Handbook of EEG Interpretation [3rd Edition] 9780826147080, 9780826147097, 0826147089

Citation preview

TH I R D ED I TI O N

THIRD EDITION

Ha ndbook of

INTER PR ETATION

William O. Tatum, IV, DO

Using a visual approach to identifying EEG waveforms, this handbook is the prime point-of-care reference on all major EEG topics: normal and abnormal variants, epileptiform and nonepileptiform abnormalities, adult and pediatric seizures, status epilepticus, ICU EEG, and sleep; in addition to ambulatory and video-EEG monitoring, electrocorticography, and magnetoencephalography. Essential “bottom-line” information in every chapter helps guide clinicians through the many challenges of EEG interpretation to improve patient outcomes. Practical tips from authors are included in a user-friendly manner. Designed for rapid retrieval and structured review, this handbook is a highly useful tool for neurology residents and fellows, clinicians, and technologists in search of reliable EEG information, regardless of specialty or level of training.

EEG INTER PR ETATION

Thoroughly updated and expanded Third Edition of the most trusted resource for anyone involved in EEG interpretation. Designed for on-the-go reference in the clinic or at the bedside, Handbook of EEG Interpretation concisely covers the fundamental components of EEG in clinical practice with graphic examples of classic EEG presentations and essential text throughout. Six new chapters have been added to address areas of growing importance with new dedicated chapters on technical aspects and artifacts of recording. With chapters written by prominent experts, this portable reference includes updated examples and color images new to this edition to reflect current advances in the field.

Handbook of

EEG

An Imprint of Springer Publishing

THIRD EDITION

Ha ndbook of

EEG INTER PR ETATION

Key Features:  hird edition of the comprehensive, easy-to-read, quick access handbook on T EEG interpretation n Updated to reflect advanced clinical EEG applications and techniques n Expanded coverage with the addition of six entirely new chapters n Provides a visual approach to identifying EEG waveforms and understanding the essence of their clinical significance with over 300 color tracings n Purchase includes digital access for use on most mobile devices or computers n

An Imprint of Springer Publishing

11 W. 42nd Street New York, NY 10036-8002 www.springerpub.com

Tatum

Recommended Shelving Category: Neurology

William O. Tatum, IV

Handbook of

EEG IN T ER PR ETAT ION

Handbook of

EEG IN T ER PR ETAT ION

Third Edition Editor William O. Tatum, IV, DO Professor of Neurology Mayo Clinic College of Medicine & Health Sciences Director, Comprehensive Epilepsy Center Mayo Clinic Jacksonville, Florida

An Imprint of Springer Publishing

Copyright © 2022 Springer Publishing Company, LLC Demos Medical Publishing is an imprint of Springer Publishing Company. All rights reserved. First Demos Medical Publishing edition 2008; subsequent editions 2014. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Springer Publishing Company, LLC, or authorization through payment of the appropriate fees to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400, fax 978-646-8600, [email protected] or on the Web at www.copyright.com. Springer Publishing Company, LLC 11 West 42nd Street, New York, NY 10036 www.springerpub.com connect.springerpub.com/ Acquisitions Editor: Beth Barry Compositor: diacriTech ISBN: 9780826147080 ebook ISBN: 9780826147097 DOI: 10.1891/9780826147097 21 22 23 24 25/ 5 4 3 2 1 Medicine is an ever-changing science. Research and clinical experience are continually expanding our knowledge, in particular our understanding of proper treatment and drug therapy. The authors, editors, and publisher have made every effort to ensure that all information in this book is in accordance with the state of knowledge at the time of production of the book. Nevertheless, the authors, editors, and publisher are not responsible for any errors or omissions or for any consequence from application of the information in this book and make no warranty, expressed or implied, with respect to the content of this publication. Every reader should examine carefully the package inserts accompanying each drug and should carefully check whether the dosage schedules therein or the contraindications stated by the manufacturer differ from the statements made in this book. Such examination is particularly important with drugs that are either rarely used or have been newly released on the market. Library of Congress Control Number: 2021903787 Contact [email protected] to receive discount rates on bulk purchases. Publisher’s Note: New and used products purchased from third-party sellers are not guaranteed for quality, authenticity, or access to any included digital components. Printed in the United States of America.

This edition of the Handbook is dedicated to my daughter Kelsey, who embodies strength, courage, and the ability to see people for who they are. Thank you for teaching me the passion and beauty of what life has to offer.

DKWILY

Contents Contributors  ix Preface  xiii 1. Technical Aspects of EEG  1 Jonathan J. Halford 2. Normal EEG  19 William O. Tatum, IV 3. Artifacts of Recording  49 William O. Tatum, IV 4. Abnormal EEG: Nonepileptiform  75 Selim R. Benbadis and Elson L. So 5. Abnormal EEG: Epileptiform  99 William O. Tatum, IV 6. Pediatric Seizures  123 Douglas R. Nordli, Jr. and Phillip L. Pearl 7. Ambulatory EEG  143 William O. Tatum, IV 8. Video-EEG and Adult Seizures  165 Michael Sperling and William O. Tatum, IV 9. Invasive Video-EEG  193 Stephan Schuele and Giridhar Kalamangalam 10. The EEG in Status Epilepticus  209 Frank W. Drislane and Peter Kaplan 11. ICU EEG  239 Nicolas Gaspard and Lawrence J. Hirsch

viii Contents

12. Polysomnography  279 Madeleine Grigg-Damberger and Steven A. Lopez 13. Neurophysiologic Intraoperative Monitoring  327 Aatif M. Husain 14. Electrocorticography  363 Marc R. Nuwer and Inna Keselman 15. Magnetoencephalography  377 John S. Ebersole and Michael E. Funke Index  401

Contributors Selim R. Benbadis, MD Professor of Neurology Director, Division of Epilepsy, EEG, and Sleep Medicine University of South Florida Tampa, Florida Frank W. Drislane, MD Professor of Neurology Comprehensive Epilepsy Center Beth Israel Deaconess Medical Center Harvard Medical School Boston, Massachusetts John S. Ebersole, MD Medical Director MEG Center Atlantic Health Neuroscience Institute Summit, New Jersey Michael E. Funke, MD, PhD Medical Director MEG Center McGovern Medical School University of Texas Health Science Center at Houston Houston, Texas Nicolas Gaspard, MD, PhD Associate Professor of Neurology Comprehensive Epilepsy Center Université Libre de Bruxelles Hôpital Erasme Bruxelles, Belgium Madeleine Grigg-Damberger, MD Professor of Neurology University of New Mexico Albuquerque, New Mexico Jonathan J. Halford, MD Professor of Neurology Medical University of South Carolina Ralph H. Johnson VA Medical Center Charleston, South Carolina

x Contributors

Lawrence J. Hirsch, MD Professor of Neurology Division Chief of Epilepsy and EEG Co-Director of Comprehensive Epilepsy Center Co-Director of Critical Care EEG Program Yale School of Medicine New Haven, Connecticut Aatif M. Husain, MD Professor of Neurology Chief, Division of Epilepsy, Sleep, and Neurophysiology Duke University Veterans Affairs Medical Centers Durham, North Carolina Giridhar Kalamangalam, MD, DPhil Wilder Family Professor of Neurology Chief, Division of Epilepsy University of Florida College of Medicine Gainesville, Florida Peter Kaplan, MBBS Professor of Neurology Johns Hopkins School of Medicine Baltimore, Maryland Inna Keselman, MD, PhD Assistant Professor Department of Neurology David Geffen School of Medicine at UCLA Los Angeles, California Steven A. Lopez, RPSGT Technical Supervisor University Hospital Sleep Disorders Center Department of Internal Medicine University of New Mexico Albuquerque, New Mexico Douglas R. Nordli, Jr., MD Professor of Pediatrics Chief, Section of Child Neurology Co-Director, Comprehensive Epilepsy Center University of Chicago Chicago, Illinois

Contributors  xi

Marc R. Nuwer, MD, PhD Distinguished Professor Department of Neurology David Geffen School of Medicine at UCLA; Department Head Department of Clinical Neurophysiology Ronald Reagan UCLA Medical Center Los Angeles, California Phillip L. Pearl, MD Professor of Neurology & William G. Lennox Chair Director, Epilepsy and Clinical Neurophysiology Boston Children’s Hospital Harvard Medical School Boston, Massachusetts Stephan Schuele, MD, MPH Professor of Neurology, Physical Medicine and Rehabilitation Chief, Section of Epilepsy Medical Director, Neurological Testing Center, Northwestern Memorial Hospital Feinberg School of Medicine Northwestern University Chicago, Illinois Elson L. So, MD Professor of Neurology Director, EEG Laboratory Department of Neurology Mayo Clinic Alix School of Medicine Rochester, Minnesota Michael Sperling, MD Baldwin Keyes Professor of Neurology Director of Jefferson Comprehensive Epilepsy Center Sidney Kimmel Medical College Thomas Jefferson University Philadelphia, Pennsylvania William O. Tatum, IV, DO Professor of Neurology Mayo Clinic College of Medicine & Health Sciences Director, Comprehensive Epilepsy Center Mayo Clinic Jacksonville, Florida

Preface This edition of the Handbook of EEG Interpretation promises to be the best iteration yet. Each chapter is written by a professor of Neurology subspecializing in the field of epilepsy. In conjunction with my colleagues at Demos Medical Publishing, I am delighted to expand on the strong response we have had in the past to the content of the handbook. We have written this book to fit into the lab coat pockets of clinicians in the hope of delivering a transportable reference to the bedside. As in any field of medicine, the best quality of care evolves and is reflected by new knowledge borne by the practitioner. In the case of electroencephalography, this knowledge is a function of exposure and of experience. Therefore, within the chapters outlined in this book, exposure to the functional uses of EEG is provided not as a sole representation, but rather to supplement the clinical experience with the essential "bottom line" information regarding information directed to helping clinicians with the challenges of EEG interpretation. While guidelines exist to identify waveforms, it is the “big picture” that determines normalcy and the boundaries of abnormality during final interpretation. Historically, EEG interpretation begins with didactic framework, though it is “on-the-job” training, usually in a one-on-one fashion, that has been the standard by which most beginning EEG-ers acquire competence and confidence by exposure to experts more senior in the field of clinical neurophysiology. Today, while much of the same methods continue in large university settings to educate neurologists and neurophysiologists with the means to interpret the EEG, the role of the Internet and classroom educational experiences are not capable of being retained “at the bedside” during encounters with “reallife” EEG recordings. Hence, the Handbook of EEG Interpretation hopes to fill a void that exists by providing quick and easy access to topics in EEG in the hopes of ultimately providing better patient care. Terminology throughout the book will adhere to that proposed by the International League Against Epilepsy (ILAE) in 2017. Understanding the technical aspects of EEG lies at the foundation of interpretation and this is skillfully presented by Dr. Halford. Correctly identifying abnormal EEG requires an understanding of Normal EEG, which is outlined to provide samples that include variations of normal and benign (normal) variants in addition to a dedicated chapter to outline Artifacts of Recording, given their role in challenging interpretation of the signal source. Abnormal Nonepileptiform EEG is reviewed by Drs. Benbadis and So. In addition to the chapter on Abnormal Epileptiform EEG, these are foundational chapters necessary to provide an accurate clinical correlation and facilitate appropriate patient management of patients with seizures and neurological illnesses that may have management decisions predicated upon interpretation of EEG. Pediatric Seizures are summarized in the excellent chapter by Drs. Nordli and Pearl. The role of Ambulatory EEG, in addition to Video-EEG and Adult Seizures, written by Dr. Sperling, are new chapters to emphasize the increasing role of event recording with EEG. Drs. Schuele and Kalamangalam update our knowledge on Invasive Video-EEG to reflect upon new techniques that distinguish intracranial from scalp EEG by recording directly from the brain. Status Epilepticus is reviewed by Drs. Drislane and Kaplan to provide expert electroclinical emphasis on persistent seizures in concert with the well-written chapter provided by Drs. Hirsch and Gaspard on ICU EEG to address electrographic interictal and ictal abnormalities in critically ill patients during encephalopathic states involving varying degrees of stupor or coma. The chapter on Polysomnography is updated by Drs. Grigg-Damberger and Lopez to provide a focus on and

xiv Preface

correlation of patients with sleep disorders and epilepsy. The chapter on Neurophysiologic Intraoperative Monitoring, by Dr. Husain, adds very instructive information to performing a variety of monitoring techniques in the operating room, in addition to a chapter on Electrocorticography, to expand on examples that involve patients with epilepsy, brain tumor, vascular malformations, and other diseases of the cerebrum. The excellent final chapter, Magnetoencephalography, written by Drs. Ebersole and Funke, includes examples to illustrate the powerful source localizing capability of this technique. It is with great humility and pride that I thank my amazing colleagues for their willingness, dedication, and time to promote the importance of EEG education by donating their expertise to this work. With completion of the third edition of the handbook, it is our hope that it will continue to serve clinicians young and old, experienced and novice, physician and technologist alike. To benefit the reader, quick and easy “pattern recognition” has been our format, though by the end of this book it is hoped that the reader will see EEG as much more within the vast and important field of clinical neurophysiology. Above all, realizing that clinical interpretation of EEG is an art must be recognized. We recognize the many excellent works that have contributed to advancing our knowledge of EEG and its interpretation. Realizing they are unable to be fully represented within a portable handbook, it is hoped that this “taste” of interpreting EEG will lead to organized, formal, scientific knowledge beyond simple interpretation of waveforms. The intent for the reader of our handbook is to provide a “bullet” of information with a graphic representation of the principal features in EEG, and hence provide a quick neurophysiology “go to” that is a time-efficient means of crosschecking a patient’s “brainwaves” in real-time. With the unique characteristics provided by EEG, we can only expect that as our knowledge base grows within the field of clinical neurophysiology, the expanding reach of EEG into other areas of medicine will grow and have more widespread implications in the future. William O. Tatum, IV, DO

1 Technical Aspects of EEG Jonathan J. Halford

I

nterpreting EEG requires both the ability to recognize patterns and an understanding of the underlying physics and engineering principles. Although some of the concepts underlying the technical background of EEG are complex, most concepts which are relevant to clinical practice can be expressed using only simple equations. Unfortunately, the number of concepts required to understand how EEG works increases every year. The main reason for this is the digitization of recording. We all live in an analog world, in which measured quantities can take on an infinite number of positive or negative values. Several decades ago, this abruptly changed with the addition of substantial digital components to EEG equipment, which represent measurements using a limited number of possible discrete quantities and require computers to be integrated in the device. Digital components have been added in the amplifier, storage, and visualization stages as well as in the recording electrodes themselves in some recent investigational systems. Digital signal analysis has also changed the interpretation of EEG signals, with digital improvements to filters, digital remontaging, and the addition of automated detection and artifact removal. Although digital systems force the recorded signal to become discrete (potentially leading to loss of relevant information) and are slower than analog systems, their use is spreading throughout devices because they allow greater control on signal information flow, offer lower power consumption, and are less susceptible to noise.

ANALOG RECORDING Basic Electrical Principles The movement of electric charge in currents within the nervous system leads to the EEG signal. Electric charges exist with either positive or negative polarity and are found in packages of a particular unit size called “quanta,” such as the positive charge of one proton or the negative charge of one electron. Protons are usually fixed within atoms and cannot move, but some electrons from each atom, especially in certain materials, are free to move between atoms and their movement creates electrical current flow. Charges also create electrical fields which cause them to act at a distance upon other charges, with like charges repelling each other and opposite charges attracting. Electric current flows best through materials called conductors, which are mostly metals or water solutions. When electric current flows through a conductor, it also creates a magnetic field around it. The measurement unit for charge (Q) is the coulomb, which has the charge of 6.24 × 1018 electrons. Flow of electric current (I) is measured in amperes, also called “amps,” and is equal to the flow of one Coulomb of charge per second. I=

dQ dt

( dQ is the change in Q; dt is the change in time)

Electric current does not flow very fast, with electrons moving only at an average speed of approximately 0.25 mm/sec, but the electric field spreads out within the conductor at the speed of light. This is similar to what happens in a pipe filled with water in which, when

2  Chapter 1: Technical Aspects of EEG

a small quantity of water is forced into one end of the pipe, the pressure increase within the water spreads very rapidly and a small quantity of water is immediately forced out of the other end of the pipe, despite the water itself in the pipe not having travelled far. Any changes to a magnetic field created by changes in the movement of electric charge also spread out from the conductor at the speed of light. Electrical devices work by causing current to flow around in a circle, called a circuit. The voltage (V) is the amount of potential energy between two points in that circuit. The unit measure of V is the volt, which is the energy, measured in joules, which will be delivered per coulomb of charge when that charge passes between two points in the circuit. A common analogy to understand electric current flow is water flow to and from a water tank. The volume of water is equivalent to charge (Q), the rate of flow is equivalent to the current (I), and the voltage (V) is equivalent to the difference in the pressure of water between two points in the flow system (such as between the top of the water tank and the pipe in your home). Like an electric circuit, water delivery in a city runs in a circle, with water flowing down from the tank to various destinations from which it is eventually recollected and pumped back into the tank. But unlike water which usually flows only one way through pipes, electronic circuits are often designed to create current flow which alternates in direction very rapidly. A direct current (DC) is current which flows only in one direction. Alternating current (AC) rapidly switches direction of flow continuously. The electrical resistance (R) is a measure of the difficulty of passing an electrical current through a conductor. Generally, the amount of current (I) is linearly proportional to the voltage and inversely linearly proportional to the resistance: I=

V R

(This can also be represented as V = I × R)

The unit of measure for R is ohms (Ω), which equals 1 volt per ampere of current. A resistor is an electrical circuit element added to create an impediment to current flow, like a sponge or filter placed in a pipe to slow the flow of water. A material with a high R, such as rubber, is called an insulator. Conductance (G) is defined as the opposite of resistance such that G = 1/R. The human body predominantly consists of saltwater, which acts as a passive volume conductor, a relatively low R medium for electric current through which an electric field can spread easily throughout. This is the reason why signals generated in one area of the body (such as the heart) can be measured in completely different areas of the body (such as on the head or foot). Two circuit elements serve to store electric charge, capacitors and inductors. Capacitors are devices used to store charge, consisting of one or more pairs of conductors (usually metal plates) separated by a thin insulator (Figure 1.1). A voltage source, such as a battery, pushes electrons (negative charge) onto one of the conductive plates, creating an electrical field which pushes electrons off the opposite plate, leaving positive charges behind. How much charge a capacitor can store, or its capacitance (C) is linearly proportional to the amount of charge stored in the capacitor (Q) and inversely linearly proportional to the voltage difference (V) between the two metal plate conductors. C=

Q V

The unit of measurement of capacitance is the farad (F), which equals the capacitance in which one coulomb of charge causes a potential difference across the capacitor of 1 volt. A capacitor will resist the flow of DC but will allow AC to “flow” through, although this AC flow is more of an illusion since charge just accumulates and leaves the two plates of the capacitor. This apparent “flow” of current across a capacitor is termed capacitive coupling.

Analog Recording  3

Electrons crowd on making this plate negatively charged e– e– e– e– e– e–

e– e– e– e– e– e–

Conductive plate

Insulator (dialectric)

Conductive plate + + + + + +

+ + + + + +

Electrons repelled from this plate, leaving positive charges behind, making this plate positively charged

FIGURE 1.1.  Diagram of a capacitor.

Inductors are electrical components that store energy in a magnetic field when electrical current is flowing through it. Typically an inductor consists of a coil of wire wrapped around a magnet (Figure 1.2). If the electric current flowing through the inductor decreases or stops, the magnetic field quickly collapses and induces a current in the coil which resists that decrease in current flow. In this way, an inductor resists changes in the current, therefore resisting AC but permitting DC. The strength of an inductor to perform this function is termed inductance (L) and its unit of measurement is a henry. In a circuit in which the current is decreasing at a constant rate of 1 ampere per second, an inductance of one henry results in the generation of one volt of potential difference across the inductor. This induced voltage, called the electromotive force, is directly proportional to the inductance (L) and directly proportional to the change in current (−dI). The change in current (−dI) is negative because the current is decreasing. V=L

− dI dt

( dI is the change in current; dt is the change in time)

Often there are multiple circuit elements which store charge in an electrical circuit, and these sum in different ways depending on whether they are in series or in parallel, as described in Table 1.1.

