been developed, which on the one hand are intuitively comprehensible for physicians (most of them are excellent pattern recognisers) and still convey the kind of information used in nonlinear sciences. Re-reviewed under the synergetic and physiological “logic” here delineated, one must take biological “ordering” in the realm of the cutaneous and the mucosal microcirculation as the consequence of “drive dependent consensualisation” of a priori existing movement patterns. In being able to identify quite easily the temporal ordering in the “double plots”, one can apply conventional physiological concepts in terms of “latent information” extractable from simple non-invasive measurement. The logic presented concerns the dynamics of cardiac and non-cardiac determinants for the displacement modes of blood cells in the microvascular beds. The techniques used in “computer-based non-invasive cardiovascular diagnostics” indeed prove to be ideal for the task of providing kinetic long term information from a wide spectrum of “sources”. As one can monitor both the movement patterns of accelerated erythrocytes and the variations in blood content of microvascular beds, one can quite easily identify the main “attractors” for rhythmically modulated processes. They depend on the periodically varying influence of other physiological systems onto the dynamics of a hypercomplex flow situation which cannot possibly be assessed by direct means. Simple concept makes it comprehensible that there must be a superposition of the movement pattern of these “attractors” upon the cardially accelerated basic acceleration–deceleration scenarios in the blood movement driven by the heart through microvessels. The pragmatics results of such a straightforward logic spring to the eye: since in the normal cutaneous microcirculation there are no sinusoidal fluctuations (in

7 Synergetic Interpretation of Patterned Vasomotion Activity …

133

either blood displacement rate or blood content) the so-called flux motions obtained by the majority of users of Laser-Doppler equipment in the realm of the skin in awake human subjects most likely have nothing in common with “myogenic vasomotion” but most likely represents consequences of clearly separable neuronally caused events. Since the combined neurographic and rPPG and/or LDA investigations, we know that rapidly accelerating and slowly decelerating episodes are associated with bouts of actions potentials and the pauses between them. They represent a scenario called “relaxation oscillations” (or the effect of “kicked oscillators”) leading to limit cycle dynamics which have nothing in common with the well-studied spontaneous fluctuations in arteriolar tone seen in experimental animals under anaesthesia. Each individual limit cycle episode, in turn, is reflecting topologically and physiologically analogous acceleratory phases which are totally independent from “physiologically analogous” deceleratory phases. Therefore, the limit cycle patterns are bound to be influenced by additional “attractors”, including either boosting or suppressing effects of respiration. Note that even variably operational “muscle pumps” exert influences on the blood content of the skin: this is a fact put to practical use by Blazek’s tests of venous competence. It is difficult to underestimate the significance of these concepts regarding our comprehension of the “control” of the microcirculation in awake human subjects. In having rigorously applied the concepts of synergetics, and here especially the concept of temporal binding on the one hand, and of phase synchronisation on the other, a conclusion refuting the widely held conviction concerning the role of “myogenic vasomotion” as prime control agency for cutaneous microcirculation is being refuted. As further “fine-graining” in the portraying of the temporalities of periodically varying movement patterns has now become possible, more subtle display of rhythmically varying influences onto the ensemble of vascular smooth muscles, onto the cardiac muscle as a “load dependent pump” and, of course, of those exerted by rhythmically modulated neurodynamic drive and inhibition by the two parts of the autonomous nervous system (v.i.) has become easily detectable. Since the vagal part of the autonomous nervous system is only represented in the cardiac activity, fluctuating parasympathetic activity is a priori felt in the respiratory arrythmia of the heart. Since, however, there can be coherency between the vagally exerted influence upon the instantaneous cardiac 1/f activity and the respiration-related increase in venous return and thence cardiac filling: for this reason, respiration-related activities can be exerted onto blood content and/or blood cell movement in the ca 0.25 Hz–0.3 Hz band activity. The results reviewed in this lecture have clearly shown that we are now in command of clearly identifiable criteria allowing “differential diagnosis” of certain clearly identifiable, mostly transiently dominating “attractors” as sources of comprehension in a system not amenable for direct inspection. For example, since we can identify the general states (either clearly diagnosable neuropathy or thermal interruption of natural neuronal activity), we can reproduce at will (by warm ambient temperature) the very situation under which myogenic vasomotor activity emerges: then, there conductance alterations include sinusoidally modulated myogenic vasomotion most readily operational under the influence of tension-induced smooth muscle

134

H. Schmid-Schönbein

constriction and shear-induced smooth muscle relaxation (NO-vasodilatation mechanism). In contrast, it is equally easy for any user of LDA or rPPG techniques to provoke by simple cooling very powerful, aperiodically modulated and asymmetrically reacting attracting transient processes in the form of strong peripheral effects of neurogenic vasoconstriction. Having performed both provocation tests, each user of LDA and/or rPPG equipment will easily comprehend that that the “normal” mixtures of reactions that are recorded at “normal temperature” are simply the reflection rapidly changing quasi-attractors upon the “tone” of vascular smooth muscle. It is well known, of course, that under the influence of phasically acting alpha adrenergic (“activating“) and beta-adrenergic (“relaxing“) mechanisms, the fluctuating activity of various spinal efferent sympathetic neurons can be detected. If fine graining of noninvasive microcirculatory measurements is further improved, even lymphomotoric activity can, in principle, be measured in clinical settings, they are bound to exert an effect on the “attractor configuration“(a situation not yet incorporated into the analysis of microcirculatory dynamics, but see Christ). In addition, there are other aperiodic influences that can modify the exact configuration and positioning of attractors within the phase space. These include such parameters as hemorheological events, endothelial leakage with hemoconcentration and, most importantly in many severe clinical situations, a pressure effect that can be paraphrased as compartment symptomatology (compaction of tissue content and/or the sequelae of abnormal external compression). It can be expected that in these, still other attractor combination dominate the picture, an aspiration born out by occasional applications of our measuring devices to isolated cases of hyperviscosity, of chronic vasogenic and lymphogenic edema and to the symptomatology found subsequent to crush injury and ulcerations. As can be seen, the strategem we developed many years ago is beginning to yield pragmatically relevant dividends: where all attempts to find “quantitative” abnormalities of microvascular perfusion in clinical settings have largely disappointed, the potentials of the combined assessment of as many non-invasively obtainable data as possible should be pursued. For many applications, the ingenious FINAPRESS® system (allowing to record phasically the arterial pressure in a “non-traumatising” fashion) can be recommended. However, for the interpretation of discrete alterations this method is wanting: owing to the unpleasant pain of the inflated cuff this method does not really qualify as “non-invasive”.

