Epileptic Seizures and the EEG

Measurement, Models, Detection and Prediction

Andrea Varsavsky, Iven Mareels, Mark Cook

• Publisher: CRC Press
• Year: 2016
• ISBN:
  9781000218923
• Edition: 1
• Topic: Medizin
• Subtopic: Biochemie in der Medizin

A study of epilepsy from an engineering perspective, this volume begins by summarizing the physiology and the fundamental ideas behind the measurement, analysis and modeling of the epileptic brain. It introduces the EEG and provides an explanation of the type of brain activity likely to register in EEG measurements, offering an overview of how these EEG records are and have been analyzed in the past. The book focuses on the problem of seizure detection and surveys the physiologically based dynamic models of brain activity. Finally, it addresses the fundamental question: can seizures be predicted? Based on the authors' extensive research, the book concludes by exploring a range of future possibilities in seizure prediction.

1

Introduction

The brain is a very complex system composed of billions of interconnected neurons. As observers we understand the brain system from the measurements or signals we obtain from it. If the spatial scale at which measurements are made is very small, the measurement could pertain to a single neural cell. In contrast, at larger scales the measurement pertains to large collection of neurons. In Figure 1.1 the issue of spatial scale is illustrated.

In this book we concentrate on the EEG as the measurement that is used to acquire a signal. The EEG records the time-evolving voltages generated by brain activity, and is described in Section 1.2. However the measured signal is not necessarily the real signal generated by the brain, but its projection onto the recording equipment. Thus the measurement process is a system itself, with its inputs arising from the outputs of the original system, and its output the resultant measured data. This is shown in Figure 1.2. A system as a whole is not restricted to the generating system (the brain), but must incorporate the measurement system (the EEG). Analysis of the measurements can provide insight into function and dysfunction of the original system, so long as the recording process itself is well understood.

This chapter is dedicated to presenting concepts necessary to understand the brain as a system. In Section 1.1 the physiology of the brain relevant to epilepsy is summarized. Section 1.2 then concentrates on the measurement and analysis of this brain activity, and Section 1.3 summarizes how physiology and measurement can be turned into a mathematical model of brain dynamics. Section 1.4 discusses how the presence of stochastic elements affects EEG sources, measurement, analysis and modeling.

Each section is introductory only because more detail is included in later chapters, with the exception of Section 1.1 where we attempt to contain all relevant (albeit simplified) physiology. All sections focus on epilepsy and the specific problem of differentiation between seizure and non-seizure activity.

FIGURE 1.1: A system is defined by the measured signals and thus the scale of the system depends on the scales of activity that significantly affect this measurement. In this figure example systems of different scales are shown. If a measurement of the brain only involves activity of single neurons, then the smallest of these systems may be used. However in a signal such as the EEG larger scales must be incorporated, including the effects of the entire head, the recording equipment and maybe even environmental factors not shown. The smaller sub-systems may be used to explain the larger system.

1.1 The Brain and Epilepsy

The brain is part of the central nervous system (CNS) and is responsible for interpreting sensory information received from the environment so that humans can behave as humans do. Each region of the brain has its own task in this process. Figure 1.3(a) shows a functional de-composition of the cerebral cortex — a thin layer approximately 2-3mm thick that covers the entire surface of the brain. The different functional regions of the cortex (temporal lobe, parietal lobe, occipital lobe, frontal lobe) are responsible for motor control as well as cognitive and memory functions. Sensory information is passed on to the cortex from a subcortical system known as the thalamus, shown in Figure 1.3(b). The thalamus also plays an important role in regulating the interaction between different regions of the brain.

In the order of 10-100 billion densely interconnected nerve cells, called neurons, make up the cerebral cortex. How individual neurons work is understood quite well, but it is the complex ways in which they inter-connect and interact that determine brain function. How these networks function is less well understood.

The general structure of these networks is illustrated in Figure 1.3(b) and (c). All mammalian brains look roughly like this, although the details vary. For example neurons in the cerebral cortex of humans are much more densely inter-connected than in animals, and is believed to be one reason for the more sophisticated capabilities of humans. The gray matter in this figure is the cortex, folded to form gyri and sulci and containing all the cortical neurons. Between cortex and subcortex is the white matter — a region composed mostly of connections made between different areas of the brain. The majority of these connections are between different cortical regions, but subcortical systems such as the thalamus also communicate with the cortex through the white matter. Very few neurons can be found in this region.