Magnetic field (B) I Current (I) in

Current (I) out

B (A)

(B)

FIGURE 1.2.  In (A), the right-hand rule illustrates the direction of the magnetic field if the electrical current is flowing upward. (B) Depicts the magnetic field generated by electrical current flow through a coiled conductor.

4  Chapter 1: Technical Aspects of EEG TABLE 1.1.  Rules for combining resistors, inductors, and capacitors in electronic circuits

Resistors and Inductors In Series Series In

Capacitors

R = R1 + R2 + R3

1 C

L = L1 + L 2 + L 3 In Parallel Parallel

1 R 1 L

=

=

1 R1 1 L1

+

+

1 R2 1 L2

+

1 C1

+

1 C1

+

1 C1

C = C1 + C2 + C3

1

+

=

R3 1 L3

The fundamental circuit for understanding the analog components of neurophysiology recording is the resistor–capacitor (RC) circuit (Figure 1.3). The current flow in this circuit varies over time when a voltage change occurs.

Analog Amplifiers The changes in the current in an AC electrical signal are sinusoidal, meaning they follow the shape of a sine wave. Many biological signals have a sinusoidal shape as well, and any time-varying signal can be broken down into a combination of sinusoidal waves of different frequencies and times of occurrence. One of the primary functions of ­neurophysiology instruments is to amplify a biological signal, which means to increase the

1.2 R

× 10 –4

1

V in C (A)

Vc

I (amperes)

0.8 VR

0.6 I( ) = 0.368 I max

0.4

I

0.2

= RC

0 0 (B)

0.5

1

1.5 time (sec)

2

2.5

3

FIGURE 1.3.  (A) A series resistor–capacitor (RC) circuit is shown with three circuit elements: the resistor (R), capacitor (C), and a power source (Vin). Three voltage potential changes are labeled: the voltage potential increase from a power source (Vin), the voltage potential decrease across a resistor (VR), and the voltage potential decrease across a capacitor (VC). The current (I) flows in a clockwise direction around the circuit (Inductiveload, 2017). (B) Plots the current flow as a capacitor charges. Since the time constant for the RC circuit (τ) is 0.5 seconds, current drops to the 1/e (0.368) fraction of the starting current by that timepoint.

Analog Recording  5

power of the voltage and/or current of the time-varying electrical signal. Neurophysiology amplifiers are predominantly voltage amplifiers and therefore the output of the amplifier is measured in volts. The gain of an amplifier (G) is the factor by which the amplifier increases the amplitude of the input signal. A voltage amplifier with a G of 2, for example, outputs a signal with double the voltage of the input signal. A simple amplifier with one channel has two input lines and two output lines. It amplifies the voltage potential difference between the two input lines (the input channel) to produce one output signal carried by two output lines (which are together called the output channel). These lines are usually wires and each input line is connected to an electrode on or in biologic tissue. Usually one input line is connected to a signal electrode, which is placed in a region of interest in the biologic tissue, and the other input line is connected to the reference electrode, which is placed in a location with relatively low noise. Biosignal recording equipment contain multichannel amplifiers made up of multiple single channel amplifiers, which permits recording from multiple signal electrodes. If a single reference electrode is connected to the second input of all of the amplifiers, this is called a referential recording setup. A ground electrode is also placed, usually on the forehead in EEG recording. The location of the referential electrode in EEG recording is usually chosen to be at the top of the head, at a distance from eye movement artifacts near the front of the head, muscle artifacts in the temporal regions, and the posterior dominant background rhythm in the posterior region. Figure 1.4 contains a diagram of the wiring of an amplifier system in a referential recording setup with two signal electrodes, one referential electrode and two amplifier output channels. This figure also shows the setup for bipolar recording, in which three electrodes are connected in a chain in which the middle electrode is connected to both amplifiers. Because there is so much noise in the environment of neurophysiologic recordings, a special type of amplifier, called a differential amplifier, is used in neurophysiology recording equipment. This type of amplifier is essentially a combination of three amplifiers wired together. In the first stage of a differential amplifier, two amplifiers working in parallel amplify (a) the potential difference between the signal electrode and the patient ground electrode and (b) the potential difference between the reference electrode and the patient ground electrode. In the second stage, a third amplifier amplifies the difference between the outputs of the two first-stage amplifiers. A differential amplifier can be used for both a referential and a bipolar recording setup, but Figure 1.5 depicts only the referential setup. Environment noise usually causes a voltage potential difference between the recording electrodes and the ground electrode, which is called the common mode signal, and this is of relatively similar magnitude among the recording channels. The differential signals are the voltage potential differences between each signal electrode and its reference electrode, which are the signals of interest. Differential amplifiers work to cancel out much of the common mode signal, a process is called common mode rejection, and to amplify the differential signals. The degree to which a differential amplifier rejects the common mode signal is called the common mode rejection ratio (CMRR). This is calculated using the amplification of the common mode signal, called the common mode gain (Gcm) versus the amplification of the referential signal, called the referential gain (Gd). This is usually expressed in the unit of decibels (dB), which expresses the ratio of two numbers (usually signal magnitude) on a logarithmic scale multiplied by a factor of 20. G  CMMR = 20 log 10  d   G cm  An amplifier which reduces the common mode signal by a factor of 10,000 (104) relative to the differential signals has a CMRR of 80 dB. Most modern biosignal amplifiers

6  Chapter 1: Technical Aspects of EEG

Referential recording

Electrode 1 Differential electrode

Differential Amp 1

Output 1

Differential Amp 2

Output 2

Differential Amp 1

Output 1

Differential Amp 2

Output 2

Electrode 2

Bipolar recording

Electrode 1

Electrode 2

Electrode 3

FIGURE 1.4.  Referential and bipolar recordings setups.

have a CMRR of over 100 dB. The entire bioamplifier system (consisting of multiple channel amplifiers) has a total input impedance (I in) which can be calculated using a set of equations (not given here) which incorporate the total resistance, capacitance, and inductance of the amplifier as well as the frequency content of the AC signal. The Iin is designed to be as high as possible, in order to improve the performance of the amplifier. Most modern amplifiers have an Iin of 100 MΩ (100 megaohms) or more. But a high Iim is not sufficient to assure a high-quality recording. Each electrode has an impedance value as well, which is primarily due to skin resistance and electrode impedance (both capacitive and resistive). If there is a difference between the impedance of the electrodes, this can cause leakage of common mode signal into the output of the differential amplifiers. For this reason, it is not just important to keep the impedance of each electrode as low as possible, but also important to keep the impedance of each electrode within a similar range (such that a few electrodes don’t have significantly higher impedance than the other electrodes).

Analog Recording  7

Differential amplifier 1 Diffe Biologic tissue

Amp A1

Electrode 1 Circuit ground Patient ground electrode

Amp C1

Channel 1 output

Amp B1

Reference electrode Differential amplifier 2 Amp A2 Electrode 2

Circuit ground

Amp C2

Channel 2 output

Amp B2

FIGURE 1.5.  Differential amplifier diagram for two recording electrodes in a referential recording setup.

Electrodes Recording electrodes provide an interface point between biologic tissues and metal recording electrodes. Within biologic tissues, electric current flows in a low-resistance saltwater solution by the movement of positively and negatively changed ions. The metal of electrodes and the wires that connect them to amplifiers also provide a low-resistance media in which current can flow. But the boundary between these two media is difficult to bridge electrically because of multiple sources of impedance to current flow which must be overcome. An electrolyte solution or gel is placed between the skin and the electrode to help improve conduction. The first source of impedance is the outer layer of the skin, the epidermis, which contains keratin and is covered in natural oils, both of which are good insulators. Skin preparation by gentle abrasion removes some of this keratin by removing dead skin cells and the natural oils that cover the epidermis. At the boundary between the electrolyte solution and the metal electrode, current can pass back and forth either through the movement of electrons due to chemical reactions between the metal and the electrolyte (called redox reactions) or through capacitive coupling. Movement of charge through capacitive coupling is preferred, because redox reactions at the electrolyte solution/electrode boundary can produce toxic chemicals which can damage tissues. Capacitive coupling can occur because a charge imbalance develops between the electrolyte solution and the metal electrode, either from redox chemical reactions or the exchange of electrons between ions in solution and the metal. This charge imbalance, which acts as a capacitor, is called the half-cell potential.

8  Chapter 1: Technical Aspects of EEG

An ideal electrode develops only a small half-cell potential because if this potential is too large, it leads to problems due to its instability and capacitance. This half-cell potential can break down abruptly for uncertain reasons, leading to discharges of current which causes “electrode pop” artifacts, or can be altered unpredictably by the influx of sweat into the electrolyte. Because capacitance acts to impede the flow of DC current, the half-cell potential also decreases the transmission of low-frequency signals. Silver– sliver chloride electrodes, which have silver salt intermixed with the silver metal within the electrode, produce the lowest half-cell potentials and make the best scalp EEG electrodes, particularly if DC potentials need to be recorded. Because electrical charges flow directly between the silver–silver chloride electrode and the electrolyte gel, not much of a charge imbalance develops between the electrode and the biological media and there is not polarization of charge between the electrode and the conductive media. For this reason, silver–silver chloride electrodes are termed nonpolarizing electrodes. Nonpolarizing electrodes are good for recording both high frequencies and DC potentials (very low frequencies). Gold and platinum electrodes are low-resistance conductors and do not cause chemical reactions with the electrolyte. Electrical charges do not move between these metals and the electrolyte media because these metals are inert. Electrical current flows between the metals and biologic media through capacitive coupling due to the large half-cell potential which develops between the electrode and the biologic media. This charge imbalance produces capacitance and therefore capacitive impedance, which impedes the transmission of low-frequency signals. Gold, platinum, and stainless steel electrodes are termed polarizing electrodes. The larger half-cell potentials of polarizing electrodes make them more susceptible to electrode pop and movement artifacts, due to rapid unpredictable release of charge from the electrode. Gold electrodes (which are actually gold plated over silver) are considered acceptable for recordings, although their recording properties are not as good as silver–silver chloride electrodes, but they are often used because they are cheaper to fabricate.

Analog Filters An analog signal filter is a device that removes some unwanted components in an electrical signal, usually certain frequencies. Analog filters are built using a combination of resistors and capacitors. The design of a simple analog filter is based on the concept of the voltage divider (Figure 1.6A), which is a circuit consisting of two resistors (R1 and R2) which produces an output voltage (Vout) which is a fraction of its input voltage (Vin).  R  1 Vout = Vin    R1 + R 2  A highpass filter, which allows high frequencies to pass on through (and blocks lower frequencies), is a circuit with a combination of a capacitor and resistor similar to a voltage divider, as illustrated in Figure 1.6B. In this design, all current flow must pass through the capacitor, which will only pass current through capacitive coupling, and therefore preferentially passes AC current of higher frequencies better than that of lower frequencies. The resistor, which does not impede lower frequencies, lets a certain amount of the lower frequency content of the signal leak through and therefore not get passed on to the output of the circuit. A lowpass filter is designed with the resistor and capacitor in opposite positions in the circuit (Figure 1.6C). Since the resistor does not alter the transmission of current based on its frequency content, the low pass filter allows both the low and high frequency components of the signal to pass through. But then the higher frequency components of the signal are allowed to leak out through

Analog Recording  9

Voltage divider

Vin

R2

Vout

R1

(A)

Highpass (low frequency) filter

Vin

Vout

(B)

Lowpass (high frequency) filter

Vin

Vout FIGURE 1.6.  Voltage divider and simple highpass and lowpass analog filters.

(C)

the capacitor (which preferentially passes higher frequencies) and therefore they do not appear as much in the output of the circuit. Filters are defined mainly by three design characteristics—filter type, cutoff frequency (or frequencies), and filter order. There are four filter types, as illustrated by Figure 1.7. The first two filter types (highpass and lowpass) have already been discussed. Figure 1.7 illustrates the frequency response of the four types of filters. Both highpass and lowpass filters have one cutoff frequency, which is defined as the frequency at which the voltage output of the filter is −3 dB (or 0.707) of the voltage of the input signal. The cutoff frequency can be calculated from the time constant of the filter (τ). As described above in the discussion of RC circuits, τ = RC. f cutoff =

1 1 = 2 πRC 2 πτ

A bandpass filter, as depicted in Figure 1.7 C, is a combination of a lowpass and a highpass filter which has two cutoff frequencies, one for each filter. Frequencies between the two cutoffs are allowed to pass. A bandstop (or notch) filter is designed to block frequencies in a very narrow range and is usually used to remove 50-Hz or 60-Hz power line noise. Filters can also change the timing of when certain features appear in an input signal. This is called the phase-response of a filter. Generally, all analog filters change the appearance of the signal so as to make waveforms in the input signal appear to occur slightly earlier or slightly later than in the original signal, and this is termed phase advance and phase delay, respectively. Whether a filter causes phase advance or delay depends on the frequency of the waves in the signal and on whether the filter is a lowpass or highpass filter. Generally, this has become less of a problem in modern equipment because, unlike for analog filters, most modern digital filters do not induce much change in the phase of the signal for signal frequency components which are within the passband of the filters (if the digital filters are applied to the signal data after it is recorded).

10  Chapter 1: Technical Aspects of EEG Highpass filter

0

0

–10

–10

–20

–20

–30 –40 –50

–30 –40 –50

–60

–60

–70

–70

–80

0

20

40

(A)

60

–80

80 100 120 140 160

Frequency (Hz)

Lowpass filter

10

Magnitude (dB)

Magnitude (dB)

10

0

2 4 6

(B)

Frequency (Hz)

0

0

–10

–10

–20

–20

–40 –50

–30 –40 –50

–60

–60

–70

–70

–80 (C)

Bandstop (notch) filter

Magnitude (dB)

Magnitude (dB)

Bandpass filter

–30

0

20 40 60 80 100 120 140 160 180 Frequency (Hz)

8 10 12 14 16 18 20 22 24

–80 (D)

0

20

40 60 80 Frequency (Hz)

100

120

FIGURE 1.7.  (A) Highpass filter with cutoff frequency of 100 Hz. (B) Lowpass filter of cutoff frequency of 1 Hz. (C) Bandpass filter of cutoff frequencies of 40 Hz and 100 Hz (highpass and lowpass, respectively). (D) 60-Hz bandstop (notch) filter. All filters are Butterworth third order except for the bandstop filter, which is 10th order.

Grounding In electrical engineering, the ground is the reference point in an electrical circuit from which voltages are measured and a common path for electric current. Connecting a circuit to earth ground means physically connecting to the earth through conductors. Patients connected to a modern biosignal amplifier system are usually not connected to earth ground but connected to the amplifier common ground. Amplification is performed in two stages. The first stage involves differential amplification within the preamplifier, which is connected to the amplifier common ground. The second stage of amplification is performed by a more powerful amplifier which is connected to earth ground. This second amplifier includes an isolator which transmits the signal through a nonelectrical mode of signal transmission,

Analog Recording  11

Main power line

Preamplifier

+

Signal input

–

Isolation amplifier

+ –

SSignal o output

Amplifier common ground

Earth ground

FIGURE 1.8.  Connection between the main power line, the patient, amplifiers, and two types of ground.

such as light waves (called optical isolation). This isolates the patient from the more powerful second amplifier and earth ground. This protects the patient from exposure to accidental electrical currents in the environment which are not from the neurophysiologic recording equipment (and which want to flow to earth ground) and isolates the delicate preamplifier stage from connection to the more powerful isolation amplifier equipment, decreasing noise. Figure 1.8 is a diagram of a typical connection between one channel of a biosignal amplifier and the two types of ground. Because the body of the recording subject, power lines, lines connecting the electrodes to amplifiers, and ground lines all contain electrical charge, capacitive coupling can develop between them all, even if there is not a physical connection through which electrical charge can flow. This creates multiple potential circuit loops through which current can flow due to magnetic induction generated by any magnetic fields in the environment. These circuit loops are called ground loops. Note in Figure 1.8 how many potential ground loops there are just for one biosignal recording channel. The magnetic flux noise in the environment which creates the current in these ground loops is usually caused by other electrical equipment nearby.

12  Chapter 1: Technical Aspects of EEG

DIGITAL RECORDING Basic Digital Recording Principles Digital electronic systems represent the amplitude of time-based signals (such as EEG) in discrete bands, as opposed to the continuous ranges used in analog electronics. As depicted in Figure 1.9, the time-based signal is sampled at discrete points which are arranged to occur at regular intervals in time. Additionally, the amplitude of the signal (usually in volts) is discretized at certain specified levels within a minimum and maximum range. The frequency at which the system measures the amplitude of the input signal is called the sampling rate, which is expressed as the number of measurements per second in hertz (Hz). EEG is typically recording with a sampling rate between 256 Hz and 2,048 Hz (2 kHz). Since digital computers work using the base two number system, technical specifications for the features in digital systems, such as sampling rate, are often numbers of base two such as 28 = 256 or 211 = 2,048. The Nyquist Theorem states that in order to adequately reproduce a signal it should be periodically sampled at a rate that is at least two times the highest frequency you wish to record. In practice, the sampling rate must be at least 5 times the highest frequency of interest for faithful analog to digital (A-D) conversion of the signal. The minimum number of points at which the signal amplitude is discretized is termed the bit depth. Figure 1.10 depicts two approximately 1 Hz sine waves which have been sampled with two different bit depths. One sine wave has been sampled at a bit depth of 210 (1,024 discrete levels of amplitude) and the other sine wave has been sampled at a bit depth of 25 (32 discrete levels of amplitude). For the sine wave sampled at a bit depth of 5, it is clear that the bit depth is not sufficient to fully render the shape of the wave, and the wave appears “jagged” with abrupt transitions from discrete levels of amplitude clearly visible. A bit depth of 10 is sufficient to render most EEG recordings, although most modern EEG amplifiers have a bit depth of 16 or higher. A much higher bit depth than 10 is helpful because biosignal amplifiers must record not only the signals from the nervous system but also noise and artifactual signals which commonly have a much higher amplitude than brain signals.

Signal Digitization The basic stages of biosignal recording are depicted in Figure 1.11. The signal from the electrodes is initially amplified to a small degree by the preamplifier, using a combination of amplifiers in the differential amplification process as described above (Epstein, 2011). The purpose of this preamplifier stage is (a) to use a very high amplifier

f(t)

t

FIGURE 1.9.  An analog signal (in red) and a discrete digital representation of that signal. The horizontal axis is time (t) and the vertical axis is the quantity of the signal, f(t).

Digital Recording  13

10 bit

5 bit

1 0.8 0.6

Amplitude

0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1 0

0.5

1

1.5 Time (sec)

2

2.5

FIGURE 1.10.  Sine waves of approximately 1-Hz frequency rendered with two sampling rates. The blue sine wave is rendered at 3 a bit depth of 10 and the red sine at bit depth 5.

impedance (so as to accurately measure the signal), (b) to remove the common mode signal, and (c) to boost the amplitude of the signal to provide sufficient input amplitude to the next amplifier stage. Next, (either before or after the next amplification stage, depending on the system), analog filters are applied. A highpass analog filter with a low cutoff frequency is used to remove the very low frequency upward or downward voltage drift of each channel (called the “DC shift”). An analog lowpass filter with a high cutoff near the Nyquist frequency is used to remove noise and to improve the accuracy of A-D conversion. The second amplifier stage brings the signal voltage up to the level needed by the A-D converter (usually in the −5 V to +5 V range) and uses built-in signal isolation. Within this amplifier is a complete (isolation) barrier to electrical current flow across which the signal is carried by light (optical isolation) or using a transformer to transmit the signal using changing magnetic flux (transformer isolation). The output of

Jackbox Preamp (differential amplifier) (gain: 10–50)

Analog to digital (A-D) converter

Analog filters

Isolation amplifier (Gain:10–1000)

Digital storage Digital signal

Biological tissue Digital visualization

FIGURE 1.11.  The basic components of a digital biosignal amplifier in a recording jackbox (for one recording channel).