7.6 Outlook By a combination of physical, physiological and—most importantly—pathophysiological situations, we were able to overcome one of the most sterile attempts for the comprehension of cardiovascular functions, namely the delineation of putatively “chaotic” dynamics as derived from merely phenomenological interpretation of in the non-stationary and nonlinearly regulated time series (e.g. ECG-recordings). During the 1990s, a serious confoundation of “order” with “pathological stereotypes” had occurred, and thence it had been concluded that it were “healthy to be chaotic”.

7 Synergetic Interpretation of Patterned Vasomotion Activity …

135

Not surprisingly, a number of amiable layman in command of powerful computers went on to “portray” the dynamics what they mistook as “the activity of the heart”: this then led to the horrendous misconception that chaoticity were a sign of normal function, whereas “order” were a sign of pathology. Without such prejudices, it has now become possible to solve clinical problems. Most importantly, this is achieved by minimally invasive techniques and the utilisation of the plethora of data compression techniques put at the disposal of medical practitioners. As we are beginning to develop problem adapted semantics on the one hand, and powerful problem adapted means of “feature extraction” from prolonged time series on the other the time has come to shoulder tasks that cannot be mastered in any other fashion. As shown in a recent publication concerning hypercomplex data gathered in animal experiments, this set of algorithms can in fact be utilised in comprehensive analysis of far more complicated and extended processes recorded invasively (which are more complex than those occurring in clinical settings (see Lambert et al.). However, the same technique is also capable of demonstrating order under load in human subjects by non-invasive means. As a demonstration of the most important potentials of such a routine, which in the future will allow to display the highly informative frequency chromatograms as a series of “time –frequency plots” , the periodically fluctuating activities are being displayed in an intuitively comprehensible fashion. As shown in Fig. 7.10, the data obtained from ergometer exercise in a task where the volunteer was asked to strain himself in such a fashion that he felt “subjectively comfortable” while working at about 120 W (displayed to him as feed-back of rotational speed of the bicycle), a remarkably regular activity of the respiration, the cardiac activity and the skeletomuscular was seen, in which there was a remarkable integer number relationship between their periodicities. In other words, the apparent chaoticity (or the emergence and submergence of quasi-attractors) at rest had not just made room to a most regular configuration of powerful stationary attractors, but tended to so-called n:m synchronisation: under physical load, there is 1 to 4 relation of respiration and cardiac frequency, and 1– 2 correlation between skeletomuscular activity and ventilation. While at present it has to remain unsettled weather this remarkable entrainment reflects just frequency synchronisation or phase synchronisation, this simple experimental paradigm can be accepted as a typical case of load dependent plasticity (or drive dependent “movement consensualisation”), see Tass and Varela et al. This is obviously a situation, where with the help of non-invasive computer-aided diagnostic tools a most remarkable and highly stable “dynamic order” can be documented in an easily intuit able fashion. Moreover, many of the parameters representing the diagnostic instrumentarium developed in nonlinear dynamics can be applied directly or indirectly in the future application of the “double plot” concepts, i.e. the simultaneous display of the original time series (and thence the projection of the amplitude dynamics) and the time – frequency plots (i.e. projection into the frequency space). Most importantly, these easily comprehensible plots reflect the appearance of gliding coordination in the sense of Erich von HOLST, the divergence of trajectories (or the occurrence of positive Lyapunov exponents and, most interestingly, the fine-grained or the coarse-grained

136

H. Schmid-Schönbein

Fig. 7.10 Prototypical demonstration of a time –frequency plot in a subject exposing himself to “comfortable” physical (bicycle ergometer) exercise of 120 W. The subject was asked to adjust the rotational speed so that he felt physically comfortable. Note the stable frequencies of skeletomuscular activity, respiratory activity and cardiovascular activity (recorded by FINAPRESS® signal) an the integer number (n:m) correlation of ca. 0.6 Hz, ca 0.12 Hz and ca 2.4 Hz in ventilation, skeletomuscular activity and cardiac 1/f dynamics

features underlying fractal dimensionality of complex physiological time series). In order to understand these in a more fundamental context, it will be necessary to relate PMA patterns observable in non-invasive techniques to the concepts we have concerning the underlying TMI evasion mechanisms and their direct and indirect sequelae: this will likewise be the topic of a communication to the forthcoming conference on computer-based non-invasive diagnostics.

7 Synergetic Interpretation of Patterned Vasomotion Activity …

137

Reference 1. H. Schmid-Schönbein et al., New paradigm: objective quantification of temporal patterns in skin perfusion, in Proc. 9th Int. Symp. CNVD 2000, VDI, Düsseldorf, ISBN 3-89653-881-0, pp. 75–98 (2001)

Chapter 8

A Self-Organized Rhythm in Peripheral Effectors: The Intermediary Rhythm Appears as 0.15 Hz-Band Activity Volker Perlitz

Abstract The dynamics of a rhythm band observed first in the ear skin microcirculation of awake human subjects were scrutinized using a naturalistic study design and nonlinear frequency and phase analysis. Since this frequency band with its centre between 0.12 and 0.18 Hz was slower than the respiratory rhythm but faster than the 0.1 Hz sympathetic dominated rhythm, this rhythm band is referred to as intermediary rhythm. Using nonlinear frequency and phase detection methods pronounced differences between episodes exhibiting amplitude modulated intermediary rhythm activity and episodes void of it were identified. Varying coherence, the modulation of peak to peak distances and fluctuations of energy transfer (modulation of the dV/dt in the protosystolic and the diastolic phase of cardiac driven pulsations) and, most importantly, the objective documentation of phase jumps relate the intermediary rhythm to the principles of ‘synergetic self-organization’ as discussed by Haken. The emergence of the intermediary rhythm is suggested to originate in non-equilibrium phase transitions in the network of lower brainstem neurons and is further linked to parasympathetic neuronal effectors, e.g. parasympathetic innervations of facial skin microcirculation. Thus, the intermediary rhythm comprises oscillations of a dynamic equipoise allowing either deceleration to slower sympathetic rhythms or acceleration to faster vagal and respiratory related rhythms.

8.1 Introduction Gestalt can be both bound to physical structure or temporal order. The latter is called rhythm and is an essential feature of physiological processes. In the cardio-vascular as well as in the respiratory system rhythms often exhibit changes in amplitude and frequency at widely stable margins, thus forming bands. Known for long are various rhythms which were identified and correlated to physiological functions of e.g. the autonomic nervous system (ANS). There is ample consent that high frequency (HF, V. Perlitz (B) Simplana GmbH, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_8

139

140

V. Perlitz

0.15 Hz to 0.4 Hz) and low frequency (LF, 0.04 Hz to 0.15 Hz) oscillations dominate in blood pressure fluctuations, heart rate variability, skin blood flow, sympathetic nerve discharges and neuronal activity in the brain stem. This classification is widely agreed to reflect the so-called sympatho-vagal balance [1–9]. Recently, our group has described a highly stable intermediary rhythm manifesting itself as a band centered on 0.15 Hz ± 0.03. Hence, this band is exactly at the watershed between high and low frequency oscillations. First noted in cutaneous vascular bed circulation of the human ear lobe and forehead skin, this intermediary rhythm emerged gradually as test subjects adjusted to the experimental setting. This process was interpreted to correspond with a reduction in psychophysical load [10–16]. In this communication, we further elaborate physiological principles as well as the relevance of the intermediary rhythm in the context of synergetics [17].

8.2 Methods Detailed descriptions of our methodological approach have been given elsewhere [12, 13, 18]. In general, informed written consent, a thorough medical history, and socio-demographic data were obtained from test subjects before the study, which was approved by the University’s (RWTH Aachen) review board. We renounced all invasive interventions, such as pharmacological interventions to reduce psychomotor drive but used instead of a well-known auto-suggestive relaxation technique, the autogenic training (AT). This modality has been readily demonstrated to induce predictably psychophysical responses in volunteers as well as in a great number of clinical conditions in patients [11, 14, 19, 20]. Hence, it qualifies as a naturalistic study design in that physiological coordination responses during complex cognitivepsychophysical drive reduction become observable. Data presented were chosen from a healthy subject recruited from the college community (female, age 24 years, nonsmoker). The tests she participated in were recorded on three different occasions (T0, T1, T2) for 10 min during naïve relaxation, followed by 10 min training the practice of AT. Lacking any experience with AT, these tests were performed on the test subject in a supine horizontal position, at an ambient temperature kept constant at 22 ± 2 °C. Prior to all recording sessions an initial period of 30 min passed which was needed to attach probes (see below). This also allowed the test subject to become familiar with the laboratory situation. Only then the recording session was initiated. It is of paramount relevance that she was encouraged during all recording sessions to keep her eyes shut yet avoiding to fall asleep.

8 A Self-Organized Rhythm in Peripheral Effectors: The Intermediary …

141

8.2.1 Data Acquisition All physiological data were recorded on-line with a personal computer equipped with an analog-to-digital (AD) converter (Data Translation, DT 2812 a), as well as a software package (DiaDem®, National Instruments). Data were recorded at a sampling rate of 300 Hz with 12-bit resolution. They were stored for further processing on digital media. We used a photoplethysmographic probe (PPG) which was attached to the glabella forehead region known to display little or no vasoconstrictor activity at room temperature for this could have altered rhythms in these delicate signals. Irrespective of certain interpretative deficiencies regarding the amplitude of these signals, sufficient precision in the temporal domain of these signals was demonstrated by simultaneous recording of PPG and LASER-Doppler [21]. To detect respiration-induced fluctuations contained in cardiovascular signals, a belt equipped with a strain-gauge (ADMS) was used for the registration of the frequency and amplitude of respiratory induced thoracic movements.

8.2.2 Data Analysis Data were analyzed off-line using software designed for the non-linear analysis of multiple time series of bio-physiological data (commercially available via Procalysis®, Simplana GmbH/Aachen-Germany). Slow trends in DC-offset were removed. The following nonlinear algorithms were applied to data sets. Multiscaled Time-Frequency Distribution (mTFD) The changes of frequencies contained in physiological signals can be ideally analyzed using Morlet-wavelet based multiscaled Time–Frequency-Distribution (mTFD) as this method combines analysis of temporal information concerning fluctuations in both frequencies and amplitudes [18, 22]. For mTFD computations, the interval length analyzed comprised usually the entire recording time with an average of approx. 180.000 samples. The frequency range was 0.04–0.6 Hz which equaled period length from 25 to 1.66 s. As each window contained 7 periods, window lengths were from 11.66 to 175 s. With frequency and time resolution set in general to 200, this resulted for an epoch of 600 s an overlap for the lowest frequency of 172 s, for the highest frequency of 8.66 s. This overlap helped smooth spectra within the mTFD.