14  Chapter 1: Technical Aspects of EEG

the isolation amplifier is directed to the A-D converter where the analog electrical signal is discretized at a specified sampling rate and set of amplitudes and then transmitted out as digital information. The digital information is stored in a computer in memory and/or in a hard drive. From there, the digital signal information is visualized using computer software and a monitor display.

Digital Filtering Once a digital signal is stored, it can be processed mathematically by a computer. Digital filtering is the process of using mathematical algorithms to apply filters to remove certain frequency components. This process acts in a similar way to analog filtering but unlike analog filters, which can process incoming signals quickly in real time, digital filters are usually only applied to the signal after the recording is completed. Digital filters have some advantages over analog filters. First, a well-designed digital filter has a very “flat” response in the passband of the filter, meaning that digital filters do not affect some frequency components more than others (as much as analog filters) in the range of frequencies that are intended to be passed through the filter without attenuation. Second, digital filters can be designed to not have any effect on the phase of the signals in the frequency ranges of the passband (low phase distortion). Analog filters, on the other hand, often cause slight phase delay or phase advance in the signal, even for those signal frequency components that are intended to pass through filter without modification. Third, digital filters can adapt their characteristics to match characteristics of the input signal, whereas analog filters are not as flexible.

Digital Display EEG is displayed using modern neurophysiology equipment on computer monitors, which depict information as individual points of light, called pixels. Characteristics of computer monitors include luminance (brightness), aspect ratio (ratio of the height to width of the viewable screen), and display resolution (measured based on number of pixels along the height axis versus the number of pixels on the width axis). Equipment manufacturers agree to standard display specifications, which are listed in Table 1.2. The total number of pixels on a screen is calculated based on the number of pixels of width versus the number of pixels of height. Note that the total number of pixels on a screen (and therefore the amount of memory required to store the information in the video) is increasing exponentially as video format specification formats are advancing. In a digital review system, the reviewer sets the height at which a unit of voltage is rendered on the screen, called the vertical scaling or digital display gain. A standard setting for the digital display gain is 7 microvolts per mm of computer monitor screen. Additionally, the reviewer can set the time base for review, which is the number of seconds of neurophysiology signal which is viewed on the monitor, which defines the horizontal time scaling. A sufficient screen resolution is necessary to render a neurophysiology waveform clearly, as illustrated in Figure 1.12. Inadequate screen resolution can limit rendering of neurophysiology waveforms if the time base and/or gain are set insufficiently low by the reviewer. Note in Figure 1.12 that the screen resolution is insufficient to clearly render these waveforms, particularly for the higher frequency waves on the right.

Aliasing During the process of A-D conversion, and also during the process of signal display on a computer monitor, a time-based signal is forced to take certain discrete values, as explained in the process of discretization above. Aliasing can occur when the sampling rate for this discretization is insufficient to represent the frequencies of the analog signal or

Digital Recording  15 TABLE 1.2.  Computer monitor screen resolutions, aspect ratios, and total number of pixels per frame

Video Standard

Width

x

Height

Aspect Ratio

Pixels

8K UHD

7,680

×

4,320

16:9

33,177,600

3,840

×

2,160

16:9

8,294,400

2,560

×

1,440

16:9

3,686,400

1,920

×

1,080

16:9

2,073,600

1,440

×

1,080

4:3

1,555,200

1,280

×

720

16:9

921,600

“High” VGA

640

×

480

4:3

307,200

QVGA

320

×

240

4:3

76,800

QQVGA

160

×

120

4:3

19,200

(4,320p) 4K HD (2,160p) FHD+ (WQHD) (1,440p) FHD (FullHD) (1,080i, 1,080p), HDV (1,080i) WXGA-H (720p)

FIGURE 1.12.  Inadequate rending of waveforms in pixels because of inadequate screen resolution.

digital signal being converted or displayed, respectively. The Nyquist Theorem states that the highest frequency that can be represented in a digital time-based signal is half of the sampling rate. Of course, sampling at twice the frequency of the original analog waveform is not nearly enough to guarantee an accurate representation of the waveform (as illustrated in the right side of Figure 1.12), and usually a minimum of five sampling points per cycle of a waveform is suggested. So, for example, the minimum adequate sampling frequency to record a 10 Hz signal would be approximately 50 Hz.

16  Chapter 1: Technical Aspects of EEG

Signals of frequency close to or higher than the Nyquist frequency which are present in the signal before digitization can appear in the signal after digitization at lower frequencies and this is called aliasing. This artifactual aliased signal (fa) appears at a frequency lower than the Nyquist frequency (f N) based on what magnitude of the original signal frequency is (fo), as described in the equation below. fa = fo – fN The problem of aliasing in A-D conversion can be minimized by applying an analog lowpasss filter (called an anti-aliasing filter) with a cutoff near the Nyquist frequency before the signal undergoes digitization, as portrayed in Figure 1.13. This decreases the amplitude of the frequencies which are most likely to be aliased before the A-D conversion process. Aliasing of a digital signal can also occur during the process of display on a computer monitor. The resolution of the computer monitor used to display a time-based signal

2 Hz sampled at 20 Hz 18 Hz sampled at 20 Hz

2 Hz sampled at 1000 Hz 18 Hz sampled at 1000 Hz

1 0.8 0.6 0.4

Amplitude

0.2 0 –0.2 –0.4 –0.6 –0.8 –1 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Time (sec) FIGURE 1.13.  Aliasing of an 18-Hz signal which is acquired at a 20-Hz sampling rate to a frequency of 2 Hz. The red sinusoidal wave is the 18-Hz signal sampled at 1,000 Hz. The blue signal is the 2-Hz signal sampled at 1,000 Hz. The black circles represent the 2-Hz sinusoidal wave created by aliasing because the sampling frequency (20 Hz) is very close to the frequency of the signal (18 Hz).

1

Digital Recording  17

imposes a sampling frequency on the signal to be displayed. This sampling frequency is best understood as the maximum frequency that a monitor can display. This maximum display frequency (fd) can be calculated from the number of horizontal pixels dedicated the display (h) and the number of seconds (s) of time-based signal displayed: fd =

1 h ( ) 2 s

Signals of frequency near this maximum display frequency are not rendered well in many cases and aliasing occurs. This aliasing is prevented by a digital antialiasing filter which is built into the neurophysiology review software. This digital filter can be quite complex and in some systems performs a type of “intelligent down-sampling” such that when the frequency of the signal to be rendered exceeds the maximum display frequency, the system uses an algorithm to render the waveform such that certain frequencies that are above the maximum display frequency are rendered at precisely the maximum display frequency. The use of modern high resolution widescreen monitors can minimize the aliasing problems caused by a suboptimal monitor maximum display frequency. Monitors with Full HD Plus (FHD+ or WQHD) resolution at 2,560 × 1,440 or higher resolutions offer many more pixels than traditional monitors and are now inexpensive enough for most labs.

CONCLUSION One of the main reasons to learn about the technical background to EEG recording is to be able to understand and identify artifacts. As digital equipment becomes more complex, equipment in the clinical recording environment can create artifactual signals which are quite complex and easily mistaken for biological signals. An EEG interpreter’s ability to recognize artifacts is enhanced by a firm understanding of the technical background of neurophysiology data acquisition. Since new technologies and artifacts are constantly emerging, the ability to recognize the difference between a biological signal and a signal generated by a machine is also important. In probability theory, a stochastic (or random) process is a collection of random variables used to represent the evolution of a value over time. It is not possible to predict how a stochastic process will evolve over time if you know the starting conditions. A deterministic process is a process in which, given the starting point, you can know with certainty how the process will evolve over time. Biological signals have a strong stochastic component, which is due partially to noise but also due to complexity of the system generating the signal. Artifactual signals generated by equipment in the environment often have the appearance of deterministic signals, which appear in a mathematically predictable pattern. Understanding this basic difference in types of signals is important to identifying many types of artifacts.

ADDITIONAL RESOURCES Epstein CM. Analog signal recording principles. In: Schomer DL, Da Silva FL, eds. Niedermeyer’s Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Lippincott Williams & Wilkins; 2011:111–118. Inductiveload. Own work, Public Domain. https://commons.wikimedia.org/w/index.php?curid=5801614. 2017. Maus D, Litt B. Engineering principles. In: Ebersole JS, Husain AM, Nordli DR, eds. Current Practice of Clinical Electroencephalography. Lippincott Williams & Wilkins; 2014:45–77.

18  Chapter 1: Technical Aspects of EEG Metting van Rijn A, Peper A, Grimbergen C. High-quality recording of bioelectric events. Med Biol Eng Comput. 1990;28(5):389–397. https://doi.org/10.1007/BF02441961 Supek S, Aine CJ. Novel Noise Reduction Methods. Magnetoencephalography: From Signals to Dynamic Cortical Networks. Springer; 2014:35–68. White RL, Gross TJ. An evaluation of the resistance to electrolysis of metals for use in biostimulation microprobes. IEEE Trans. Biomed. Eng. 1974(6):487–490. https://doi.org/10.1109/TBME.1974.324339

2 Normal EEG William O. Tatum, IV

T

he EEG is a unique and valuable measure of the brain’s electrical function. It is a graphic display of a difference in voltages from two sites of brain function recorded over time. EEG involves the study of recording these electrical signals that are generated by the brain. Extracranial EEG provides a broad survey of the electrocerebral activity throughout both hemispheres of the brain. Intracranial EEG provides focused EEG recording directly from the brain from surgically implanted electrodes that are targeted at specific regions of the brain. Information about a diffuse or focal cerebral dysfunction, the presence of interictal epileptiform discharges (IEDs), or patterns of special significance may be revealed. For the successful interpretation of an abnormal EEG, one must first understand the criteria necessary to define normal patterns. While a normal EEG does not exclude a clinical diagnosis of epilepsy, an abnormal finding on EEG may be supportive of the clinical diagnosis. Furthermore, it may be indicative of cerebral dysfunction by demonstrating focal or generalized slowing. The results of a standard EEG may have nothing to do with the reason that the study was performed (i.e., patients with headache) and the results need to be taken in context with the individual reason for referral. It is the clinical correlation applied to the findings of the EEG that imparts its utility. Electrical signals are created when electrical charges move within the central nervous system. Neural function is normally maintained by ionic gradients established by neuronal membranes. Sufficient duration and length of small amplitude potentials (measured in microvolts) of electrical currents generated by cerebral activity are required to be amplified and displayed for visual interpretation. Normally, a resting (diffusion) membrane potential exists through the efflux of positive-charged (potassium) ions maintaining electro-chemical equilibrium at −75 mV. With depolarization, an influx of positive-charged (sodium) ions exceeding the normal electrochemical resting potential occurs. Ion channels open within the lipid bilayer via a voltage-dependent mechanism, and closure is timedependent. Conducting the change in cellular depolarization to adjacent portions of the nerve cell membranes results in an action potential when threshold is exceeded. However, it is the synaptic potentials as opposed to the action potentials that are the most important source of the extracellular current flow to produce the EEG. Excitatory postsynaptic potentials (EPP) flow inwardly (extracellular to intracellular) to other parts of the cell (sinks), via sodium or calcium ions. Inhibitory postsynaptic potentials (IPP) flow outwardly (intracellular to extracellular) in the opposite direction (source) and involve chloride or potassium ions. These summed potentials are longer in duration than action potentials and are responsible for most of the EEG waveforms. The brainstem and thalamus serve as subcortical generators to synchronize populations of neocortical neurons in both normal (i.e., sleep elements) and in abnormal situations (i.e., generalized spike-and-wave complexes). Volume conduction characterizes the process of current flow from the brain generator and recording electrode. Layers of cortical neurons are the main source of the EEG. Pyramidal cells are the major contributor of the synaptic potentials that make up EEG (Figure 2.1A). These neurons are arranged in a perpendicular orientation to the cortical surface from layers III, IV, and VI.

20  Chapter 2: Normal EEG

Volumes large enough to permit measurement at the surface of the scalp require areas that are >6 cm2 though >10 cm2 for recording most IEDs on the scalp using the International 10–20 system of electrode placement due to the attenuating properties incurred by the scalp and skull. All generators have both a positive and negative pole and function as a dipole (see Figure 2.1B). The EEG displays the continuously changing voltage fields at different locations on the scalp.

(A)(A) (A)

S

(B) (B)(B)

FIGURE 2.1.  (A) A pyramidal cell with excitatory postsynaptic potentials and inhibitory postsynaptic potentials. (B) Dipole depicting a field of charge separation.

tandard EEG recording displays the difference in electrical potentials between two different sites on the scalp overlying cerebral cortex that is closest to the recording electrode (Figure 2.1). During routine use, electrical potentials are acquired indirectly from the scalp surface, and incorporate waveform analyses of frequency, voltage, morphology, and topography. However, most of the human cortex is buried deep beneath the scalp surface. In addition, EEG represents a 2-dimensional projection of a 3-dimensional source. This presents challenges to solve the inverse problem of source localization using scalp EEG. Furthermore, the waveforms that are recorded from the scalp represent pooled synchronous activity from large populations of neurons that create cortical potentials and therefore may not reflect small interictal or ictal generators. Initial one-channel EEG recordings in the late 1920s have evolved to sophisticated digital-based computerized recording devices. From the patient scalp, electrodes conduct electrical potentials to an electrode box (jackbox). Thereafter, the montage selector permits EEG signals to pass through amplifiers before filtering and ancillary controls regulate the signal output. Data display follows acquisition and processing and has a wide variety of data presentation for EEG interpretation. Electrode placement has been standardized by an international system that uses anatomic landmarks on the skull. These sites are then subdivided by intervals of 10% and 20% to designate the site where an electrode will be placed. A minimum of 21 electrodes are commonly used for clinical studies, though digital EEG now has the capability for a greater number and 25 electrodes to accent basal temporal structures are recommended. During infant EEG recordings, fewer electrodes are often used and vary depending upon age and head size. A newer modified combinatorial electrode system uses electrode placement with more closely spaced electrodes in a 10–10 system (see Figure 2.2). The designations, Fp (frontopolar), F (frontal), T (temporal), O (occipital), C (central), and P (parietal) are utilized in the 10–20 system. Subsequently, numbers combined following the letters for location reflect either the left (odd numbers) or right (even numbers) hemisphere

Normal EEG

Normal EEG  21

of electrode placement. The “z” designation reflects midline placement (i.e., Cz = central midline). In the 10–10 system, lower numbers in their positions reflect locations closer to the midline, and T3/T4 become T7/T8 while T5/T6 become P7/P8. Electrode impedances should be maintained between 100 and 5,000 ohms. Special electrodes may also be added, such as sphenoidal, true temporal, or fronto-temporal electrodes. Most are employed for the purpose of delineating temporal localization. True temporal electrodes (designated T1/T2 and FT9/FT10) are placed to help distinguish anterior temporal or posterior inferior frontal location to enhance recovery compared with F7 and F8 electrode positions. Combining the 10–20 system with electrodes from the 10–10 system may be most practical for routine clinical use when additional electrodes are desirable. Collodion is a compound used to secure electrodes during prolonged recording techniques, such as during video-EEG, continuous EEG in the intensive care unit, or ambulatory EEG monitoring. Paste used for routine recordings is more temporary. Subdermal electrodes are used when other recording techniques are not feasible, such as in the OR and ICU.

FIGURE 2.2.  Electrode placement systems use either a 10–20 system (blue) or modified combinatorial system with 10–10 electrode placement (blue + gold).

O

ther additional electrodes may include EKG (recommended with every EEG), eye movement monitors, EMG, and extracerebral electrodes to aid in artifact differentiation (i.e., tremor monitor) and sleep staging. Respiratory monitors may also be important if respiratory problems are identified or sleep monitoring is performed.

22  Chapter 2: Normal EEG

(A)

FIGURE 2.3.  (A) Bipolar montage demonstrating phase reversal. (B) Referential montage demonstrating absolute voltage.

(B) (B)

T

he electrical “map” obtained from the spatial array of recording electrodes used to interpret EEG is the montage (Figure 2.3). Several montages are used throughout a 20- to 30-minute EEG recording. Every standard EEG should include at least one montage using an anterior-posterior longitudinal bipolar, reference, and traverse bipolar montage ­(Figure 2.4). A reference montage uses an active electrode site as the initial input, and then at least one “neutral” electrode to depict absolute voltage. The greatest amplitudes measured are commensurate with the site of maximal electro-negativity or positivity. One midline reference electrode (i.e., Pz), may be useful for a lateralizing temporal recording. However, two references (i.e., ipsilateral ear reference) may be useful for more generalized discharges. Even multiple “averaged” sites of reference (or Laplacian montages for very focal recordings) may be useful for localized discharges. Bipolar montages may be arranged in many different spatial formats, including longitudinally, transverse fashion, or in a circumferential pattern. The longitudinal bipolar (aka “double banana”) is the one that is most frequently represented in clinical practice and throughout this text. An anterior to posterior temporal and central transverse connecting chain of electrodes arranged left alternating to rightsided placement is an example of a targeted array to lateralize and cross-localize temporal abnormalities. Bipolar montages compare two active electrodes sites adjacent to each other and signify absolute electrographic sites of maximal negativity (or positivity) by phase-reversals.

(A)

(B) (B)

FIGURE 2.4.  EEG demonstrating bipolar (A) and reference (B) montages to illustrate a left anterior temporal sharp wave at the F7 electrode (left) and T1 derivation (right).

Normal EEG  23

EEG

Electrode 1

Electrode 2

Negative

Up

Down

Positive

Down

Up

FIGURE 2.5.  The rules governing polarity and convention are relative to deflection of the waveform. By convention, when input 1 is negative, the deflection is up. Polarity of the EEG thus has two reasons for an upward or downward deflection.

B

y convention, when the voltage difference between electrode 1 is more negative than electrode 2, deflection of the waveform is up (Figure 2.5). Recordings are usually performed with a visual display speed of 30 mm/sec (slower with sleep studies). Amplifier sensitivities of 7 μV/mm and filter settings of 1 to 70 Hz are routine parameters during standard EEG. Reducing the low filter settings promotes slower frequency representation, while reducing high filter settings attenuates high frequencies. A narrow band reduction is possible using a “notched” filter setting to limit 60-Hz interference (or 50 Hz in some countries outside the United States). Proprietary software offers digital seizure and spike detection capabilities for digital EEG. These systems are commercially available for both standard and prolonged EEG monitoring. This section will encompass patterns of cerebral and extracerebral origin as well as patterns of uncertain significance to illustrate the range of normal EEG encountered in clinical practice. Like most interpreters for clinical EEG, the orderly approach includes (a) review of the parameters of recording, (b) identifying the state of the patient and background activity, and (c) the most salient feature(s) of the epoch.

24  Chapter 2: Normal EEG

NORMAL EEG

FIGURE 2.6.  Normal 10-Hz alpha rhythm “blocked” by eye opening and returning on eye closure. Note the faster frequency immediately on eye closure (“squeak”).

T

he alpha rhythm is the starting point in the orderly approach to analyze EEG. In the normal EEG, a posterior dominant rhythm is represented bilaterally over the posterior head regions and lies within the 8- to 13-Hz bandwidth (alpha frequency). When this rhythm is attenuated with eye opening, it is referred to as the alpha rhythm (Figure 2.6). During normal development, an 8-Hz alpha frequency appears by 3 years of age. The alpha rhythm remains stable between 8 and 12 Hz even during normal aging into the later years of life. In approximately 25% of normal adults, the alpha rhythm is poorly visualized, and in 50% suggests a cortical grey matter abnormality within the region or hemisphere having the lower amplitude. However, lesser asymmetries may simply reflect normal skull asymmetries. A skull defect may produce a breach rhythm with focal, asymmetric, higher amplitudes (this relative increase may be >3 times). Beta activity is more prominent without the skull to attenuate the faster frequencies. A breach rhythm (Figure 2.8) on the EEG is considered normal for the physiological conditions of recording when it is independent of spikes or focal slowing.