142

V. Perlitz

8 A Self-Organized Rhythm in Peripheral Effectors: The Intermediary …

143

Fig. 8.1 a–e: a Original PPG recording showing typical patterns of waxing and waning: emerging and submerging of large so-called beat oscillations result in spindle-shaped envelopes, taking turns with unmodulated sections. b The Time Frequency Distribution analysis of the PPG original time series gives an overview of the two dominating frequency bands: the ca 1.2 Hz band representing sinus node activity (detailed in Fig. 8.1c), and the 0.15 Hz rhythm band representing the dynamics of fluctuating central lower brainstem activity (Fig. 8.1d). The interaction between both bands is governed by a set of surprising yet fundamental physiological features. The dynamics of the ca 1.2 Hz band activity is inversely related to amplitude modulations contained in spindles: the power of 1.2 Hz activity decreases as spindles peak, and vice versa. This is paralleled by a pattern of frequency modulation in the ca 0.15 Hz band activity: during each unmodulated section the 0.15 Hz band activity reaches a minimum; during the 480th and the 540th s this activity completely vanishes. As shown during this episode, the 1.2 Hz band is powerful, yet oscillating around its mean frequency. Extending an interpretation initially proposed by Winfree who referred to these incoherent activities as “black holes”, we label these episodes “white holes". Figure 8.1e shows the respiration rhythm at ca. 0.3 Hz which is at a 2:1 coupling ratio with the 0.15 Hz rhythm band in the PPG signal

8.3 Results A „representative “example for a plethora of similar recordings are given in Fig. 8.1a: a rhythm which envelopes the 0.15 Hz beat reappears (with great variation) and subsequently disappears about once every 100 s (i.e. exhibits a 0.0 l Hz band activity). Magnification displays that.

Fig. 8.2 a–b: Another example is shown in these graphs. This readily illustrates principles outlined for figs. 1a-d above. Here, spindles stretch over an extended period on a rather constant level of power. In the absence of beat oscillations, the power of the 1.2 Hz band is greatly enhanced while the 0.15 Hz band activity is interrupted (at the 450th s).

144

V. Perlitz

Fig. 8.3 Two sections cut from the above PPG recording subjected to high-pass filtering exceeding the frequency of respiration; rescaled to superimpose for easy comparison. Upper trace: spindle episode; lower trace: unmodulated “white hole”

• acivity in the 0.15 Hz band differs in its long term evolution and that • amplitudes of the 0.01 Hz-activity is much more pronounced as it is interrupted with zero power in the pauses. We assume that the dominating frequency of the superimposed rhythms and their varying „amplitudes “ is not attributable to simple superposition of two closely related frequencies in the sense of an interference, but that. • emerging and submerging patterns results from a specific recruitment of cooperating oscillators whereas • submergence is associated with superposition of „non-coherent oscillators and thus be likened to „white noise“ (and/or „weakly coloured“ superposition).

8.4 Discussion Employing time-frequency distribution analysis we suggest that during natural, yet triggered psychophysical relaxation the human facial skin microcirculation is dominated by a rhythm at ca. 0.15 Hz. This is in keeping with similar findings presented by other researchers using of spectral analysis [23–32]. Analogously, a rhythm at ca. 0.15 Hz was also observed to emerge in reticular neurons of the lower brainstem of freely breathing anaesthetized dogs in response to maneuvers lowering the global level of activity, such as an, e.g., administration of pharmaceutics. Due to dynamic variations in the given frequencies, the rhythm was termed 0.15 Hz frequency band.

8 A Self-Organized Rhythm in Peripheral Effectors: The Intermediary …

145

These variations itself represent an essential physiological property, namely adapting to order various rhythms in a n:m integer number mode. Thus, the function of this 0.15 Hz rhythm band elicited coupling between cardiovascular and respiration time series which was enhanced as amplitudes of the 0.15 Hz rhythm in reticular neurons increased.[4, 18]. Analogies between canine and human were also found in amplitude dynamics. Spindle wave shaped amplitude modulation epochs were prevalent in the recordings of the reticular neurons as well as in human recordings of facial skin blood content oscillations [4, 11–13, 18]. These analogous findings in canine and human lead us to suggest that the 0.15 Hz rhythm band in facial skin microcirculation of humans results from coherent oscillatory neuronal activity in the lower brain stem. In the present communication, we are able to further our understanding of essential rhythmic principles governing the cardiovascular-respiratory coordination in human. We demonstrate the interaction between spindles shaped amplitude modulations of the 0.15 Hz rhythm band and the cardiac attractor. There is an evident inverse relationship between the power of the spindle and the power of the cardiac 1.2 Hz frequency band. As spindles reach a maximum, the power of the 1.2 Hz frequency band becomes distinctly weaker. When lacking amplitude modulations in the 0.15 Hz rhythm band, the power of the cardiac attractor reaches a maximum. This suggests closer considerations of the interplay of frequency and amplitude modulations. As for the dynamics of the frequency dynamics exhibited by the 0.15 Hz rhythm band, we have expanded our understanding labelling this rhythm since 0.15 Hz is the frequency precisely between the frequencies of rhythms exhibited by the two branches of the autonomic nervous system, the sympathetic and parasympathetic nervous system. We ,therefore, chose to refer to this rhythm as the intermediary rhythm abandoning a merely phenomenological term in favour of a term which outlines an important physiological feature. As this concept of the amplitude modulated „intermediary rhythm” is subjected to closer scrutiny, it is important to start from the well-known fact that relay neurons are surrounded by pools of small interneurons. As one further assumes that the latter are predisposed to function as “inhibitors” (owing to their neurosecretory products, namely glycine or γ-amino buturylic acid (GABA), it can be insinuated that by mechanisms of “collateral inhibition”, the supra-threshold excitation of any relay neuron is associated with delayed inhibition due to circulating excitations within the pool of small interneurons. Furthermore, in assuming that small neurons are easily excited but poorly inhibited, while large neurons are difficult to excite but easily inhibited, the emergence, submergence and amplitude modulation phenomena of the intermediary rhythm can be explained by straightforward application of conventional concepts of sequential excitation and inhibition in neuronal pools. In closing the description of this putative mechanism, it can be assumed that on the one hand, the global efficacy of the excitatory and inhibitory ionic currents are “integrated” at the axon hillock, where on the other hand peripheral and central input is either augmenting the probability of supra-threshold relay excitation or the opposite, namely the hyper-polarization and subsequent sustained inhibition (a phenomenon also referred to as “occlusion” in terms of conventional neurodynamics). Anatomy teaches, of course, that there are

146

V. Perlitz

many small neurons surrounding each relay neuron. Thence, the finite duration of synaptic delay adds up to several hundred μs up to as many as several ms. Consequently, circulating excitatory currents can easily survive the rapidly inhibited activity in the concomitant relay neuron. This will then justify a holistic interpretation of selfterminating “discharge trains” well known from the EEG (especially in the “spindle” shaped alpha waves found in the occipital cortex). The intermediary rhythm is not only found in human, but also in the canine experiments performed by Lambertz et al. [4, 18] is readily explainable by straightforward consideration of initial phase synchronization and subsequent phase desynchronization in neuronal pools (in this case of the common brain stem). One can therefore insinuate that all rhythmically modulated discharge patterns exhibiting emerging and submerging trains of action potentials can be explained by the unavoidable interference between phase synchronization and phase desynchronization. However, this assumption includes the additional concept that the temporal beat phenomena must be accompanied by rhythmically modulated spatial expansion and retraction of excited sub-pools within a given neuronal ensemble as it has been assumed for the explanation of the so-1called synaptic “occlusion”. When viewed from system theoretical aspects, the described behavioural traits of agito-inhibited neuronal pools are in keeping with a concept concerning “information” presented many decades ago by Ernst and Christine von Weizsäcker. These authors claimed that as a system is being sequentially informed, there is a rapid (e.g. linear) increase in “corroboration”, which, however, is automatically accompanied by an equally rapid (also linear) fall in novelty value. This concept can be operationalized not only for linear increments but for non-linear and/or exponential increments as well. Furthermore, it can be “biologised” by assuming the re-iteration of an efficient stimulus input to depend on an excitation threshold. It can then be stated that initially, the “novelty value” is associated with spread of excitation (by divergence), whereas corroboration is associated with automatic increment in the inhibitory activity just referred to. As the intermediary rhythm was originally discovered under the conditions of naïve psychomotor relaxation, its manifestation with typical beat phenomena was taken as a “tool” to differentiate “excited” from “unexcited”, and the former condition was considered to allow reproducible PPG and Laser-Doppler measurements in clinical routine. The newer insight into the “stability” of ca O.15 Hz-Band activity during triggered auto-suggestive relaxation including closure of the eyes makes it necessary to re-evaluate the situation in terms of “non-linear dynamics”: the beat phenomenon found originally can now be taken to represent a “threshold phenomenon” (v.i.), where its “stable” manifestation is indicative of “synergetic self-organization” far from thermodynamic equilibrium [17, 21, 33–35].

8 A Self-Organized Rhythm in Peripheral Effectors: The Intermediary …

147

8.5 Conclusion The improvement of computer-based evaluation of non-invasively obtained cardiovascular data is important for future research activities since the reduction of primary experimental expenditure and progressively lowered invasiveness extends the range of application. This brings about a dualistically operational advantage: • recordings are far less invasive than any currently known invasive cardiovascular technique (including the electrocardiogram which many patients detest due to fear to be electrocuted); • patients are not only able to naively relax under these conditions but they are willing to subject themselves to long term measurement. Two important advantages follow suit: a. The fidelity of the measurement increases due to the lack of adverse effects of the experimental setting (i.e. less noise is being recorded), and b. information yet concealed can be retrieved by digital conversion of the data from the conventionally analyzed amplitude domain into the frequency domain. Looking at the different patterns, an entirely new “class” of scientific interpretations concerning the instantaneously changing forcefulness of activating and inhibiting influences can be proposed. Where in the past, a somewhat trivializing physical interpretation of the “superposition” of co-operating and “antagonizing” influences was attempted we can now differentiate between variable “stability” of a given “mode of operation”. Thence, entirely new information concerning subtle differences in the coping of the subject with the experimental boundary conditions becomes available. It follows from this statement that perhaps the most important progress lies in the discovery that using a minimally invasive method a “behavioural trait” of awake human subjects could be discovered which not only relates to theoretically important medical “categories”, but which aids in the practical application of the non-invasive tool. While our investigation shows that a simple “definition” of “resting state” is perhaps a difficult task, the insights now available make it possible to exclude an individual subject and/or patient from measurement when the intermediary rhythm has not yet emerged.

References 1. L. Bernardi, A. Radaelli, P.L. Solda, A.J.S. Coats, M. Reeder, A. Calciati, C.S. Garrard, P. Sleight, Autonomic control of skin microvessels: assessment by power spectrum of photoplethysmographic waves. Clin. Sci. 90, 345–355 (1996) 2. L. Bernardi, D. Hayoz, R. Wenzel, C. Passino, C. Calciati, R. Weber, G. Noll, Synchronous and baroreceptor-sensitive oscillations in skin microcirculation: evidence for central autonomic control. Am. J. Physiol. Heart Circ. Physiol. 273, H1867–H1878 (1997)