Normal EEG  27

FIGURE 2.9.  Normal frontocentral theta in an 18-year-old while awake.

T

heta rhythms are composed of 4- to 7-Hz frequencies of varying amplitude and morphologies. Approximately one-third of normal awake, young adults show intermittent 6- to 7-Hz theta of 5,000 ohms when a single electrode is involved, as well as ensuring that ground loops and double grounds do not put the patient at a safety risk when generalized 60-cycle artifact is found, as in the examples provided in Figures 3.13A–C. (Continued)

68  Chapter 3: Artifacts of Recording

FIGURE 3.13B.  After notched filter application. Note the persistence of myogenic artifact that persists in the right temporal derivations. (Continued )

FIGURE 3.13C.  60-Hz artifact present in a compressed spectral array of frequency versus time. Note the discrete representation of a 60-Hz frequency (oval) using the fast Fourier transform.

FIGURE 3.14A.  Vagus nerve stimulator artifact (right half of the EEG tracing) recorded during electrical stimulation.

FIGURE 3.14B.  Deep brain stimulator artifact (ovals) during continuous video-EEG monitoring. Note the subclinical seizure that is ending (arrows) despite ongoing artifact. The second patient was admitted for convulsive status epilepticus with persistent nonconvulsive status epilepticus and stimulator artifact simulating nonconvulsive seizures.

E

lectrical artifact may occur from internal mechanical sources from electronic circuits surgically implanted (such as pacemakers or neurostimulators) devices to produce undesirable signals that contaminate the EEG or ECG recording. In this way, the patient or unshielded electrodes act as an antenna and produce extracerebral sources of artifact, like nearby power lines, that create external 60-Hz interference (Figure 3.14A–B). This is accomplished by the inductance of magnetic fields created from nearby current flow. It is the current flow from the neurostimulator that results in the artifact on EEG. Electrodes are depolarized by the electrical current and is amplified by the machine amplifiers to result in the “noise” that is visually apparent on EEG.

70  Chapter 3: Artifacts of Recording

FIGURE 3.15.  Mechanical artifact induced by continuous positive airway pressure (CPAP) in a comatose patient in the ICU. Note the alternating polarity of the mechanical artifact and low voltage.

A

variety of artifacts may be seen in the electrically hostile environments of the ICU, CCU, CSU, and OR that are produced by many different mechanical and instrumental sources. Electrical induced “noise” can be more evident for routine mechanical function at low gain (high sensitivity) settings. Alternating movement generated by a respirator is noted in the above example using high sensitivities of 3 μV/mm in a patient who is intubated and mechanically ventilated with continuous positive airway pressure (CPAP; Figure 3.15).

Artifacts of Recording  71

FIGURE 3.16.  Phone ring artifact on the EEG noted during in-patient long-term video-EEG monitoring.

E

nvironmental artifacts may often not be readily identifiable or correctable within the confines of a “hostile” environment when performing EEG in the ICU or CCU. Others may be common sources that are ubiquitous and intermittent within the environment, defying identification. Many environmental artifacts in special care units may be generated by high frequencies produced by nearby electrical machinery not directly connected to the patient. Equipment such as blood warmers, bovies, and electrical beds in the OR may make it challenging to locate the source. Other radio frequencies may be eliminated by simply unplugging or moving the source to redirect the electrical current flow away from the recording electrode. Telephone lines may interfere with EEG recording and produce artifact (Figure 3.16) that typically appears in all the channels during recording.

72  Chapter 3: Artifacts of Recording

FIGURE 3.17A.  EEG during a psychogenic nonepileptic attack with rhythmic unilaterally predominant movement artifact that simulates an electrographic focal seizure.

P

eople with psychogenic nonepileptic attacks experience recurrent episodes of paroxysmal movement or neurological impairment that resemble epileptic seizures but are not associated with electrophysiologic abnormalities of the brain. About 20% to 30% or more of patients admitted for video-EEG monitoring have psychogenic nonepileptic attacks. Distinguishing between epilepsy and nonepileptic attacks may require review of the EEG in concert with video analysis. The EEG may display rhythmic movements that may appear to be unilateral, especially when unilateral motor movements occur (Figure 3.17A). These artifacts may hamper the interpretation of the EEG and lead to a false diagnosis of epileptic seizures. The time-frequency analysis of EEG artifact in patients with psychogenic nonepileptic attacks remains stable and unchanged throughout the duration of the event. In addition, while amplitude often may vary, the morphology is monomorphic without evolution of the frequencies or evidence of a postictal state. Rhythmic epileptiform activity that manifests temporal and spatial evolution represent a characteristic feature on EEG present in patients with epileptic seizures. The pattern of artifact may be helpful identifying some patients with focal seizures when artifactual evolution is present, from a continuous pattern to a phasic decrescendo pattern of varying frequencies prior to termination (Figure 3.17B). EEG artifact may be helpful even when semiology appears confusing. Some basic rules underlying identification of artifact (Box 3.1) apply whether scalp EEG or intracranial EEG is being performed. Infrequently, a physiologic nonepileptic event (notably syncope) may mimic patients with epilepsy. In this case the EEG and ECG are important to record. ECG like the EEG is subject to artifact and needs to have artifact considered when abnormalities are suggested (Figure 3.17C). (Continued)

Artifacts of Recording  73

FIGURE 3.17B.  A frontal lobe seizure on EEG manifest as evolving diffuse myogenic artifact with abrupt termination. (Continued )

FIGURE 3.17C.  Pseudo-ventricular tachycardia due to electrode artifact (placed on the chest) in an asymptomatic patient during video-EEG monitoring.

74  Chapter 3: Artifacts of Recording

CONCLUSION Artifact is present in essential every EEG. There are many different types of nonphysiologic and physiologic artifacts that are encountered during the process of recording EEG. Not all artifacts are contaminants or noise, and occur as essential features of the EEG to understand basic functions of the brain. Still, other forms of artifact may mimic abnormalities and beguile the EEG interpreter to misidentify waveforms as an abnormality. When it becomes continuous or involves most of the channels for most of the recording, then artifact limits the clinical usefulness of the EEG. Skilled EEG technologists are critical to recognizing artifact and identifying the source. Elimination of or camouflaging unhelpful sources of artifact is best performed during the recording and requires a team of experienced personnel, including physicians, EEG technologists, nurses, and informatics and technology specialists, to ensure optimal EEG calibration and recordings. National and international society guidelines are available that have set minimum standards for recording EEG. With the complex nature imposed by artifact, even experienced EEG technologists and interpreting clinicians will be challenged to recognize every artifact that appears in the EEG. It is the common occurrence of artifact that emphasizes the need for conservative interpretation when challenging waveforms arise.

ADDITIONAL RESOURCES Bayly J, Carino J, Petrovski S, et al. Time-frequency mapping of the rhythmic limb movements distinguishes convulsive epileptic from psychogenic nonepileptic seizures. Epilepsia. 2013;54:1402–1408. https://doi.org/10.1111/epi.12207 Beniczky S, Conradsen I, Moldovan M, et al. Quantitative analysis of surface electromyography during epileptic and nonepileptic convulsive seizures. Epilepsia. 2014;55:1128–1134. https://doi. org/10.1111/epi.12669 Gaspard N, Hirsch LJ. Pitfalls in ictal EEG interpretation: critical care and intracranial recordings. Neurology. 2013;80:S26–S42. https://doi.org/10.1212/WNL.0b013e31827974f8 Mathias SV, Bensalem-Owen M. Artifacts that can be misinterpreted as interictal discharges. J Clin Neurophsiol. 2019;36(4):264–274. https://doi.org/10.1097/WNP.0000000000000605 Maulsby RL. Some guidelines for assessment of spikes and sharp waves in EEG tracings. Am J EEG Technol. 1971;11:3–16. https://doi.org/10.1080/00029238.1971.11080808 McKay JH, Tatum WO. Artifact mimicking Ictal Epileptiform activity in EEG. J Clin Neurophysiol. 2019;36(4):275–288. https://doi.org/10.1097/WNP.0000000000000597 Mizrahi EM. Avoiding the pitfalls of EEG interpretation in childhood epilepsy. Epilepsia. 1996;37(suppl 1):S41–S51. https://doi.org/10.1111/j.1528-1157.1996.tb06021.x Newman-Toker DE, Pronovost PJ. Diagnostic errors: the next frontier for patient safety. JAMA. 2009;301:1060–1062. https://doi.org/10.1001/jama.2009.249 Rosado Coelho C, Fernandez-Baca Vaca G, Luders HO. Electrooculogram and submandibular montage to distinguish different eye, eyelid, and tongue movements in electroencephalographic studies. Clin Neurophysiol. 2018;129:2380–2391. https://doi.org/10.1016/j.clinph.2018.09.011 Tatum WO. Artifact-related epilepsy. Neurology. 2013;80(suppl 1):S12–S25. https://doi.org/10.1212/ WNL.0b013e3182797325 Tatum WO, Dworetzky BA. Artifacts of recording and common errors in interpretation. In: Schomer DL, Lopes da Silva F, eds. Niedermeyer’s Electroencephalography. 7th ed. Oxford University Press; 2018:266–316. https://doi.org/10.1093/med/9780190228484.003.0011 Tatum WO, Dworetzky BA, Freeman WD, Schomer DL. Artifact: recording EEG in special care units. J Clin Neurophysiol. 2011;28:264–277. https://doi.org/10.1097/WNP Tatum WO, Dworetzky BA, Schomer DL. Artifact and recording concepts in EEG. J Clin Neurophysiol. 2011;28:252–263. https://doi.org/10.1097/WNP.0b013e31821c3c93 Tatum WO, Rubboli G, Kaplan PW, et al. Clinical utility of EEG in diagnosing and monitoring epilepsy in adults. Clin Neurophysiol. 2018;129:1056–1082. https://doi.org/10.1016/j.clinph.2018.01.019.

4 Abnormal EEG: Nonepileptiform Selim R. Benbadis and Elson L. So

I

nterictal EEG provides useful information about the presence of nonepileptiform neurophysiological dysfunction. When abnormalities are encountered, they are not specific for an underlying etiology, and as such represent abnormalities without further differentiation of the pathological process. While neuroimaging demonstrates anatomical definition of an abnormality, the EEG provides evidence of abnormal neurophysiological function when neuroimaging is normal. The EEG is sensitive to cerebral dysfunction but may have a lag during clinical improvement or lead relative to maximal clinical symptomatology. Many of the patterns that are nonepileptiform are due to a nonspecific etiology. Still, the presence of a nonepileptiform abnormality reflects the clinical presence of abnormality and often parallels the degree of dysfunction. Acuity is unable to be demonstrated by EEG in nonepileptiform abnormalities, although serial tracing may further help to define the trend of neurological improvement or deterioration. Therefore, the EEG is objectively able to substantiate and quantify the degree or depth of encephalopathy when diffuse nonepileptiform abnormalities are encountered. Furthermore, they may lateralize (or even localize) abnormalities when focal areas of slowing are evident. Many nonepileptiform and epileptiform abnormalities may help characterize the encephalopathy when the two features are identified on the EEG. This chapter will discuss and focus on generalized and focal nonepileptiform abnormalities.

76  Chapter 4: Abnormal EEG: Nonepileptiform

DIFFUSE ABNORMALITIES Diffuse slowing on the EEG may have various morphologies, and occur intermittently or continuously, to reflect abnormal cerebral function. The presence of diffuse slowing suggests a bilateral disturbance of cerebral function and represents an encephalopathy that is nonspecific for etiology.

FIGURE 4.1.  An abnormal high amplitude burst of diffuse intermittent theta in an awake adult following a motor vehicle accident associated with driving under the influence. Superimposition of myogenic “spikes” can make this burst of theta appear falsely epileptiform.

I

ntermixed diffuse intermittent theta in the most alert state is normal in young adults. When theta frequencies are seen in the frontal or frontocentral regions and voltages are greater than 100 µV or when theta is present greater than 10% of the time in an adult (not in childhood or elderly), then theta may reflect a nonspecific abnormality similar to diffuse intermittent slowing or background slowing. The slower the frequency, the higher the amplitude, and the greater the persistence, the more likely intermittent theta is abnormal (Figure 4.1).

Diffuse Abnormalities  77

FIGURE 4.2.  Generalized monomorphic 5- to 6-Hz theta frequencies obtained during a syncopal episode in a patient undergoing head-up tilt table testing for neurocardiogenic syncope.

D

iffuse (or generalized) slowing of the background electrocerebral activity reflects a nonspecific abnormality (Figure 4.2). It is indicative of a bilateral disturbance of cerebral function. With progression of cerebral dysfunction, the degree of generalized abnormal nonepileptiform abnormalities increases. Abnormally intermixed intermittent slowing that is manifest initially as intermittent theta (sometimes normal as discussed above) progresses to involve a greater degree of intermittent slowing, first becoming continuous theta slowing that is subsequently replaced by greater amounts of higher amplitude delta frequencies.

78  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.3.  EEG demonstrating diffuse slowing of the posterior dominant rhythm to 6 Hz. This degree of slowing of a well-defined background below 8 Hz is abnormally slow even in a 65-year-old man.

B

ackground slowing is defined as slowing of the posterior background activity to a frequency slower than the lower limits of a normal alpha rhythm frequency of 8 Hz (Figure  4.3). Diffuse slowing of the posterior dominant rhythm is a feature of encephalopathy. The degree of slowing of the background reflects the degree of cerebral dysfunction. The greater degree of background slowing reflects a more severe encephalopathy. Abnormality is defined when a posterior dominant rhythm that is present and that is normally reactive appears too slow for the patient’s age. The lower limits of normal for an alpha rhythm is 5, 6, 7, and 8 Hz at ages 1, 3, 5, and 8 years, respectively. Often times diffuse slowing of the background is associated with other stigmata of mild diffuse encephalopathy, such as intermittent bursts of generalized theta or delta activity.

Diffuse Abnormalities  79

FIGURE 4.4.  An intermittent 4-sec burst of irregular 1- to 2-Hz delta activity occurring on a diffusely slow posterior dominant rhythm of 6 Hz. This 55-year-old woman was clinically confused and disoriented, with multiple metabolic and systemic disturbances.

D

iffuse intermittent slowing is characterized by intermittent bursts of diffuse slow activity. This usually appears in the delta range and often appears in addition to background slowing of the posterior dominant rhythm (see Figure 4.4). Like background slowing, with which it frequently coexists, it is indicative of a diffuse encephalopathy. The bursts are usually irregular or polymorphic but can occasionally be rhythmic. As the severity of the encephalopathy increases, the bursts will increase in duration and frequency and merge into or become continuous generalized slowing (see continuous generalized slowing, page 60). Like other encephalopathic patterns, the presence of diffuse intermittent slowing is nonspecific relative to an etiology and may reflect either the presence of cortical or subcortical cerebral dysfunction.

80  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.5.  Frontal intermittent rhythmic delta activity in a 67-year-old patient with noncommunicating hydrocephalus. Note the slower 1.0- to 1.5-Hz frequency and the cerebral origin that is verified by the eye movement monitors demonstrating “in phase” deflection.

F

rontal intermittent rhythmic delta activity (FIRDA) appears as bursts of intermittent delta. These intermittent bursts are often high voltage, bisynchronous, and monomorphic waveforms (Figure 4.5). FIRDA may rarely appear asymmetric, especially when a focal structural lesion is present. However, FIRDA typically has a bilaterally symmetric electrophysiological field. FIRDA may appear normally during the buildup associated with hyperventilation (see Chapter 1). An abnormal pattern exists when FIRDA occurs in the waking adult EEG consisting of bilateral rhythmic monomorphic delta waves. The frequency is usually consistent throughout the EEG when it appears. Bifrontal predominance is typical in adults, and occipital predominance is more typically seen in children, shifting with brain maturation. FIRDA is most often associated with encephalopathies of toxic or metabolic origin. It also occurs with subcortical lesions, such as a deep midline lesion, or increased intracranial pressure.

Diffuse Abnormalities  81

FIGURE 4.6.  Occipital intermittent rhythmic delta activity (OIRDA) in a 6-year-old child with absence epilepsy.

O

ccipital intermittent rhythmic delta activity (OIRDA), like FIRDA, is a nonspecific finding in the EEG relative to etiology but in children often associated with generalized epilepsies. OIRDA is demonstrated as a posterior predominant bisynchronous rhythmic delta slowing appearing in bursts (Figure 4.6). OIRDA has the same features as FIRDA but occurs almost exclusively in children as a maturation-related spatial feature in EEG. OIRDA appears maximal over the occipital region instead of appearing with frontal predominance. OIRDA has been noted to occur in association with generalized (absence) epilepsy but is not an epileptiform abnormality unless intermixed spikes are present.

82  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.7.  Continuous irregular 1.5- to 3.0-Hz delta in a 66-year-old man with encephalopathy who was unresponsive. The above sample of EEG was representative of the entire record. No reactivity of the background was present to somatosensory stimulation.

C

ontinuous generalized slowing consists of polymorphic delta activity that is continuous or near-continuous (greater than 80% of the record) and unreactive (Figure 4.7). Unreactive implies no change in the background electrocerebral activity produced with external stimuli in addition to the absence of sleep–wake patterns. Unlike the prior two examples of background slowing and intermittent generalized slowing, continuous generalized unreactive polymorphic delta slowing is indicative of a severe diffuse encephalopathy. Most patients with this feature are comatose or stuporous. Like the other nonepileptiform abnormalities associated with encephalopathy, this finding is nonspecific as to etiology. The most common causes by far are toxic-metabolic or systemic disturbances. However, severe diffuse, bilateral structural lesions or injury affecting the brain can also produce this pattern (e.g., traumatic brain injuries or advanced neurodegenerative diseases).

Diffuse Abnormalities  83

FIGURE 4.8.  Low-voltage EEG recording in a patient with head injury following a motor vehicle accident. The recording was obtained at a sensitivity of 2 µV/mm with no waveforms demonstrating a voltage of greater than 20 µV.

L

ow-voltage EEG is typically associated with diffuse slowing of the background rhythm.  In general, the state of the patient is the best indicator of abnormality, with some lowvoltage EEGs of less than 10 to 20 µV found in a subset of normal individuals. When seen during encephalopathy or coma, low-voltage EEG (Figure 4.8) is typically associated with diffuse slowing and poor reactivity to somatosensory stimulation. One distinguishing characteristic is the lack of admixed alpha and beta frequencies in this low-voltage recording.

84  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.9.  Three-year-old presented with first-time nonfebrile convulsion, but had prior staring spells. She was born full-term, but she sat up at 9 months, spoke first words at 14 months, walked at 16 months, and continued to walk “clumsily.” She was usually happy and smiling.

T

he notched delta pattern characteristic of Angelman syndrome typically shows runs of high amplitude 2- to 3-Hz delta waves maximal at the bifrontal regions, with a spike or sharp wave superimposed on the delta wave close to its peak, at the peak, or following the peak. There is no behavioral alteration when runs of notched delta waves are occurring, and therefore, the pattern on EEG is not ictal (Figure 4.9). In Angelman syndrome children less than 12 years old, 4 to 6 Hz rhythmic slowing can be seen with spikes at the occipital region, often enhanced with eye closure. Angelman syndrome patients have higher incidence of epilepsy, but the notched delta EEG pattern does not distinguish those with from those without epilepsy. The genetic basis of the syndrome involves chromosome 15q11-13. It has been reported that patients with deletion abnormality involving this gene have a more prominent typical EEG pattern than patients who have other mechanisms of the chromosome abnormality.