148

V. Perlitz

3. C. Cogliati, R. Magatelli, N. Montano, K. Narkiewicz, V.K. Somers, Detection of low- and high-frequency rhythms in the variability of skin sympathetic nerve activity. Am. J. Physiol. Heart Circ. Physiol. H1256–H1260 (2000) 4. M. Lambertz, P. Langhorst, Simultaneous changes of rhythmic organization in brainstem neurons, respiration, cardiovascular system and EEG between 0.05 Hz and 0.5 Hz. J. Auton. Nerv. Syst. 68, 58–77 (1998) 5. A. Malliani, M. Pagani, F. Lombardi, S. Cerutti, Cardiovascular neural regulation explored in the frequency domain. Circulation 84, 482–492 (1991) 6. N. Montano, T. Gnecchi-Ruscone, A. Porta, F. Lombardi, A. Malliani, S.M. Barman, Presence of vasomotor and respiratory rhythms in the discharge of single medullary neurons involved in the regulation of cardiovascular system. J. Auton. Nerv. Syst. 57, 116–122 (1996) 7. N. Montano, C. Cogliati, V.J.D. Dias da Silva, T. Gnecci-Ruscone, A. Malliani, Sympathetic rhythms and cardiovascular oscillations. Auton. Neurosci. 90, 29–34 (2001) 8. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation 93, 1043–1065 (1996) 9. S. Zhong, S. Zhou, G.L. Gebber, S.M. Barman, Coupled oscillators account for the slow rhythms in sympathetic nerve discharge and phrenic nerve activity. Am. J. Physiol. 272 (Regulatory Integrative Comp Physiol 41), R1314–1324 (1997) 10. V. Perlitz, B. Cotuk, S. Haberstock, E.R. Petzold, Selbstorganisation kutaner Perfusionsrhythmik bei therapeutisch induzierter psychovegetativer Entspannung. In: Neurobiologie der Psychotherapie (Hrsg. G. Schiepek). Schattauer, Stuttgart-New York (2003) 11. V. Perlitz, B. Cotuk, G. Schiepek, A. Sen, S. Haberstock, H. Schmid-Schönbein, E.R. Petzold, G. Flatten, Synergetik der hypnoiden Relaxation. Psychother. Psych. Med. 54, 1–9 (2004) 12. V. Perlitz, M. Lambertz, B. Cotuk, R. Grebe, R. Vandenhouten, G. Flatten, E.R. Petzold, H. Schmid-Schönbein, P. Langhorst, Cardiovascular rhythms in the 0.15 Hz band: Common origin of identical phenomena in man and canine in the reticular formation of the brain stem? Pflugers Archiv—Eur. J. Physiol. 448, 579–591 (2004) 13. V. Perlitz, B. Cotuk, M. Lambertz, R. Grebe, G. Schiepek, E.R. Petzold, H. Schmid-Schönbein, G. Flatten, Coordination dynamics of circulatory and respiratory rhythms during psychomotor relaxation. Auton. Neurosci. 115(1–2), 82–93 (2004) 14. V. Perlitz, H. Schmid-Schönbein, A. Schulte, J. Dolgner, E.R. Petzold, W. Kruse, Effektivität Des Autogenen Trainings. Therapiewoche 26, 1536–1544 (1995) 15. S. Ziege, Optoelektronische Analyse von aktiven und passiven Hautperfusionsrhythmen und deren Bedeutung hinsichtlich der zentralen vegetativen Regulation. Ph.D. Thesis, RWTH Aachen University (1992) 16. S. Ziege, H. Schmid-Schönbein, R. Grebe, E. Martin, Long-term registration of cutaneous microcirculation during general anesthesia. Int. J. Microcirc. 17, 385–394 (1997) 17. H. Haken, Information and Self-Organization in Complex Systems. Springer Series in Synergetics. (Berlin, Heidelberg, New York, 2000), pp. 1–35 18. M. Lambertz, R. Vandenhouten, R. Grebe, P. Langhorst, Phase transitions in the common brainstem and related systems investigated by nonstationary time series analysis. J. Auton. Nerv. Syst. 78, 141–157 (2000) 19. G.D. Jacobs, J.F. Lubar, Spectral analysis of the central nervous system effects of the relaxation response elicited by autogenic training. Behav. Med. 15, 125–132 (1989) 20. J.H. Schultz, Das Autogene Training (Thieme Verlag, Stuttgart, New York, 1973) 21. H. Schmid-Schönbein, S. Ziege, R. Grebe, V. Blazek, R. Spielmann, F. Linzenich, Synergetic interpretation of patterned vasomotor activity in microvascular perfusion: discrete effects of myogenic and neurogenic vasoconstriction as well as arterial and venous pressure fluctuations. Int. J. Microcirc. 17, 346–359 (1997) 22. J. Morlet, G. Arehs, I. Fourgeau, D. Giard, Wave propagation and sampling theory. Geophysics 47, 203–236 (1982) 23. I.M. Braverman, The cutaneous microcirculation: ultrastructure and microanatomical organization. Microcirculation 4, 329–340 (1997)