Focal Abnormalities  85

FOCAL ABNORMALITIES Focal abnormalities on the EEG provide electrographic evidence of a localized abnormal cerebral function. They are not specific for etiology and may be seen with many different underlying structural lesions that affect the brain. They may also be encountered as a temporary nonstructural physiological effect (i.e., following a seizure). The location, morphology, persistence, and poor reactivity are features that suggest an underlying structural lesion, but because the specificity is low, a broad differential is required.

FIGURE 4.10A.  Alpha asymmetry (slower over the right hemisphere) in a patient with an acute right fronto-parietal ischemic infarction. (Continued)

A

lpha asymmetries depict an abnormality on the side ipsilateral to the hemisphere and  characteristically involve a slow posterior dominant rhythm (Figure 4.10A). Additional focal, regional, or lateralized abnormalities are often seen in conjunction with alpha asymmetries. A persistent hemispheric difference of greater than 1 Hz should be regarded as abnormal when alpha asymmetry is seen. Additionally, while the right hemisphere is normally asymmetrical with respect to the higher amplitude, a persistent amplitude asymmetry of greater than 50% should be regarded as abnormal (Figure 4.10B). (Continued)

86  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.10B.  Right hemisphere suppression in a patient with a right hemisphere intracerebral hemorrhage. Note the dramatic difference in amplitude at standard sensitivity settings of 7 uV/mm.

Focal Abnormalities  87

FIGURE 4.11A.  Focal delta in a 28-year-old patient with right temporal polymorphic delta due to an anterior temporal ganglioglioma. Note the anterior-mid-temporal localization with loss of the intermixed faster frequencies.

F

ocal polymorphic delta is confined to one or two electrode contacts and indicates a more restricted disturbance of cerebral dysfunction affecting the white matter tracts (Figure 4.11A). When concomitant loss of faster frequencies is seen (above), these findings on the EEG may be more suggestive of a structural lesion that affects both the ipsilateral gray and white matter of the hemisphere. The more focal and persistent the polymorphic delta slowing, the greater the likelihood a structural lesion will be present. Continuous regional delta slowing may also appear on invasive EEG with the same frequencies represented as scalp recording (Figure 4.11B). (Continued)

88  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.11B.  Sample of intracranial EEG using subdural electrodes. Note the focal slow (delta) activity at contacts OF3, grid 22 to 24, and grid 16 to 18. Rules for localization are similar on invasive recordings, and on referential montages this indicates a focal area of dysfunction that was found to be caused by a small hemorrhage.

Focal Abnormalities  89

FIGURE 4.12.  Temporal intermittent rhythmic delta activity (TIRDA) in a patient with left temporal lobe epilepsy. Note the brief burst of regional bisynchronous bitemporal TIRDA (line).

T

emporal intermittent rhythmic delta activity (TIRDA) is a unique form of intermittent rhythmic delta activity. It consists of intermittent focal bursts of monomorphic delta frequencies maximal typically in a unilateral temporal derivation (Figure 4.12). The presence of TIRDA has a strong association with focal seizures. It may provide localizing capabilities in patients with temporal lobe epilepsy. TIRDA is often associated with interictal epileptiform discharges (IEDs) and is abnormal when it occurs during the awake state and is persistent.

90  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.13.  There is a brief 2-sec burst of polymorphic delta activity in the posterior temporal–parietal region of the left hemisphere in a 55-year-old patient with a left subcortical white matter lacunar infarction. Note the phase-reversals at T5 and P3 (arrows).

I

ntermittent irregular slowing (Figure 4.13) has a low correlation with an underlying lesion compared to focal slowing that is continuous. Focal slowing suggests an underlying structural lesion involving the white matter tracts of the brain. Definite statements about the specific etiology due to the intermittent irregular slow activity cannot be derived by the appearance of the EEG as with most nonepileptiform abnormalities.

Focal Abnormalities  91

FIGURE 4.14A.  A 75-year-old patient with an acute left frontal ischemic infarct. Note the left hemispheric temporally predominant polymorphic delta that affects the entire hemisphere.

C

ontinuous regional delta slowing on the EEG has a high correlation with an underlying structural lesion involving the white matter of the ipsilateral hemisphere (Figure 4.14A). The area of slowing usually overlies the hemisphere containing the structural lesion but does not necessarily reflect the precise location as the one represented by EEG. For example, frontal lobe lesions may appear as ipsilateral continuous regional delta slowing on EEG. Trauma, tumor, stroke, intracranial hemorrhages, and infection all create a similar appearance on the EEG without specific features. The EEG may be abnormal and demonstrate continuous regional delta slowing that indicates a reversible structural lesion with resolution (Figure 4.14B). (Continued)

92  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.14B.  Typical focal or “regional” (not hemispheric) slowing in a patient with transient aphasia but no stroke on imaging. The delta activity is present in the left frontotemporal region, but not in the paracentral area. In the absence of a structural abnormality on brain MRI, this patient was felt to have had a transient ischemic attack, and less likely a focal seizure with postictal slowing on EEG.

Focal Abnormalities  93

FIGURE 4.15A.  This EEG was taken from a 64-year-old after a right hemisphere ischemic infarct. Note that a well-formed alpha rhythm is not present in the right hemisphere compared with the left and is instead replaced by continuous right hemispheric polymorphic delta slowing.

L

ateralized polymorphic delta slowing may consist of theta or delta frequencies that are focal, regional, or lateralized. Delta that is polymorphic (and arrhythmic) is composed of slow-wave activity that is 3.5 Hz (or less) and is composed of waveforms that vary in frequency and duration (Figure 4.15A). Polymorphic delta activity when localized is indicative of an underlying supratentorial lesion affecting the white matter of the ipsilateral hemisphere. The more state-independent and persistent the delta slowing appears, the greater the likelihood that a structural lesion will be present. Lateralized or even localized polymorphic delta, however, may exist as a transitory phenomenon on the EEG due to head injury, transient ischemic attack, migraine, and during a postictal state. Iatrogenic causes such as the intracarotid amobarbital procedure (aka Wada test) may also produce hemispheric slowing on the EEG due to the transitory effects of drug-induced hemianesthesia when it is used for presurgical lateralization of language and memory function (Figure 4.15B). (Continued)

94  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.15B.  Right hemisphere slow activity induced by intracarotid methohexital (3 mg) during a Wada test. This illustrates typical lateralized (“focal”) hemispheric dysfunction. This only lasts a few minutes due to the short half-life of methohexital.

Focal Abnormalities  95

FIGURE 4.16.  Asymmetry of sleep spindles in a 36-year-old patient with a right thalamic glioma.

S

leep spindles are initially evident in the first 2 months, and by 2 years of age are synchronous in normal children. Sleep elements are normally maximal in frequency in the central location, although they may appear in the frontal regions as well. A frequency of 12to 14-Hz is observed in the central regions and is the distinguishing characteristic of stage N2 sleep. Spindles are very stable in their bilateral appearance and a persistent slowing of the spindle frequency or the unilateral appearance of sleep spindles should be regarded as an abnormal nonepileptiform feature (Figure 4.16).

96  Chapter 4: Abnormal EEG: Nonepileptiform

FIGURE 4.17.  Fast focal midline central rhythm (arrow) in a 67-year-old man who had discomforting sense of unsteadiness when standing for a prolonged period of time, such as when queuing at a store cashier line. The discomfort was attenuated by leaning against a support, or by walking. His evaluation supported a diagnosis of orthostatic tremor (aka Shaky Leg Syndrome).

A

lthough this finding can be nonspecific, it can be helpful in diagnosing orthostatic tremor in the setting of an appropriate clinical picture. Nonetheless, it is present in only about 25% of the patients with orthostatic tremor. The fast rhythm is maximal at Cz, and its frequency ranges from 14- to 24-Hz (Figure 4.17). Motion analysis of the tremor with electromyogram (EMG) shows that the tremor rate ranges from 13- to 18-Hz, compared with 4-to 8-Hz in Parkinson’s tremor and approximately 12-Hz in essential tremor. The patient may have to be standing to bring out either the EEG or EMG feature.

CONCLUSION EEG can reveal many types of nonepileptiform abnormalities corresponding to varied cerebral dysfunctions occurring at any age. Despite advancements in neuroimaging, EEG has continued to be a valuable diagnostic tool for assessing cerebral functions and revealing underlying mechanisms or etiologies. It provides objective assessment of patients with altered cognition or consciousness. Serial EEGs are useful in evaluating the clinical course of these patients. It therefore guides therapeutic management and clinical prognostication. Traumatic brain injury, strokes, and dementia remain common neurological disorders. Neuroimaging in these disorders is complemented by EEG for objective assessment of the degree of brain injury and dysfunction. Inflammatory cerebral disorders, especially

Additional Resources  97

autoimmune encephalopathies, are gaining attention and concern. Nonepileptiform EEG abnormalities are present more often than not in all these cerebral disorders. The predominant type of EEG abnormalities in these disorders is slowing of the background EEG. The location and extent of the slowing should be carefully noted. Clinical implications vary among unifocal, multifocal, and generalized slowing. The EEG interpreter should also closely observe for EEG slowing that has a characteristic or known pattern, because patterns can suggest the underlying basis of the clinical condition. EEG recording should not be abbreviated after nonepileptiform abnormalities are first noted. Continued recording permits observation of the degree of persistence of the slowing and the effect of environmental or applied stimulation. Reactivity of nonepileptiform abnormalities is an important feature that helps assess the cause and prognosis of the clinical condition.

ADDITIONAL RESOURCES Benbadis SR. Focal disturbances of brain function. In: Levin KH, Lüders HO, eds. Comprehensive Clinical Neurophysiology. Saunders; 2000:457–467. Epstein CM, Riecher AM, Henderson RM, et al. EEG in liver transplantation: visual and computerized analysis. Electroencephalogr Clin Neurophysiol. 1992;83:367–371. https://doi.org/ 10.1016/0013-4694(92)90072-P Gloor P, Kalabay O, Giard N. The electroencephalogram in diffuse encephalopathies: electroenephalographic correlates of gray and white matter lesions. Brain. 1968;91:779–802. https://doi. org/10.1093/brain/91.4.779 Kaplan PW. Metabolic and endocrine disorders resembling seizures. In: Engel J Jr, Pedley TA, eds. Epilepsy: A Comprehensive Textbook. Lippincott Raven; 1997:2661–2670. Laan LAEM, Vein AA. Angelman syndrome: is there a characteristic EEG? Brain & Develop. 2005;27:80–87. https://doi.org/10.1016/j.braindev.2003.09.013 Liporace J, Tatum W, Morris GL, et al. Clinical utility of sleep-deprived versus computerassisted ambulatory 16-channel EEG in epilepsy patients: a multicenter study. Epilepsy Res. 1998;32:357–362. https://doi.org/10.1016/S0920-1211(98)00069-2 Luders H, Noachtar S, eds. Atlas and Classification of Electroencephalography. Saunders; 2000. McManis PG, Sharbrough FW. Orthostatic tremor: clinical and electrophysiologic characteristics. Muscle & Nerve. 1993;16:1254–1260. https://doi.org/10.1002/mus.880161117 Schaul N, Gloor P, Gotman J. The EEG in deep midline lesions. Neurology. 1981;31:157–167. https://doi. org/10.1212/WNL.31.2.157 Zifkin BG, Cracco RQ. An orderly approach to the abnormal electroencephalogram. In: Ebersole JS, Pedley TA, eds. Current Practice of Clinical Electroencephalography. 3rd ed. Lippincott Williams & Wilkins; 2003:288–302.

5 Abnormal EEG: Epileptiform William O. Tatum, IV

I

nterictal epileptiform discharges (IEDs) represent a distinct group of waveforms that are characteristically seen in people with epilepsy. Variations of normal background rhythms, a variety of artifacts, and variants of uncertain significance may mimic abnormal IEDs and lead to overinterpretation of the EEG and untoward patient consequences (Chapter 1). IEDs have reliably been associated with epilepsy at rates sufficient to be clinically relevant. Although prominent intra-patient and inter-patient variability in frequency and morphology of IEDs may occur, those patients with prominent IEDs on the EEG are not necessarily the patients with more severe epilepsy. Scalp detection of IEDs is based upon dipole localization and the surrounding field. The resultant scalp detection of the source may appear different on the scalp EEG than the actual site of seizure genesis. In most cases, an IED reflects a radial dipole that is oriented to be detected on the scalp. However tangential dipoles may commonly occur in certain epilepsy syndromes (i.e., self-limited epilepsy with centrotemporal spikes [SLECTS]) or from developmentally or surgically altered cortex may produce unusual dipoles that challenge the EEG reader. Horizontal dipoles are better detected by magnetoencephalography which is often complimentary to source localization. Rarely, normal individuals may possess IEDs on EEG without the phenotypic expression of seizures. The photoparoxysmal (PPR) response, generalized spike-and-wave (GSW), and centrotemporal IEDs are the most frequent asymptomatic IEDs encountered in the absence of clinical seizures. When they are encountered, they typically reflect the genotype of an inherited trait that is represented on the EEG without the phenotypic expression of clinical seizures. Additionally, focal IEDs have a variable association with clinical epilepsy and depend upon the location of their appearance. For example, central, parietal, and occipital spikes, in general, may be less likely to be associated with clinical epilepsy than IEDs associated with the frontal and temporal location in the absence of a structural lesion. IEDs have seen in migraine, certain drugs such as lithium or clozapine, the autism spectrum disorder, cerebral palsy, and blindness, among other conditions. The interictal EEG plays a pivotal role in providing ancillary support for a clinical diagnosis of epilepsy. The presence of abnormal IEDs occurs in 5 seconds prior over contact U1-2 and W3-4. Patient underwent laser ablation of the center (electrode U) and posterior aspect (electrode W) of the tumor, sparing hippocampus, and anterior border of the lesion. He has remained seizure-free for >2 years without memory decline.

FIGURE 9.12.  A 49-year-old right-handed man had a 25-year seizure history of episodes of epigastric or olfactory aura followed by staring, automatism and postictal aphasia. Surface EEG showed left temporal sharp waves and left temporal seizure onset. MRI was negative; FDG PET showed subtle left anterior temporal hypometabolism. Ictal SPECT showed marked increase perfusion in the anterior medial and lateral left temporal area. Memory function was intact. Subdural grid implant from 2012, before stereo EEG was available at our center, shows extensive lateral and basal temporal and orbitofrontal coverage with grid electrodes, a strip over the temporal tip and four depth electrodes targeting amygdala and hippocampus. Invasive EEG shows selected temporal grid and depth electrodes for 11s. Repetitive spikes are seen over the anterior-lateral grid A and E followed by abrupt attenuation and low voltage fast over A9-10 followed by propagation to the anterior hippocampus 3s later. The patient underwent anterior-lateral neocortical resection sparing the mesial structures and has remained seizure-free on anti-seizure medications.

206  Chapter 9: Invasive Video-EEG

FIGURE 9.13.  A 25-year-old right-handed woman provided a 3-year history of seizures with sudden reading difficulties and nonsensical speech lasting 30 seconds. Surface EEG showed left temporal sharp waves and left temporal seizure onset. MRI showed a nonenhancing lesion involving the left fusiform gyrus, suggestive of cortical dysplasia versus a low-grade glioma. Memory function was intact. (Left) Subdural implant shows extensive lateral and basal temporal coverage using grids and mesial depths targeting amygdala and hippocampus. (Right) 20s EEG in the upper panel shows a herald spike followed by low-voltage fast activity (LVFA) limited to the area anterior to the lesion (C4-5). 40s later, spread to the adjacent and more posterior basal temporal (B) grid is seen, still sparing the hippocampus. Patient underwent a lesionectomy including the seizure onset zone anterior to the lesion, sparing the mesial structures. Pathology showed a grade 2 astrocytoma.

ELECTRICAL STIMULATION MAPPING Electrical stimulation mapping (ESM) represents an integral part of any invasive evaluation, whether SDE or SEEG. Stimulation can be done by stimulating adjacent, bipolar electrodes or with a monopolar target electrode with an indifferent reference. In SDE studies, ESM focuses primarily on mapping brain function. Triggered responses can be either positive, that is, a visible movement or sensation reported by the patient, or negative, that is, a response which requires task-specific testing to demonstrate interruption of cognitive or motor function (such as performing alternating movements). Systematic cortical mapping with subdural grids allows a clear delineation of resection margins based on functional boundaries. Defining the suspected EZ in relation to eloquent cortex remains the primary indication for a subdural grid evaluation in many centers. In SEEG, ESM is performed to explore both eloquent function and the epileptogenic network. Compared to SDE, SEEG is limited in its ability to delineate functional boundaries, but does allow stimulation of areas inaccessible to subdural grids, such as the insula, amygdala-hippocampus, cingulate, operculum, depths of sulci, or subcortical functional pathways.

Additional Resources  207

Electrical stimulation is typically performed using a frequency of 50 Hz with a biphasic pulse, 0.3 ms pulse width, train duration of 2 to 10 s. For SDE, incremental intensities of 1 to 17.5 mA may be used, that lead to a maximum charge delivered of 17.5 mA × 0.3 ms = 5.25 µC. The functional threshold to elicit a response using SDE varies between 2 and 15 mA, though maximum stimulus intensities may be limited by afterdischarges. Commonly used grid electrodes have a surface contact diameter of 4 mm and a surface area of 0.126 cm2, leading to a maximum charge density of 5.25 µC/0.126 cm2 = 42 µC per cm2 per phase, which is below the recommended limit of 50 to 60 µC per cm2 per phase. Depth electrodes have a cylindrical surface of areas ≈0.05 cm2, less than half that of a grid electrode. Consequently, a current intensity of 7 to 8 mA achieves a similar charge density per phase: (8 mA × 0.3 ms) /0.05 cm2 = 48 µC per cm2 per phase, compared with grid electrodes. However, many recommend a stricter safety limit of 30 µC per cm2 per phase, which reduces the stimulation intensity threshold of a 0.3 ms pulse to 5 mA. It should be borne in mind that such a value may be below the threshold needed to activate some eloquent areas. A different use of ESM during SEEG is to trigger habitual electroclinical seizure activity. For this, 1-Hz repetitive stimuli may be used initially, with standard 50-Hz stimulation following as needed. The yield of a habitual electroclinical response is around 50% to 60% and higher in the temporal lobe and in SOZs characterized by LVFA. The false positive rate for stimulation-induced seizures is around 10% to 25%. Results of ESM for mapping seizures therefore need to be viewed in the context of the overall results of the SEEG evaluation.

CONCLUSION Invasive EEG monitoring remains an essential tool for defining the EZ in around 30% of patients considered for epilepsy surgery. SDEs remain popular in patients with neocortical epilepsies and in those with an EZ in proximity to eloquent cortex. Placement schemes for SDE are conceptually straightforward in overlaying the suspected SOZ and regions of rapid propagation. Interpretation of SDE in predicated similar principles of contiguous spread from localized onset. ESM in SDE serves the role of delineating functional cortex from the boundaries of proposed resection. SEEG has seen increasing use for patients with deep-seated lesions and those with nonlesional MRI. SEEG requires meticulous planning of electrode placement and data interpretation of SEEG is more complex. The importance of neurophysiologic biomarkers, SOP, and seizure stimulation remain to be further validated to provide consistent clinical reliability. However, these metrics harbor the potential for a more precise understanding of the optimal area to be resected to render an epilepsy patient seizure-free.

ADDITIONAL RESOURCES Bancaud J, Angelergues R, Bernouilli C, et al. Functional stereotaxic exploration (SEEG) of epilepsy. Electroencephalogr Clin Neurophysiol. 1970;28:85–89. https://doi.org/10.1016/0013-4694(70)90013-1 Frauscher B, von Ellenrieder N, Zelmann R, et al. Atlas of the normal intracranial electroencephalogram: neurophysiological awake activity in different cortical areas. Brain. 2018;141:1130–1144. https://doi.org/10.1093/brain/awy035 Gloor P. Neuronal generators and the problem of localization in electroencephalography. J Clin Neurophysiol. 1985;2:327–354. https://doi.org/10.1097/00004691-198510000-00002 Gnatkovsky V, Pelliccia V, de Curtis M, et al. Two main focal seizure patterns revealed by intracerebral electroencephalographic biomarker analysis. Epilepsia. 2019;60:96–106. https://doi.org/10.1111/ epi.14610

208  Chapter 9: Invasive Video-EEG

Kalamangalam GP, Long S, Chelaru MI. A neurophysiological brain map: spectral parameterization of the human intracranial electroencephalogram. Clin Neurophysiol. 2020;131:665–675. https://doi. org/10.1016/j.clinph.2019.11.061 Lagarde S, Buzori S, Trebuchon A, et al. The repertoire of seizure onset patterns in human focal epilepsies: determinants and prognostic values. Epilepsia. 2019;60:85–95. https://doi.org/10.1111/ epi.14604 Oderiz CC, von Ellenrieder N, Dubeau F, et al. Association of cortical stimulation–induced seizure with surgical outcome in patients with focal drug-resistant epilepsy. JAMA Neurol. 2019;76:1070– 1078. https://doi.org/10.1001/jamaneurol.2019.1464 Roehri N, Bartolomei F. Are high-frequency oscillations better biomarkers of the epileptogenic zone than spikes? Curr Opin Neurol. 2019;32:213–219. https://doi.org/10.1097/WCO.0000000000000663 Tao JX, Ray A, Hawes-Ebersole S, et al. Intracranial EEG substrates of scalp EEG interictal spikes. Epilepsia. 2005;46:669–676. https://doi.org/10.1111/j.1528-1167.2005.11404.x Trébuchon A, Chauvel P. Electrical stimulation for seizure induction and functional mapping in stereoelectroencephalography. J Clin Neurophysiol. 2016;33(6):511–521. https://doi.org/10.1097/ WNP.0000000000000313 von Ellenrieder N, Gotman J, Zelmann R, et al. How the human brain sleeps: direct cortical recordings of normal brain activity. Ann Neurol. 2020;87:289–301. https://doi.org/10.1002/ana.25651 Zijlmans M, Worrell GA, Dümpelmann M, et al. How to record high-frequency oscillations in epilepsy: a practical guideline. Epilepsia. August 2017;58(8):1305–1315. https://doi.org/10.1111/epi.13814

10 The EEG in Status Epilepticus Frank W. Drislane and Peter Kaplan

S

tatus epilepticus (SE) is a prolonged seizure or a series of seizures without recovery between them. There are many forms of SE, with different clinical and EEG manifestations. Deciding which type of SE a patient has is important in directing proper management and in predicting the outcome. EEG is often a crucial part of that evaluation. There are both convulsive and nonconvulsive forms of SE. Generalized convulsive SE (GCSE) is the best described type of SE and has the greatest morbidity, mortality, and clinical urgency for treatment. GCSE must be diagnosed and treated promptly, in part for concern that it may become refractory to treatment and have grave consequences. For most forms of nonconvulsive SE (NCSE), however, there is less evidence that it causes lasting (or in some cases, any) neurologic harm. Thus, the treatment imperative is less, but not negligible. NCSE has been referred to as “underdiagnosed and over-treated.” Continuous EEG recording can help elucidate the temporal pattern of recurrent seizures when subtle or no clinical signs are evident.

GENERALIZED CONVULSIVE STATUS EPILEPTICUS Clinical and EEG manifestations of generalized convulsions and GCSE are usually sym­ metric from the onset (Figure 10.1), although some may exhibit focal or lateralizing features, particularly at the onset or end of the seizure. On the EEG, true generalized seizures associated with genetic generalized epilepsies (GGEs) typically begin with bilaterally symmetric epileptiform discharges. The initial sudden interruption in behavior is often accompanied by widespread voltage attenuation and faster frequency rhythms between 20 and 40 Hz, producing an “electrodecremental” appearance of the ictal EEG (more often encountered in children). This is often the time when muscle artifact obscures the EEG recording of the tonic and clonic phases of the convulsion, although sometimes, epileptiform discharges may be evident at the vertex. The superimposed electromyography (EMG) activity itself may exhibit a characteristic rhythmic appearance in synchrony with the clinical features of repetitive clonic jerks.

210  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.1.  Generalized convulsive status epilepticus (GCSE) begins with a lower voltage faster frequency pattern, followed by prominent muscle artifact. © WO Tatum, 2013.

F

ollowing an individual seizure, repetitive epileptiform discharges may decrease from several per second to a frequency less than 1 Hz. The discharges might not disappear entirely, often giving way to generalized periodic discharges (GPDs) or lateralized periodic discharges (LPDs), usually on a suppressed, lower voltage background, and usually for several minutes before the gradual return of a normal background—or prior to another convulsion if GCSE has not been terminated. When GCSE is prolonged, the EEG becomes more discontinuous, and clinical manifestations may become minimal. When the visible motor manifestations of SE cease, this may be termed “subtle” GCSE. At this point, clinical signs are minimal and may include low amplitude eyelid or facial twitching, intermittent myoclonic jerks, repetitive or sustained nystagmus, or even the absence of all clinical movement. Persistence of regular and rhythmic GPDs at 2.5 Hz or greater suggests that nonconvulsive seizures are still ongoing and that the SE has become refractory despite the lack of clinical signs. At this stage, patients are typically comatose.

Generalized Convulsive Status Epilepticus  211

FIGURE 10.2A.  Seizures begin with rhythmic slowing (in this case focally) in right-sided channels, spreading bilaterally. © WO Tatum, 2013.

S

ome authors have proposed that there is a characteristic sequence (Figure 10.2 A–E) of EEG changes during GCSE—based on EEG recordings on animals and humans at various stages of (usually generalized convulsive) SE, although all such stages are seldom observed in an individual patient. This sequence of EEG changes includes: (A) Discrete seizures that are repetitive and separated electrographically by background slowing and attenuation between recurrent seizures. (B) Seizures that merge gradually, with some fluctuation in voltage and frequency. (C) Continuous seizure activity, but sometimes with asymmetric epileptiform discharges, reflecting the focal or lateralized onset of many seizures. (D) Ongoing seizure activity interrupted by brief periods of a suppressed background, often for just a second or so. (E) In late SE, the background becomes suppressed, with GPDs or infrequent bursts of polyspikes (Figures 10.2 D–E and 10.3). “Subtle” SE often corresponds to EEG stages D and E. (Continued )

212  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.2B.  Discrete seizures merge and become continuous, primarily over the right hemisphere. © WO Tatum, 2013. (Continued )

FIGURE 10.2C.  Seizures are continuous, still primarily over the right hemisphere. © WO Tatum, 2013. (Continued )

Generalized Convulsive Status Epilepticus  213

FIGURE 10.2D.  Seizure activity is interrupted by periods of a suppressed background. © WO Tatum, 2013. (Continued )

FIGURE 10.2E.  The background becomes more suppressed, and discharges recur with a longer periodicity. © WO Tatum, 2013.

214  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.3.  Following a seizure, there are often generalized periodic discharges (GPDs), gradually decreasing in frequency.

W

 hile the later phases of this EEG sequence are thought to reflect the uncoupling of the electrical and mechanical activity after prolonged SE, not all clinical neurophysiologists have found such a predictable sequence of EEG patterns in many patients. Still, these patterns may be useful in determining the approximate phase of SE a patient is in, and whether the SE is prolonged enough (electrographically, when not evident clinically) that treatment for refractory SE should be considered.

Focal Motor Status Epilepticus  215

FOCAL MOTOR STATUS EPILEPTICUS There are many causes of focal motor SE (FMSE). Stroke (ischemic or hemorrhagic), trauma, and infection are the most common. SE occurs in about 1% of all acute strokes, although isolated seizures are much more common. Central nervous system (CNS) infection (e.g., meningoencephalitis, often with herpes simplex encephalitis) may be manifested clinically as FMSE. Other etiologies of FMSE include mass lesions, autoimmune disorders, vasculitis, multiple sclerosis, and, rarely, mitochondrial or degenerative disorders. Rarely, benign idiopathic focal epilepsies, such as self-limited epilepsy with centrotemporal spikes (SLECTS) can lead to FMSE, but seizures in these syndromes are often treated easily and seldom lead to SE.

FIGURE 10.4.  Focal motor status epilepticus with intact consciousness, with semi-rhythmic slowing and sharp features over the left hemisphere, with EMG artifact from right facial twitching. © WO Tatum, 2013.

I

n focal motor SE, ictal EEG features are quite variable. Epileptiform activity may consist of discrete, frequently recurrent, focal motor seizures that are localized or lateralized to one hemisphere or have an asymmetric bihemispheric involvement when consciousness is impaired (Figure 10.4). There may be clinical recovery between seizures, or there may be continuous electrographic seizure activity. Following individual focal seizures, or between electrographic focal seizures on EEG, there may be continued slowing or LPDs. In terms of diagnosis and clinical outcome, there does not appear to be a large difference between the discrete and continuous forms of FMSE.

216  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.5.  Epilepsia partialis continua (EPC) in a 41-year-old patient with a sense of “tingling” and twitching on the left side of the mouth. Note primarily the rhythmic delta slowing, phase reversing at F8.

C

linically, FMSE may be manifested as prolonged regular jerking of an isolated area of the body such as the face, hand, or foot. This is referred to as epilepsia partialis continua (EPC) (Figure 10.5). EPC can last days or weeks. Almost always, there is a responsible focal lesion, but there are not always identifiable focal rhythmic epileptiform discharges on the surface EEG. Although repetitive discharges or rhythmic theta or delta slowing are also common patterns with EPC, some epileptic foci are generated in deeper cortex, have limited surface area involved (i.e., less than 10 cm2), or have a dipolar source that is not oriented “favorably” for detection by surface electrodes, so there is often minimal or no change detected on surface EEG despite ongoing focal seizures. LPDs are repetitive spike or sharp and slow wave complexes (see also Chapter 11).  They usually last 100 to 400 msec and typically recur at 0.5 to 2 Hz but sometimes have longer intervals. They are usually distributed broadly over most of one hemisphere, with an attenuated EEG background between discharges. Most epileptologists do not consider LPDs to be definite clinical seizures or SE per se, but rather on an ictal–interictal continuum, and often associated with clinical seizures and with acute, serious focal neurologic illness. LPDs have a greater correlation with the likelihood of clinical seizures occurring than do GPDs. In about 90% of cases there is a structural lesion. Stroke is the most common cause, although tumor, and occasionally, CNS infections, and severe metabolic disturbances, such as nonketotic hyperglycemia, may result in LPDs. Some types of epilepsy may also exhibit LPDs as an interictal or ictal phenomenon. In many cases, LPDs may be considered “the terminal phase of status epilepticus,” although they may also occur between seizures.

Focal Motor Status Epilepticus  217

FIGURE 10.6A.  The EEG of a 41-year-old man with a metastatic tumor shows lateralized periodic discharges over the left hemisphere. © F Drislane, 2012. Courtesy of the author. (Continued )

W

hile most LPDs are manifested on the EEGs as discharges recurring every 1 to 2 seconds, some intervals may be 10 seconds or longer. The more rapid discharges of > 2.5 Hz are a definite EEG correlate of seizures, but most electroencephalographers consider that there is a significant risk of ongoing seizures when the frequency of LPDs exceeds 1.5 or 2.0 Hz. ”LPDs+” is a term describing LPDs associated with low-voltage rhythmic epileptiform discharges or other rapid rhythms occurring between the high-voltage discharges. LPDs+ are more likely to be associated with epileptic seizures than are “LPDs proper” (Figure 10.6 A–C). (Continued )

218  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.6B.  Discharges become more complex, with frequent after-going low-voltage fast activity. © F Drislane, 2012. Courtesy of the author. (Continued )

FIGURE 10.6C.  These “lateralized periodic discharges (LPDs+)” are followed by brief bursts of electrographic seizure activity at about 3 Hz. © F Drislane, 2012. Courtesy of the author.

Myoclonic Status Epilepticus  219

MYOCLONIC STATUS EPILEPTICUS Myoclonic SE (MSE) has many causes. While the clinical appearance of recurrent lightning-like jerks may be similar in each, there are several MSE syndromes. Many cases of MSE occur in patients with syndromes associated with GGEs. In this case, inter­ictal myo­ clonus is common, typically consisting of a sudden quick muscular jerk involving primarily the upper body and often occurring in repetitive clusters. Myoclonus may involve eyelid myoclonia alone (e.g., in Jeavons syndrome). Common childhood GGEs include juvenile absence epilepsy and juvenile myoclonic epilepsy. The frequency of clinical SE is syndromedependent. MSE may occur following sleep deprivation, exposure to toxins (e.g., alcohol), when concentrations of anti-seizure medications (ASMs) are low (e.g., noncompliance), or if ASMs inappropriate for GGEs (particularly several sodium channel-based drugs such as carbamazepine or phenytoin) are used. In MSE, consciousness may be preserved, even when myoclonic jerks are frequent (a feature that should not lead to an erroneous diag­ nosis of a psychogenic disorder). The EEG usually shows epileptiform discharges at > 3 Hz. Typically, “fast” (greater than 3 Hz, but less than 6 Hz) generalized, symmetric, bifrontally predominant polyspike-and-waves and generalized spike-and-waves (GSWs) accompany the myoclonic jerks. With protracted duration or during drowsiness, lateralizing features on the EEG may be accentuated.

220  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.7.  EEG during myoclonic status epilepticus in Lafora body disease. There are frequent spike discharges with a generalized distribution, but they are often difficult to distinguish from muscle or movement artifact. © WO Tatum, 2013.

A

“secondary” group of MSE (not associated with the “primary” generalized epilepsies) is often the manifestation of serious, widespread underlying brain dysfunction, including the “progressive myoclonus epilepsies,” some of which are caused by storage diseases such as Lafora disease (Figure 10.7). These syndromes also include severe infant and pediatric conditions such as Lennox–Gastaut syndrome (LGS), in which “myoclonic astatic” and other seizures occur. MSE is more common in these epilepsy syndromes than in the GGEs. The EEG helps to distinguish the different types of MSE, providing information on the underlying brain function and often, a guide to optimal treatment. In the GGE syndromes, the EEG during myoclonic seizures typically demonstrates rapid (~4 Hz) generalized, frontally predominant polyspikes on a normal background, although in MSE, the background activity may become slow due to the frequent seizures or from the effect of medication. Sometimes, the spikes occur just before the clinical myoclonic jerk (Figure 10.7). In this case, the interictal EEG may show a normal background, with frequent spikes and polyspikes. In the second group of epilepsy syndromes, the spike and wave discharges are often slower (i.e., 2–2.5 Hz). Finally, in “symptomatic” types of MSE, there is usually an underlying encephalopathy reflected in diffuse slowing of all background activity. For example, in the MSE that follows anoxia, the EEG background is usually slow and of very low voltage, auguring very poorly for prognosis. In this case, the underlying pathophysiologic mechanism of MSE may differ significantly from that in the various epilepsy syndromes described earlier.

Myoclonic Status Epilepticus  221

FIGURE 10.8.  Serial tonic seizures in a patient with Lennox-Gastaut syndrome with altered cognition in between seizures. Note the brief post-ictal slowing following a burst of diffuse high frequency paroxysmal “fast” activity that was unresponsive to intravenous benzodiazepines.

T

onic SE (TSE) is rare in adults. It consists of maintenance of a tonic posture (typically with forward axial flexion, often with rigid arm extension and fast tremulousness), rather than of recurrent convulsions. Early in these seizures, the EEG often shows widespread fast activity or very rapid spikes. Sometimes, periods of background suppression or attenuation may be evident (Figure 10.8).

222  Chapter 10: The EEG in Status Epilepticus

NONCONVULSIVE STATUS EPILEPTICUS NCSE has many etiologies and protean manifestations. Some cases are characterized by impairment of alertness, attention, cognition, or behavior relative to baseline function. Observable signs may be absent or subtle, for example, with minimal automatisms or myoclonus as the only motor manifestation. NCSE may include primarily sensory symptoms, language deficits such as aphasia, psychiatric manifestations, and even autonomic features (e.g., in the Panayiotopoulos syndrome). There are many conditions other than epilepsy that can cause altered alertness or behavior, and NCSE can be very difficult to recognize; the EEG is crucial for diagnosis.

FIGURE 10.9.  Electrical status epilepticus in sleep (ESES) in a 9-year-old boy with Landau–Kleffner syndrome. The EEG shows generalized discharges that are not always rhythmic.

N

CSE among neonates and infants has particularly varied clinical manifestations and EEG patterns, remarkably different from those in children and adults. Some occur in the “epileptic encephalopathies” in which infants or young children have severe encephalopathies associated with very frequent interictal epileptiform discharges on the EEG, but in which the EEG pattern and the clinical seizures do not always correlate well. One example is electrical status epilepticus in sleep (ESES; Figure 10.9) in which clinical seizures may be absent from the epilepsy syndrome. ESES denotes a marked activation of epileptiform discharges during N3 sleep, often occupying the majority of the EEG and suggesting SE then. The waking EEG usually shows focal or occasionally generalized interictal epileptiform discharges, with the spike component usually more prominent than the slow wave.

Generalized Nonconvulsive Status Epilepticus  223

GENERALIZED NONCONVULSIVE STATUS EPILEPTICUS In early reports, NCSE was oversimplified into “absence” SE (if NCSE was associated with GSWs on the EEG), and “complex partial” SE (if there were focal epileptiform discharges or a clear focal onset noted clinically). Apparently-generalized NCSE, however, can arise from focal-onset NCSE with subsequently generalized discharges and no focal clinical features. In the 2015 International League Against Epilepsy (ILAE) classification of NCSE, there remain the (“truly”) generalized forms of NCSE and others with a presumed or proven focal onset.

FIGURE 10.10.  “Classic” 3-Hz spike-and-wave discharges during absence status epilepticus in a 23-yearold with drug-resistant juvenile absence epilepsy. © WO Tatum, 2020.

T

ruly generalized NCSE is that occurring in the GGEs. “True” or “typical” absence SE occurs in patients with absence epilepsy. Of all types of NCSE, this is a relatively small minority, but easily recognizable. Typical clinical manifestations include confusion, with occasional minimal motor abnormalities such as blinking or myoclonus, or brief, minimal automatisms, and often with some intermittent preservation of awareness. On EEG, very rhythmic and stereotyped generalized, bifrontally predominant 3-Hz spike-and-slowwaves occur in prolonged runs (Figure 10.10). The epileptiform discharges of absence SE may occur at a frequency of up to 4 Hz, especially at the onset. Rhythmic repetitive generalized epileptiform discharges on the EEG also occur in other GGEs and are not specific for an individual GGE syndrome.

224  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.11.  A 40-year-old woman with Jeavons syndrome, including eye blinking. She had absence seizures beginning in childhood, with myoclonic and generalized convulsive seizures in adolescence and frequent hospitalizations in adulthood. At the time, she was evaluated for confusion with eyelid myoclonia. The EEG shows prolonged bursts of nearly continuous generalized spikes and polyspikes. Courtesy of P Kaplan.

C

linically, the manifestations of SE in the GGEs include absence SE, but also tonic–clonic (or convulsive) SE; clonic SE; (very rare) tonic SE; and myoclonic forms of SE. Absence SE is the classic and most common form of GGE, followed by tonic–clonic SE, and then MSE (as in cases of juvenile myoclonic epilepsy). In all types of GGEs, SE is often precipitated by inadequate or inappropriate use of ASMs, or from behavioral triggers such as sleep deprivation or alcohol use. Absence SE or MSE may terminate in a generalized convulsion. Absence SE may last for several minutes, or, very rarely, up to months. MSE often last hours or even days, usually waxing and waning over its course. With both absence SE and MSE, the patient may be awake and appear alert, but detailed testing may demonstrate varying diminution of abstract thinking, personal awareness, and other higher cortical function. Some absence SE includes frequent eyelid myoclonia, with or without GSWs (e.g., in Jeavons syndrome; Figure 10.11).