8 A Self-Organized Rhythm in Peripheral Effectors: The Intermediary …

149

24. P.D. Drummond, The effect of sympathetic blockade on facial sweating and cutaneous vascular responses to painful stimulation of the eye. Brain 116, 233–241 (1993) 25. P.D. Drummond, Lacrimation and cutaneous vasodilatation in the face induced by painful stimulation of the nasal ala and upper lip. J. Auton. Nerv. Syst. 51, 109–116 (1995) 26. H. Izumi, Nervous control of blood flow in the orofacial region. Pharma. Ther. 81, 141–161 (1999) 27. M. Podgoreanu, R.G. Stout, H. El-Moalem, D.G. Silverman, Synchronous rhythmical vasomotion in the human cutaneous microvasculature during nonpulsatile cardiopulmonary bypass. Anesthesiol 97, 1110–1117 (2002) 28. E.G. Salerud, T. Tenland, G.E. Nilsson, P.A. Öberg, Rhythmical variations in human skin blood flow. Int. J. Microcirc. Clin. Exp. 2, 91–102 (1983) 29. J.S. Schechner, I.M. Braverman, Synchronous vasomotion in the human cutaneous microvasculature provides evidence for central modulation. Microvasc. Res. 44, 27–32 (1992) 30. D.G. Silverman, R.G. Stout, F.A. Lee, E.M. Ferneini, Detection and characterization of cholinergic oscillatory control in the forehead microvasculature in response to systemic alpha-agonist infusion in healthy volunteers. Microvasc. Res. 61, 144–147 (2001) 31. D.G. Silverman, R.G. Stout, Distinction between atropine-sensitive control of microvascular and cardiac oscillatory activity. Microvasc. Res. 63, 196–208 (2002) 32. T. Smits, J. Aarnoudse, J. Geerdink, W. Zijlstra, Hyperventilation-induced changes in periodic oscillations in forehead skin blood flow measured by Laser Doppler Flowmetry. Int. J. Microcirc. Clin. Exp. 6, 149–159 (1987) 33. H.P. Koepchen, Physiology of rhythms and control systems; an integrative approach, in Rhythms in Physiological Systems. Springer Series in Synergetics, ed. by H. Haken, H.P. Koepchen (Berlin, 1991), pp. 3–20 34. H. Schmid-Schönbein, S. Ziege, The high pressure system of the mammalian circulation as a dynamic self-organizing system, in Rhythms in Physiological Systems. Springer Series in Synergetics, ed. by H. Haken, H.P. Koepchen (Berlin, Springer, 1991) pp. 77–96 35. H. Schmid-Schönbein, S. Ziege, A. Scheffler, V. Blazek, R. Grebe, Attractors and quasiattractors in the cutaneous perfusion in human subjects and patients: “chaotic” or adaptive behaviour? J. Auton. Nerv. Syst. 57, 136–140 (1996)

Chapter 9

Analyzing Pain and Stress from PPG Perfusion Signal Patterns Marcus Koeny

Abstract In this chapter, different methods for pain and stress assessment based on PPG perfusion signal pattern analysis are presented. Hence, patients under narcosis cannot be asked for their pain sensation, and the assessment of pain and stress must be done based on vital signs measurement, like heart rate or blood pressure. Hence, PPG and ECG are the vital signs, which are going to be assessed during every surgical intervention, and different methods for stress assessment based on PPG and ECG are introduced.

9.1 Introduction Pain is a subjective feeling, which influences patient’s well-being. Even narcotized patients react to pain, for example, with an increase of heart rate or blood pressure depending on the depth of narcosis. Strong reactions to intraoperative pain can cause postoperative stress and complications. In the recovery room and postoperative care, pain therapy is important. An optimized pain therapy improves patient’s recovery from surgical interventions and increases the well-being of the patient. Intraoperative pain assessment is done by observing vital signs, and postoperative pain assessment is done by patient’s self-evaluation.

9.2 Pain and/or Stress? The autonomic nervous system (ANS) controls the most important body functions automatically. It is divided into sympathetic and parasympathetic nervous systems. M. Koeny (B) Medical Information Technology, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University, Aachen, Germany e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2021 V. Blazek et al. (eds.), Studies in Skin Perfusion Dynamics, Biological and Medical Physics, Biomedical Engineering, https://doi.org/10.1007/978-981-15-5449-0_9

151

152

M. Koeny

The sympathetic nervous system increases vital functions of the body, like heart rate and blood pressure during episodes of stress. The parasympathetic system brings vital functions back to a normal level. Though most functions are regulated automatically, some functions can be influenced such as breathing. This fact has a significant influence on pain or stress assessment, which is discussed later in detail. In fact, the ANS is influenced by stress, where pain is one of the most important stressors that influence the ANS. Different indices have been developed to measure the level of stress and pain. The indices that can be computed from ECG and PPG are introduced in the following section.

9.3 Fundamentals Stress influences ANS, which in turn influences measurable parameter like heart rate and blood pressure. Hence, most algorithms for pain and stress assessment are based on evaluation of these parameters. Pain may be the most significant stressor for the body, but other stressors too influence the ANS. In fact, the stress measurement leads to pain measurement, when the influence of other stressors are reduced.

9.3.1 Heart Rate Variability Heart rate is controlled by ANS and the variation in heart rate (heart rate variability) enables us to quantify the function of ANS [1]. Time domain analysis of heart rate variability (HRV) is often used as a parameter to assess the health of heart [2]. However, the frequency domain analysis is important for stress assessment. Frequency distribution is computed using Fourier transformation or wavelet transformation. Figure 9.1 shows a frequency spectrum with three significant peaks. The frequency spectrum is divided into four main frequency bands (Table 9.1). The first peak at 0.1 Hz (a) is caused by the baroreflex, the second peak (b) is caused by respiratory sinus arrhythmia (RSA), and the third peak (c) is the first harmonic of the RSA. The baroreflex is associated with internal blood pressure regulatory mechanisms [1]. The RSA peak moves according to the breathing frequency and amplitude. Further, breathing amplitude varies depending on the stress level of the patient [3]. Therefore, the RSA forms the basis of computation of the analgesia nociception index.