Generalized Nonconvulsive Status Epilepticus  225

FIGURE 10.12.  Fixation-off absence SE in a 25-year-old woman with earlier absence epilepsy beginning at age 8. In adulthood, she had episodes of eye blinking, staring, and slurred, interrupted speech. [EO = eyes open; EC = eyes closed]. Courtesy of P Kaplan.

S

E is relatively infrequent in the GGEs, but there are several syndromes that can lead to SE (Figure 10.12). They may present as ongoing seizures lasting more than 5 min and often continue beyond 30 min in duration, the traditional definition of SE.

226  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.13.  Rhythmic 4- to 6 Hz-spike and slow wave activity of de novo absence status epilepticus (SE) in a 78-year-old with no clinical signs except impaired responsiveness. © WO Tatum, 2013.

R

elatively frequent episodes of NCSE have been reported in older patients without epilepsy (“de novo absence SE of late onset”), some in association with withdrawal of benzodiazepines, intercurrent infection, or after a convulsion. Some of these patients may have had features of a GGE earlier in life but have remained undiagnosed. In de novo absence SE, EEG discharges are often less rhythmic and regular than in typical absence SE (Figure 10.13), but they are usually generalized and rapid. De novo absence SE often responds quickly to benzodiazepine treatment, even in low doses.

Generalized Nonconvulsive Status Epilepticus  227

FIGURE 10.14.  Atypical absence status epilepticus (SE) with a slower (2.5 Hz) spike and wave pattern and intermixed fast activity. © WO Tatum, 2013.

A

typical absence SE (AASE) is infrequent and occurs primarily in children with  frequent seizures of mixed types. AASE may be manifested as a change in the baseline mental status or level of alertness. An underlying severe encephalopathy with frequent seizures, such as in LGS, is common. These conditions often include multiple seizure types (including tonic and myoclonic seizures), and severe developmental delay and neurocognitive impairment. In most patients, the baseline EEG shows a widespread slow background, with frequent, generalized or multifocal spike discharges. Ictally, there is often a characteristic “slow spike-and-wave” (SSW) repetitive discharge at 1- to 2.5-Hz (Figure 10.14). In early childhood, the SSW may occur in prolonged runs and suggest AASE, but without clear clinical signs of seizures and, rather, just a subtle decline in mental status. The SSW on EEG is key to the diagnosis of LGS, but this pattern may be lost in adults. AASE has less rhythmicity and symmetry than is seen in typical ASE.

228  Chapter 10: The EEG in Status Epilepticus

FOCAL-ONSET NCSE In the 2015 ILAE classification, focal-onset NCSE is, for the most part, divided into “focalonset NCSE with impaired consciousness (or awareness)” and “focal-onset NCSE with altered cognitive function.” The latter would include aphasic and hemianopic SE, among many possible manifestations. (“NCSE without impaired consciousness” would also include [the rare] purely sensory NCSE.)

FIGURE 10.15.  Focal-onset nonconvulsive status epilepticus (NCSE) with impaired awareness (complex partial SE) in a 21-year-old patient following an episode of encephalitis. The primary symptom was confusion. Epileptiform activity is prominent broadly over the right hemisphere. © WO Tatum, 2013.

F

ocal-onset NCSE with impaired consciousness or awareness. The persistence or recurrence of focal seizures with altered or impaired awareness is probably the most common form of NCSE in adults; most of these cases constitute what was called complex partial status epilepticus (CPSE). This focal NCSE usually begins with a focal-onset seizure and progresses to include an alteration in consciousness (i.e., responsiveness). Seizures may be continuous or may wax and wane, with fluctuating clinical manifestations. The EEG develops over time, with repetitive focal-onset seizures in which epileptiform discharges may merge gradually to produce continuous focal seizure activity (Figure 10.15). During a seizure, the discharges tend to increase in voltage and later slow in frequency. Many seizures have subsequent spread to involve a greater area of the cortex, and some spread to generalized epileptiform discharges or potentially even to clinical convulsions. Between discrete seizures, there may be repetitive or periodic focal discharges and focal slowing before the next seizure begins.

Focal-Onset NCSE  229

FIGURE 10.16.  A patient with ongoing left occipital SE manifested by persistent visual hallucinations of colored spheres in the right visual field, without impairment of consciousness. © WO Tatum, 2013.

F

ocal-onset NCSE with altered cognitive function: Focal-onset NCSE has a variety of clinical features that may include sensory, visual, auditory, or olfactory symptoms that may appear to represent nonepileptic hallucinations (Figure 10.16). Manifestations may also be autonomic, psychic, or cognitive, with impairment of attention, language, mood, or behavior. The clinical deficits may resemble those of other focal deficits such as stroke or mass lesions. Examples of focal-onset NCSE include aphasic status (probably the most common NCSE mistaken for strokes), as well as hemianopic, dysmnesic, and neglect ­ ­syndromes (Figures 10.17–10.18). (Continued )

230  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.17.  Aphasic status epilepticus. Focal rhythmic left temporo–parietal discharges, with clinical manifestations of a posterior aphasia, with fluency but impaired comprehension and “jumbled speech.” © WO Tatum, 2020.

FIGURE 10.18.  78-year-old man with amyloid angiopathy and focal status epilepticus (SE) precipitated by antiseizure medication (ASM) taper in the epilepsy monitoring unit (EMU). Clinical manifestations included retained awareness (“focal aware”) but right hemianopia, right-sided neglect, and right body pain. EEG shows widespread rhythmic slowing over the left hemisphere with superimposed rhythmic ictal fast rhythms. © WO Tatum, 2020.

Electrographic Status Epilepticus  231

ELECTROGRAPHIC STATUS EPILEPTICUS Probably the most important NCSE to recognize is the continuation of SE after the motor manifestations of GCSE cease—likely the most common NCSE in hospital settings. (This entity could also be discussed above, under the topic of generalized convulsive SE, as its continuation.) This NCSE is often secondarily generalized from a focal onset, sometimes an underlying structural lesion (Figure 10.19). Others are due to severe metabolic derangements or infection; the causes are those of the preceding GCSE.

FIGURE 10.19.  Continuous 3-Hz generalized epileptiform discharges in an 83-year-old woman found to be stuporous after admission to the hospital, but without a clear cause. © F Drislane, 2009. Courtesy of the author.

T

he clinical manifestation can be simply an impairment of responsiveness. The generalized epileptiform discharges on EEG may not indicate a clear focal seizure origin, even when there is a focal lesion. In others, there may be a deeper focus and rapid generalization. Detection of this type of SE is one of the primary goals of continuous EEG monitoring in the ICU, where nonconvulsive seizures and NCSE are captured frequently on EEG without evident clinical signs.

232  Chapter 10: The EEG in Status Epilepticus

FIGURE 10.20.  Status epilepticus (SE) due to anoxia in a 50-year-old following cardiac arrest and hypothermia. The suppressed background between discharges augurs for an unfavorable prognosis. Courtesy of P Kaplan.

T

his ongoing electrographic seizure activity, often following the apparent termination of prior generalized convulsions or GCSE (usually with minimal or no motor convulsive activity), may be termed “electrographic status epilepticus” (ESE). It typically occurs in the setting of severe medical illness such as sepsis, anoxia, or severe metabolic derangements and is the most common type of NCSE in ICUs (Figure 10.20). It has been referred to as “status epilepticus in coma,” but not all such patients are comatose; most have severe medical and neurologic illnesses. ESE is often found in patients with severely abnormal mental status but in whom the SE was unsuspected before the EEG.

Diagnoses  233

DIAGNOSES

FIGURE 10.21.  A 41-year-old man with a generalized convulsion at age 34 was treated with valproate. The EEG was normal when the patient was asymptomatic. At a later clinic visit, he presented with chin tremor only. Courtesy of P Kaplan.

O

ften, the most difficult problem in the management of SE is arriving at the correct diagnosis. Interpreting the EEG in the context of the clinical situation can be difficult. Some patterns (e.g., LPDs) have controversial clinical implications. There is also incomplete agreement, even among experienced electroencephalographers, concerning which EEGs are diagnostic of ongoing electrographic seizures and NCSE. The following cases illustrate potential pitfalls in interpreting the EEG, especially in critically ill patients ­(Figures 10.21– 10.24) as discussed on the following pages. The first EEG (Figure 10.21) shows bursts of generalized spike- and polyspike-andslow-wave discharges, supporting a diagnosis of NCSE. ASMs led to an improvement in both the EEG and the clinical condition. The second (Figure 10.22) has frequent repetitive ­epileptiform discharges on the EEG, also suggesting NCSE. The spikes are regular, very brief, and have a generalized distribution typical for GGE. Figure 10.23 shows GPDs—with a left hemisphere predominance, repeating at a frequency of 4 Hz lasting at least 0.5s that share many characteristics with seizures (sharp morphology and/or evolution in frequency or morphology and/or field of spread), but last less than 10 seconds (Figure 11.15). They were originally described in neonates with seizures, though they may also occur in critically ill adults where they are almost always associated with clinically relevant (i.e., longer) electrographic seizures. They are also seen in awake patients with refractory epilepsy.

Periodic and Rhythmic Discharges  259

FIGURE 11.16.  Continuous generalized rhythmic delta activity (GRDA) with superimposed low-voltage beta activity (GRDA+F). The “extreme delta brush” pattern was present in a 28-year-old female with antiNMDAR encephalitis.

E

xtreme delta brushes consist of prominent and prolonged (minutes to hours) runs of rhythmic delta activity (RDA), typically generalized and frontally maximal, with superimposed low-voltage beta or gamma (30–40 Hz) activity, classically with a burst of fast activity overriding the delta waves (Figure 11.16). These complexes resemble the delta brushes of neonates, but are continuous or nearly so. They have been described primarily (but not exclusively) in anti-NMDAR encephalitis, especially with more severe cases. It is unclear if these complexes represent an ictal or interictal pattern, though most believe they are not ictal; eliminating them with treatment can be difficult and has not led to clinical improvement in most cases.

260  Chapter 11: ICU EEG

SEIZURES

FIGURE 11.17.  This is a 60-second epoch of EEG showing a left hemispheric seizure in an 81-year-old male with Herpes Simplex encephalitis 1 (HSE1). Note the presence of left hemisphere lateralized periodic discharges (LPDs) and the ictal onset pattern consisting of sharply contoured 1-Hz delta activity with admixed faster activity and slow evolution. Evolution is best seen at a display speed that is faster than the conventional speed.

S

eizures occur in 10% to 20% of critically ill patients undergoing CEEG. The majority of these (~75%) are nonconvulsive and require EEG to be detected. Risk factors include acute brain injury, sepsis, renal failure, coma, young age, and prior clinical seizures. Subtle motor manifestations may occur, including eye deviation, nystagmus, face or finger twitching. Ictal patterns on EEG that are seen in the ICU often differ from seizures in alert patients with chronic epilepsy. NCSz in the critically ill may have less distinct EEG onset and offset, slower evolution, slower maximum frequencies, and typically occur on an abnormal background. All of these features make them harder to recognize. Evolving or fluctuating RDA, PDs, and spike-and-wave complexes may all represent NCSz in some cases. Using a compressed time scale during EEG may allow better appreciation of the evolution of a slow rhythmic discharge (Figure 11.17).

Seizures  261

FIGURE 11.18.  The EEG of a 61-year-old male who suffered a right frontal hemorrhage. The patient was obtunded and had intermittent twitching of the left hand. Note the presence of evolving 5-Hz rhythmic poly-spike-and-wave complexes that are most prominent over the right parasagittal region.

A

lmost half of the seizures encountered in the critically ill are SE (i.e., lasting at least 10 continuous minutes or at least 50% of the record). Like seizures, SE is most often nonconvulsive in the ICU setting, occurring in patients with a severely impaired level of consciousness (Figure 11.18). Minor clinical manifestations, however, may be seen, including nystagmus, eye deviation, autonomic changes, repetitive eye opening, and subtle face or limb myoclonus.

262  Chapter 11: ICU EEG

(A) FIGURE 11.19A–C.  The EEG of a 25-year-old obtunded male with severe traumatic brain injury. His CT scan revealed bilateral frontal hemorrhagic contusions, more marked in the right hemisphere. Note the presence of high-amplitude continuous rhythmic 2-Hz sharp-and-slow-wave complexes (A). There was no clinical sign of ictal activity. This pattern was ultimately felt to represent generalized NCSE based on clinical factors and response to IV anti-seizure medication. The patient received two boluses of lorazepam (total dose: 5 mg) and a load of lacosamide with EEG improvement (B and C). Although he remained disoriented and combative, he progressively regained consciousness over the next 24 hours. (Continued)

C

riteria for NCSE have been proposed and validated as much as p ­ ossible without a gold standard for comparison (Table 11.3). Some patterns remain difficult to categorize and are thought to belong to a continuum that reflects a dynamic state between interictal and ictal activity, known as the ictal–interictal continuum (Figure 11.19A–C). They take into account the morphology, frequency, and spatial and temporal evolution of the discharges, and the presence of clinical correlates. Even in the absence of these clinical or EEG features, the pattern might still represent NCSE. (Continued)

Seizures  263

(B)

(C) (Continued)

264  Chapter 11: ICU EEG TABLE 11.3.  Salzburg criteria for NCSE in patients without known epileptic encephalopathy Epileptic discharges at frequency >2.5 Hza OR EDs ≤ 2.5 Hz or rhythmic delta/theta activity (>0.5 Hz) AND one of the following:   Typical spatiotemporal evolutiona   Subtle clinical ictal phenomena during the EEG patternsa   EEG and clinical improvement after IV ASM Note: aFor a discharge to qualify as NCSE, it should last >10 min but these criteria should be fulfilled only for 10 seconds. ASM, anti-seizure medication; IV, intravenous: NCSE, nonconvulsive status epilepticus. Source: Adapted from Beniczky S, Hirsch LJ, Kaplan PW, et al. Unified EEG terminology and criteria for nonconvulsive status epilepticus. Epilepsia. September 2013;54(suppl 6):28–29. https://doi.org/10.1111/ epi.12270.

In these difficult cases, the administration of a fast-acting ASM such as a benzodiazepine may help clarify the ictal nature; using a nonsedating IV ASM might make it easier to detect clinical improvement. Electrographic and clinical improvements are both required to confirm the diagnosis of NCSE. A negative test devoid of electro-clinical resolution is inconclusive and does not exclude the diagnosis of NCSE; this is usually referred to as “possible NCSE.” Clinical improvement may take up to 24 hours to be noticed.

Seizures  265

FIGURE 11.20A.  Quantitative EEG in a 50-year-old female with invasive aspergillosis. Cyclic seizures are easily noticed on the rhythmicity spectrogram (rows 2 and 3; yellow background) by the bursts of activity in blue-green, on the asymmetry index in row 4 (upward deflection = right and downward deflection = left), relative asymmetry spectrogram (row 5) showing relative power (red = right and blue = left), and amplitudeintegrated EEG (last row) as increases in amplitude. The duration of the epoch is 2 hours. The seizure periodicity is 20 minutes in the right hemisphere and 30 minutes in the left hemisphere.

FIGURE 11.20B.   This 20-second epoch of raw EEG data shows a right temporal seizure (arrowhead). Note the presence of a 1-second run of brief potentially ictal rhythmic discharges (BIRDs) 2 seconds before the seizure onset (box); and independent ongoing left hemispheric lateralized periodic discharges (LPDs) at 0.3 Hz.

C

yclic seizures are an uncommon form of SE encountered in critically ill patients in which discrete, usually brief, seizures occur with a nearly regular periodicity every 2- to 20-minute interval (Figure 11.20A–B). The reasons for a cyclic pattern are poorly understood, but may include the transient efficacy of endogenous inhibitory mechanisms involved in seizure termination.

266  Chapter 11: ICU EEG

REACTIVITY AND STIMULUS-INDUCED PATTERNS EEG reactivity is defined as a reproducible change in the EEG background upon patient’s stimulation. This may comprise changes in voltage or frequency, including attenuation of activity (Figure 11.21A), appearance, or disappearance of periodic or rhythmic discharges, seizures, or even bursts. Reactivity testing should be performed in all comatose patients undergoing EEG, as it carries important prognostic information, especially after anoxic and traumatic brain injury. Multiple types of stimulus, including auditory (calling patient’s name), tactile (shaking, nostril tickling), noxious (trapezius squeeze, nail-bed pressure), and visual (passive eye opening) stimulus, should be used.

Reactivity and Stimulus-Induced Patterns  267

FIGURE 11.21A.  This page demonstrates reactivity to auditory stimulation (clapping hands and loudly calling patient’s name; vertical bar). There is attenuation of the EEG background during 2 seconds followed by the return of background activity.

FIGURE 11.21B.  This 40-second page of compressed EEG is taken from a 60-year-old female with subarachnoid hemorrhage and left middle cerebral artery (MCA) vasospasm. Stimulation by nostril tickling (onset of stimulation indicated by arrowhead) is followed by the appearance of right posterior LPDs (onset indicated by double arrowhead).

I

n as many as 20% of comatose or stuporous patients undergoing CEEG, periodic, rhythmic, or ictal discharges are consistently elicited by various kinds of nonnoxious (noise, nursing) or noxious (nostril tickling) stimuli (known as “stimulus-induced” patterns). As a group, these patterns are referred to as SIRPIDs (stimulus-induced rhythmic, periodic, or ictal discharges). Conversely, rhythmic and periodic patterns may also disappear upon arousal following stimulation (“stimulus-terminated”), though this is much less common (Figure 11.21). The exact mechanism of stimulus sensitivity remains unclear, but probably implicates the combination of thalamocortical activation and cortical hyperexcitability.

268  Chapter 11: ICU EEG

FIGURE 11.22.  This 60-second epoch of compressed EEG occurred in a 51-year-old female with a history of generalized epilepsy of unknown origin who was admitted for SE. Stimulation (due to a nurse replacing her intravenous [IV] line is indicated by an arrowhead) consistently elicited brief electro-clinical seizures (double arrowhead) manifest as runs of frontally predominant generalized fast activity with clinical features of myoclonic jerks mainly of the proximal lower limbs. This example represents stimulus-induced electrographic seizure with a clinical correlate.

D

espite the classic EEG teaching that patterns induced by arousal are usually unrelated to seizures, it is clear that alerting stimuli are able to commonly elicit or exacerbate ictal discharges, including (less commonly) ones that are associated with clinical manifestations (Figure 11.22). The presence of a clinical correlate likely depends on the involvement of the motor pathways and the ability of a discharge to propagate caudally from the cortex.

Brain Monitoring With QEEG  269

BRAIN MONITORING WITH QEEG

FIGURE 11.23.  This EEG represents a 63-year-old male with subarachnoid hemorrhage (SAH) from a ruptured right anterior communicating artery (ACA) aneurysm. 10 seconds of EEG on day 2 after bleeding: the background consists of diffused symmetrical theta activity.

A

decrease in brain perfusion is associated with progressive EEG changes that reflect the degree of ischemia:

Cerebral blood flow (mL/[100 g/min])

EEG correlate

25–35

Attenuation of alpha and beta activities

18–25

Polymorphic theta slowing

12–18

Polymorphic delta slowing

500 msec; (b) typically first seen before dropout of the dominant posterior rhythm (socalled DPR); (c) continue to occur during NREM 1 and usually disappear before sleep spindles and K-complexes of N2 sleep appear; (d) prevalence of them progressively decreases across NREM-REM sleep cycles of a night of sleep.16 Figure 12.3 shows SEMs in N1.