9 Analyzing Pain and Stress from PPG Perfusion Signal Patterns

153

Fig. 9.1 Frequency spectrum of the HRV computed with a short-time Fourier transformation over 30 s

Table 9.1 Frequency bands of the HRV with main physiological associations according to [1] Designation

Frequency band

Reason

Ultra low frequency (ULF)

< 0.003 Hz

Unknown

Very low frequency (VLF)

0.003 Hz to 0.04 Hz

Thermoregulation

Low frequency (LF)

0.04 Hz to 0.15 Hz

Baroreflex

High frequency (HF)

0.15 Hz to 0.4 Hz

Respiratory sinus arrhythmia (RSA)

9.3.2 The Analgesia Nociception Index The analgesia nociception index (ANI) is an index that measures and quantifies RSA. It was developed by MDoloris Medical Systems (Lille, France). The ANI computation is based on the ECG signal and can be divided into three main parts according to Logier et al. [4]: 1. Determination of the HRV from ECG signal 2. Filtering the HRV in the band of RSA (0.15 Hz – 0.5 Hz) with a wavelet filter 3. Estimating the area under curve (AUC) of the filtered signal Heart rate variability is estimated by analyzing the beat-to-beat intervals (differences between the R-peaks) from the ECG signal. A special filter is used to remove extra systoles and artifacts from the beat-to-beat interval series [5]. Afterwards, the beat-to-beat interval series is resampled to 8 Hz, normalized over time windows of 64 s and filtered between 0.15 Hz and 0.5 Hz using DB2 discrete wavelet filters [4]. Figure 9.2 shows that the amplitude is higher during non-stress episodes compared to episodes with stress.

154

M. Koeny

Fig. 9.2 Comparison of RRhf between non-stress episode (top) and stress episode (bottom)

The parasympathetic tone is computed by calculating the AUC over four time periods of 16 s (A1–A4). The minimum of the four AUC episodes (AUCmin) is selected and used for ANI index computation using Eq. 9.1. AN I =

100 · (α · AU Cmin + β) with : α = 5.1, β = 1.2 12.8

(9.1)

Figure 9.4 shows a sequence of ANI associated with painful surgical events. The ANI is designed to be computed using standard ECG derived from the patient monitor with no need for additional monitoring. ANI can be computed using PPG and invasive blood pressure recordings too [6]. Computation of ANI using many sources improves the reliability, especially in situations where a signal is influenced by artifacts. In postoperative care, the ANI can be acquired without disturbing the patient with cables.

9.3.3 Surgical Stress Index The surgical stress index (SSI) was developed by Huiku et al. [7]. During evaluation, different parameters were used such as heart beat interval (HBI), amplitude of PPG signal (PPGA), blood pressure and pulse transition time. Analgesic drug concentration was used a reference to assess the stress level. Best correlation results between analgesic drug concentration and SSI were found using the combination of PPGA and HBI. Therefore, SSI is computed using Eq. 9.2. SS I = 100 − (0.7 · P P G An + 0.3 · H B In )

(9.2)

9 Analyzing Pain and Stress from PPG Perfusion Signal Patterns

155

Fig. 9.3 Example episode of surgical stress index (SSI)

with P P G An as PPG amplitude and H B In as heart beat interval. To match a range between 0 and 100, the PPGA and HBI values must be normalized [7]. A histogram transformation based on thousands of PPG signals is used to handle inter-patient variability. The SSI is available in GE Healthcare patient monitors or as standalone device. According to GE Healthcare, the SSI is only valid for narcotized patients. Hence, it cannot be used for optimization of pain therapy in postoperative care. Figure 9.3 shows the differences between the ANI as analgesia related index and SSI as stress related index.

9.4 Intraoperative Assessment of Pain Intraoperative assessment of pain is difficult, because patients under narcosis cannot be asked for their pain score (VAS score). Therefore, anesthesiologists must assess the level of analgesia by observing patients’ vital signs and other conditions during the surgery. The following analysis has been performed on the data from the University Hospital Aachen [7]. The ANI and the SSI have been computed and evaluated using data from various surgeries.

9.4.1 Analgesia Nociception Index Since the ANI based on ECG is available as a commercial product, it does not need a validation [8]. To validate ANI computation from PPG, the PPG signal must be from a valid source. Therefore, a self-developed implementation [7] was used to compute

156

M. Koeny

the ANI from ECG and from PPG signal. Then, the computation of HRV from PPG was validated, and the ANI-PPG was evaluated using recorded patient data. To validate HRV computation from PPG, many studies have been performed previously [9, 10]. Unfortunately, most of these studies were set up in a controlled environment where patients underwent no or only less hemodynamic changes. In such studies, the correlation between ECG-HRV and PPG-HRV can differ due to physiological reasons. One of the most important aspects is the different location of measurement (central vs. peripheral) and thus the influence of local phenomena, like vasomotion, to the PPG signal. Figure 9.4 shows the BB series and the corresponding ANI indices. The curves are very similar even after painful events such as skin incision and drilling. Furthermore, the ANI differs (after the suture event) when BB series differs as well. In this case, the BB series differs due to artifacts. To verify this assumption correlation analysis, overall 44 selected surgical interventions was performed. The mean correlation of RR intervals was about r = 0.75. The correlation between ECG-ANI and PPG-ANI was r = 0.5 (Fig. 9.5). Therefore, we assumed that the PPG-HRV as valid source for PPG-ANI computation [6]. In situations where narcosis is insufficient, the validity of ANI is important. Therefore, the ANI based on ECG and PPG has been analyzed focussing on hemodynamic changes in the 44 surgical interventions. Table 9.2 shows the correlation for blood pressure changes with an r = 0.892. The correlation in context of heart rate changes is quite good as well with r = 0.894 (Table 9.3).

Fig. 9.4 Comparison between BB-intervals and ANI from ECG and PPG

9 Analyzing Pain and Stress from PPG Perfusion Signal Patterns

157

Fig. 9.5 Correlation analysis of ANI during surgical interventions

Table 9.2 Correlation of ANIs and SSI with blood pressure changes Correlation Blood pressure Change

Correlation

Change

ANI-ECG

ANI-PPG

SSI

1

−0.015

0.001

−0.0101 0.221

Significance ANI-ECG

ANI-PPG

SSI

0.852

0.989

N

148

148

147

148

Correlation

−0.015

1

0.892

−0.016

Significance

0.852

0.000

0.843

N

148

148

147

147

Correlation

0.001

0.892

1

0.000

Significance

0.989

0.000

N

147

147

147

147

Correlation

−0.101

−0.016

0.000

1

Significance

0.221

0.843

0.996