Which Biophysiologic Signals are Recorded in Level 1 Polysomnography and Why  285

FIGURE 12.3.  A 30-second epoch of polysomnogram (PSG) during NREM 1 sleep slow eye movements, vertex wave, and an EEG background of low voltage mixed frequencies.

REMs are: (a) sharply peaked eye movements with an initial deflection lasting 85%; falsely high if it is 50% of the 30-second epoch contains a sinusoidal 8- to 13-Hz alpha rhythm and/or 0.5- to 2-Hz eye blinks, REMs associated with normal or high chin EMG tone, reading eye movements (Figure 12.12). The chin EMG during stage W is of variable amplitude but usually higher awake then asleep. Epochs of W are scored until sleep onset occurs. Sleep onset is the start of the first 30-second of any stage other than W (and usually N1).

Scoring Stage NREM 1 (N1) Decreased eye blinking is usually the first sign of drowsiness in humans, soon followed by slow roving eye movements (even though the DPR may be preserved), then waning of the posterior alpha rhythm which is usually replaced by low-voltage predominantly 4- to 7-Hz diffuse theta activity (termed low-amplitude, mixed-frequency [LAMF] Figure 12.13). The physiological changes of progressive sleepiness are summarized in Figure 12.14. Sleep onset in most people is N1. N1 is a transition from wakefulness to sleep. Patients are easily awakened from N1 often feel they were not sleeping. Sleep onset is best identified by a dropout of the DPR (more often preceded or accompanied by SEMs). N1 is scored in subjects who generate a DPR when the posterior rhythm is attenuated or replaced by LAMF activity >50% of the epoch (Figure 12.15).

FIGURE 12.12.  A 30-second of polysomnogram (PSG) recorded in wakefulness (stage W) which shows alpha activity and eye movements. The alpha frequencies appear more diffuse perhaps because alpha activity is contained in the mastoids, then referenced to frontal and central derivations. Note eye blinks and saccades, and preserved chin electromyogram (EMG) tone which characterize stage W.

296  Chapter 12: Polysomnography

FIGURE 12.13.  A 30-second epochs of polysomnogram (PSG) showing the transition from wakefulness to NREM 1 sleep. A dominant posterior rhythm is observed in the occipital derivations (O1-M2, O2-M1) which drops out at the last quarter of the epoch with the appearance of an EEG background of low-voltage mixed frequencies, and slow eye movements (SEMs). Because the dropout of the posterior dominant rhythm occurs in the later part of the epoch it is scored as stage W.

ALERT Mixed alpha & beta activity Eye movements Normal muscle tone

SLEEPINESS Increased alpha amplitude Decreased alpha frequency Blinking

RESTING Increased alpha activity Blinking

SLEEP ONSET Alpha activity disappears Vertex waves may occur Slow eye movements Blinking stops

FIGURE 12.14.  Physiological changes of progressive sleepiness.

However, as many as 10% of normal healthy adults have little or no DPR (and called “poor alpha generators”). In them, N1 is scored when either: (a) SEMs appear; (b) the waking EEG activity of 4 to 7 Hz slows by ≥1 Hz; or (c) vertex sharp waves appear. SEMs have an initial deflection which usually lasts >500 seconds. SEMs can be seen in W with eyes closed, N1, and may linger into N2. They are most prominent at sleep onset. SEMs, although usually present, are not needed to score N1, nor are vertex waves. Vertex sharp waves (V waves): (a) last 50% of a 30-second epoch.

REMs are sharply peaked, irregular, conjugate eye movements with an initial deflection usually lasting 200 µV) 4- to 6-Hz theta activity over the frontocentral or central regions characterizes drowsiness in children 8 months to 3 years. In children 3 years or older, drowsiness is characterized by 1- to 2-Hz slowing of the DPR and/or the DPR becomes diffusely distributed then is gradually replaced by low voltage mixed frequency activity.

Scoring Sleep/Wake States in Children 2 Months to 18 Years of Age  301

(A)

(B) Stage W age 2 years

(C) Stage W age 11 FIGURES 12.19.  (A), (B), and (C) show representative 30-second epochs of stage W in children of varying ages: (A) recorded in an infant age 2 months; (B) a child age 2; and (C) a child age 11.

302  Chapter 12: Polysomnography

FIGURE 12.20.  A 30-second epoch of polysomnogram (PSG) recorded in a child age 8 during NREM 1 sleep showing a burst of hypnagogic hypersynchrony, slow eye movements, and an EEG background of low-voltage mixed frequency activity.

Scoring NREM 1 Sleep in Children In pediatric subjects who generate a DPR score, N1 is scored if the DPR is attenuated or replaced by LAMF activity for >50% of the entire 30-second epoch. In those who do not generate a DPR, N1 is scored beginning with the earliest of any of the following: (a) 4- to 7-Hz activity with slowing of background frequencies by ≥1- to 2-Hz from those of stage W; (b) SEMs; (c) vertex sharp waves; (d) rhythmic anterior theta activity; (e) hypnagogic hypersynchrony; and/or (f) diffuse or occipital predominant high amplitude rhythmic 3- to 5-Hz activity. Hypnagogic hypersynchrony (Figure 12.20) is a distinctive pattern of drowsiness and N1 in children is characterized by paroxysmal bursts (≤2 seconds) or runs (lasting >2 seconds) of diffuse high-amplitude sinusoidal 75 to 350 µV 3- to 4.5-Hz waves usually widely distributed but often maximal over central, frontal, or frontocentral regions. Seen in 30% of infants 3 months postterm, 95% ages 6 to 8 months, hypnagogic hypersynchrony becomes less prevalent after ages 4 to 5 years, in only 10% of healthy children age 11, and rare after age 12. Rhythmic anterior theta activity is more often seen in adolescents when drowsy and characterized by runs of rhythmic 5- to 7-Hz activity maximal over frontal or frontocentral regions (Figure 12.21).

Recognizing and Scoring NREM 2 Sleep in Children Rudimentary sleep spindles first appear in infants between 43 to 45 weeks postmenstrual age (PMA) and are most consistently present after 46 weeks PMA. Sleep spindles in infants first appear over the midline central (CZ) electrodes, and are often asynchronous until 1 to 2 years of age. At 6 months of age 50% of sleep spindles are synchronous over the left and right central EEG derivations (C3-M2, C4-M1), 70% at 9 and 12 months. Asynchronous sleep spindles are considered abnormal if present after 2 years of age. A particularly striking age-related sleep spindle pattern is seen in infants 3 to 4 months PMA in which sleep spindles often last 5 to 8 seconds or longer (Figure 12.22). Observing such long-lasting sleep spindles usually tells you the infant is 3 to 4 months PMA. Eighty percent of children 20% of a 30-second epoch) is usually present in infants 5 to 6 months term. Infant sleep researchers report they could score NREM 3 as early as 3 months term, but most often 4 to 4.5 months postterm. SWA in children is often 100 to 400 µV and typically maximal over the frontal regions (although in many equally present over the frontal and central EEG derivations). Figures 12.25A and 12.25B show representative samples of SWA in NREM 3 in children.

Scoring REM Sleep in Children All of the brain neuronal wiring for REM sleep is present in healthy infants by 32 weeks gestational age (GA).40 Eighty percent of sleep time in premature infants is spent in REM sleep, 50% in infants at term. Given this, sleep onset in infants ≤3 months postterm is typically R sleep. The LVMF EEG activity of stage R in infants and children resembles adults although the dominant frequencies increase with age. At 7 weeks postterm LVMF EEG in REM sleep is predominantly 3 Hz, 4 to 5 Hz with bursts of sawtooth waves at 5 months, and 4 to 6 Hz at 9 months. Prolonged runs or bursts of often notched 5- to 7-Hz theta activity are seen in REM sleep between ages 1 to 5 years.

Scoring Sleep/Wake States in Children 2 Months to 18 Years of Age  305

(A)

(B) FIGURES 12.25.  A and B show representative examples of stage N3 in two children: (A) age 11 months; and (B) age 11. Note sleep spindles are intermixed with high-voltage slow wave activity. Sleep spindles can linger into NREM 3 sleep.

By ages 5 to 10 years, the EEG during REM sleep resembles that of adults, and characterized by LVMF activity with bursts of 4- to 6-per-second sawtooth waves maximal over the central regions and most prevalent during phasic REM and maximal over midline central (Cz) electrode. Figures 12.26A to C show how the EEG of REM sleep changes with age in children. The AASM manual emphasizes that sleep spindles may be present during the first or second REM sleep periods but if all other behavioral correlates of REM sleep are present should not prevent scoring an epoch as stage R.

306  Chapter 12: Polysomnography

(A)

(B)

(C) FIGURE 12.26.  (A–C) Representative examples of Stage rapid eye movement (REM) sleep in children of varying ages. Shown are three 30-second epochs of polysomnogram (PSG) recorded in children age 18 months (A), 8 years (B), and 11 years (C). Note in all, REMs, REM sleep with atonia, and a lowvoltage mixed frequency EEG background. Respiration is irregular and often contains brief central apneas especially when REMs are frequent.

Sleep Architecture in a Prototypical Young Adult  307

SLEEP ARCHITECTURE IN A PROTOTYPICAL YOUNG ADULT

Awake REM N1 N2 N3 1

2

3

4 Hours

5

6

7

8

FIGURE 12.27.  Normal healthy well-slept young adults exhibit alternating cycles of NREM and rapid eye movement (REM) sleep across a prototypic night of sleep (Figure 12.27). Note the evolution of different stages of sleep (sleep architecture) during the first cycle of NREM and REM sleep: a sequential descent from lighter to deeper stages of NREM sleep (stages N1, N2, then N3), then ascending (stages N3, N2) and ending with a brief period of REM sleep. A cycle of NREM followed by REM sleep is called a “sleep cycle.” Four to 5 sleep cycles are usually observed across a night of sleep in normal well-slept young adults.

Normal healthy well-slept young adults exhibit alternating cycles of NREM and REM sleep across a prototypic night of sleep (Figure 12.27). Note the evolution of different stages of sleep (sleep architecture) during the first cycle of NREM and REM sleep: a sequential descent from lighter to deeper stages of NREM sleep (stages N1, N2, then N3), then ascending (stages N3, N2) and ending with a brief period of REM sleep. A cycle of NREM followed by REM sleep is called a “sleep cycle.” Four to five sleep cycles are usually observed across a night of sleep in normal well-slept young adults. A night of sleep usually begins with 70 to 80 minutes of NREM then 10 minutes of REM sleep. N1 usually only lasts a few minutes in the first sleep cycle. N1 represents a transitional state to more restorative N2 and N3 sleep. N2 typically last 10 to 25 minutes soon followed by a gradual buildup of high-amplitude SWA (>75 μV 0.5–2.0 Hz). Within minutes of N3 onset, high-voltage SWA occupies most of a 30-second epoch. A cluster of body movements often signal an individual ascending to lighter stages of NREM sleep. During the ascent from deep to lighter NREM sleep only 1 to 2 minutes are often spent in what used to be called Stage 3 NREM sleep, 5 to 10 minutes in N2 (then associated with frequent body movements) before diving into the first period of REM sleep. The first period of REM sleep of a night is often short, lasting only 1- to 5-minutes, and often lacks requisite REMs, making it particularly hard to identify and score. Sleep cycles typically last a mean of 90 to 110 minutes, the first usually shorter (70 to 100 minutes), while later cycles range from 90 to 120 minutes. The first REM sleep usually occurs 70 to 90 minutes (range 70–110 minutes) after sleep onset. N3 is maximal during the first two sleep cycles with a smaller percentage of N3 in the second sleep cycle. Early morning sleep alternates between REM and N2. N3 dominates the first third of a night of sleep, REM sleep the last third. A healthy young adult spends about 75% to 80% of Total Sleep Time (TST) in NREM sleep. WASO should occupy less than 5% of the TST, 2% to 5% stage N1, 45% to 55% N2, 13% to 23% N3, leaving 20% to 25% for time spent in REM sleep.

308  Chapter 12: Polysomnography

SCORING AROUSALS IN A LEVEL 1 PSG USING AASM SCORING RULES An arousal from NREM sleep at any age is scored in a level 1 PSG when there is an abrupt EEG frequency shift (which can include alpha, theta, or beta activity >16 Hz but not sleep spindles) which lasts ≥3 seconds and ≥10 seconds of stable sleep precedes it.16 A concurrent increase in chin EMG muscle activity lasting ≥1 second is additionally required to score an arousal during REM sleep.16 We recognize and score arousals in a PSG first by observing for EEG changes in the frontal, central, and/or occipital derivations. Observing respiratory events, limb movements, and increases in heart rate in respiratory, EKG, and EMG channels preceding the EEG arousal help recognize them. However, increases in heart rate, respiratory effort, airflow, or limb movements cannot be used to score EEG arousals without ≥3 second-EEG shift. An arousal can be scored in an epoch of W if it is recorded between Lights Out and Lights On and 10 seconds of sleep precedes it. An arousal is scored and tallied even when it heralds a shift to wakefulness; the arousal counted and the following epoch scored as W. Arousals may or may not cause a shift from one sleep stage to another (called a stage shift). Figures 12.28A to 12.28C show arousals from NREM 2, NREM 3, and REM sleep. Arousals in a PSG from Lights Out to Lights On are tallied, and divided by the TST to determine an arousal index (mean number of arousals per hour of sleep). Moreover, we may report whether the arousal was related to a respiratory event or a PLM or without particular cause (spontaneous).

Scoring Arousals in a Level 1 PSG Using AASM Scoring Rules   309

(A)

(B)

(C) FIGURES 12.28.  (A–C) Shows arousal from NREM, rapid eye movement (REM), and during W. 30-second epochs of polysomnogram (PSG) showing arousals from NREM 2 (A), NREM 3 (B), and rapid eye movement (REM) sleep (C). Note the abrupt shift in EEG frequencies which lasts ≥3 seconds. A concurrent increase in chin electromyogram (EMG) muscle activity lasting ≥1 second is additionally required to score an arousal during REM sleep.

310  Chapter 12: Polysomnography

SCORING RESPIRATORY EVENTS Most level 1 PSG are ordered to evaluate for SDB. Respiratory events of interest in a PSG are apneas, hypopneas, RERAs, Cheyne–Stokes respiration, periodic breathing, sleep-related hypoxemia, or hypoventilation. As mentioned earlier, the Scoring Manual specifies: (a) which sensor is preferentially recommended to identify a particular respiratory event and (b) which alternative sensor(s) can be used if the recommended sensor is unreadable or unreliable (Table 12.6). The most important to remember are: (a) in a diagnostic study, the oronasal sensor is recommended to identify apneas, and the NP sensor to identify hypopneas or RERAs; (b) when titrating PAP therapy, the PAP device flow signal is the preferred sensor to identify apneas, hypopneas, and RERAs. There are relatively minor differences in scoring respiratory events in adults and children (particularly for central apneas). Understanding these will help understand the Scoring Manual rules for scoring respiratory events.

Rules for Scoring Apneas An apnea is a ≥90% fall in airflow through the nose and mouth which lasts at least 10 seconds in an adult, and two breaths in a child. The AASM recommends scoring an apnea in an adult when the peak signal excursion drops by ≥90% of preevent baseline for ≥10 seconds in the oronasal thermal sensor (during a diagnostic study), the PAP device flow (during PAP titration), or an alternative apnea sensor in a diagnostic study if the oronasal sensor fails. An apnea in a child is scored largely the same: a ≥90% decrease in inspiratory effort but the event duration is for ≥2 breaths. We classify apneas as obstructive, central, or mixed in type based upon the respiratory effort. We score an apnea as central if it meets apnea criteria and is associated with absent inspiratory effort throughout the entire period of absent airflow. An apnea is scored as obstructive if it meets apnea criteria and is associated with continuous or increased inspiratory effort throughout the entire period of absent airflow. An apnea is scored as mixed if it meets apnea criteria and is associated with absent inspiratory effort during the initial portion of the event, followed by resumption of inspiratory effort in the second part of the event. Figures 12.29 A-C illustrate these. An apnea in an adult does not need to cause any oxygen desaturation or arousal to be scored. If part of an apnea is a hypopnea, the entire event can be scored as an apnea as long as ≥10 seconds of it meets apnea criteria. As little as one obstructive breath can be observed in a mixed apnea to score it as mixed. Obstructive or mixed apneas in children also do not need to cause a desaturation or arousal to be scored. However, rules for scoring central apneas in children are more complicated. A central apnea in a child is most often scored if it lasts ≥2 breaths and is associated with an arousal and/or a ≥3% arterial oxygen desaturation (Figure 12.30). A central apnea which lasts ≥ 20 seconds in a child is sufficiently abnormal (and uncommon) that it can be scored without requiring either a desaturation or arousal. Lastly, central apneas lasting ≥2 breaths in infants less than 1-year-old can be scored if they cause bradycardia (10 seconds, cause oxygen desaturations to 83% to 85%, arousals, and are accompanied by snoring.

FIGURE 12.32.  A 30-second epoch of polysomnogram (PSG) recorded in a 2-month-old infant during rapid eye movement (REM)/active sleep. Two central hypopneas are observed. Note the absence of snoring, paradoxical breathing, or flow limitation, which distinguish central from obstructive hypopneas.

We score a hypopnea in a child if the peak signal excursions decrease by ≥30% (but 50 mm Hg measured by either arterial pCO2 or surrogate etCO2 or tcCO2. Cheyne–Stokes breathing is scored if there are: (a) ≥3 consecutive central apneas and/or c­ entral hypopneas separated by a crescendo and decrescendo change in breathing amplitude with a cycle length ≥40 seconds and (b) ≥5 cycles of central apneas and/or central hypopneas per hour of sleep associated with the crescendo–decrescendo breathing recorded ≥2 hours of monitoring. Cycle length is the time from the beginning of a central apnea to the end of the next crescendo–decrescendo respiratory phase. Central apneas and hypopneas that occur in a run of Cheyne–Stokes breathing should be scored as individual apneas and hypopneas as well. Figure 12.35 shows Cheyne–Stokes breathing in an adult. Periodic breathing is scored if there are ≥3 episodes of central apnea lasting >3 seconds separated by ≤20 seconds of normal breathing. Central apneas that occur within a run of periodic breathing should be scored as individual apneas as well. Figure 12.36 shows periodic breathing in an infant during NREM sleep.

FIGURE 12.35.  A 5-minute epoch of polysomnogram (PSG) recorded in an older adult with congestive heart failure showing Cheyne–Stokes breathing. Note the crescendo–decrescendo pattern of respiration alternating with central apneas. The arousal occurs during the crescendo breathing.

316  Chapter 12: Polysomnography

FIGURE 12.36.  A 90-second epoch of polysomnogram (PSG) recorded in an infant showing periodic breathing. Periodic breathing is scored if there are ≥3 episodes of central apnea lasting >3 seconds separated by ≤20 seconds of normal breathing.

Scoring PLMs in Level 1 PSG  317

SCORING PLMs IN LEVEL 1 PSG As mentioned earlier, we routinely record submental and lower-extremity anterior tibialis surface EMG in level 1 PSG. When recording patients for suspected RBD, we additionally record upper-extremity FDS EMG. We rarely record masseter EMG in patients with sleep bruxism. PLMs in sleep are scored from the tibialis anterior EMG signal. A LM event is scored when the EMG voltage increases by ≥8 µV above the resting EMG for ≥0.5 seconds but ≤10 seconds (Figure 12.37). The onset of a LM event is defined as the point at which there is an­ 8 µV increase in the EMG voltage above the resting EMG. The offset of a LM event is defined as a start of a period lasting ≥0.5 seconds during which the EMG does not exceed ≥2 µV above baseline EMG. LM events are scored as PLMs (called a PLM series) if they occur as a sequence of ≥4 in a row provided each LM occurs within 5 to 90 seconds of its neighbor. Period length (interjerk interval) between LMs is counted from the onset of one LM to the onset of the other LM in sec. LMs on two legs separated by