FIGURE 1.2: Above is an example of a generating system - the brain — and measurement system such as the EEG. In this over-simplistic representation of the brain system, signals that act as inputs are integrated together after some delay. The integration is the output of the generating system that becomes the input to the measurement system. After amplification, filtering and digitization the output of the measurement system is a record of the activity in the generating system. Shown also are the internal signals within each sub-system.

The way by which the brain works can be described in terms of behavior at different spatial scales, traditionally divided into the micro-scopic, mesoscopic and macro-scopic. In this book micro-scopic describes behavior at small scales (µm), encompassing a single or a few brain cells. Macro-scopic describes behavior at large scales (cm), spanning whole regions of brain. The intermediate scale, meso-scopic, describes behavior of networks of neurons spanning millimeters rather than centimeters. In particular the word is used to describe cortical columns, believed to be the main functional units of the cortex and described in more detail in Section 1.1.2.

Brain function can also be described in terms of different temporal scales. These are directly correlated to the spatial scales because of naturally occurring conditions in neural dynamics. The micro-scopic scale is generally associated with frequencies above 1000Hz because the mechanisms within single neurons are very fast. The meso-scopic scales are associated with activity between 10-1000Hz because faster events are negligible relative to the average behavior of ensembles of neurons. Activity at the macro-scopic level is associated with frequencies in the range of 1-100Hz because spatially averaging an even larger number of neurons filters out higher frequencies. These ranges are not rigid, but are used as a guideline in the expected behavior, as explained in more detail in Chapter 2.

FIGURE 1.3: A functional de-composition of the entire brain. (a) shows how the cerebral cortex is divided into the four lobes, each responsible for different cognitive and motor functions. (b) is a cross-section of the brain showing some of the major subcortical systems, the most important for this book being the thalamus whose principal role is to relay sensory information onto the cortex. Also note the relative size of gray matter — where the neurons are — and the white matter — used for connections between sub-systems. In (c) the How of information to and from one cortical column (described in Section 1.1.2) and other regions of the brain is shown. Notice that the inputs to this column come from other cortical columns as well as sub-cortical systems. Most connections are projected through the white matter, although nearby columns are connected more directly. The graphic in (b) is a modified reproduction of a figure taken from Anatomy of the Human Body ([56]), originally published in 1918 and lapsed into the public domain.

Epilepsy is a problem of scale. On the one hand it is a macro-scopic phenomenon that encompasses a large portion if not the whole of the brain. On the other hand, epilepsy’s root cause must be found in the chemokinetic processes that may be associated with meso- or micro-scopic cellular biology. In addition, understanding the macro-scopic may not be possible without knowledge of some of the micro-scopic. The relevant information must be retained. For example, knowing that gating mechanisms that transfer information between neurons is important for the understanding of macro-scopic brain activity, but knowing the individual complex protein structures involved in the different gate types is unlikely to help us understand macro-scopic recordings, and this complexity can be omitted for the purposes of this problem.

The remainder of this section provides a crash course into how the brain works at each scale, limited to the mechanisms that are believed to contribute to the understanding of epilepsy at the macro-scopic scale. ‘Believed’ is used here because epilepsy is not completely understood, and elements that are at this point thought more or less irrelevant may become relevant in the future. Wherever possible efforts are made to avoid superfluous use of medical jargon so that important points are not lost in the translation process. Of course some terminology is always necessary. In any case, as this introduction is by necessity brief and limited to the bare minimum needed to proceed with an understanding of the EEG system, readers are encouraged to obtain a detailed understanding of cellular mechanisms. An excellent introductory text is [80], although any other basic physiology book may be used.

1.1.1 Micro-Scopic Dynamics: Single Neurons

It is the neurons in the CNS that are responsible for the processing and transmission of information, but they are not the only type of cell present. Glial cells in the cortex outnumber neurons by just under 4 to 1, but their role is not as clearly understood. They are believed to be responsible for support roles such as provision of structure, insulation and maintenance [80]. More recent studies reveal that the role of glial cells may not be so passive, but in any case their contributions to neural function are more or less ignored. Since it is likely that their complete function will continue to be unknown for some time they are largely ignored in this text, although one should remember that they exist.

Figure 1.4(a) shows a picture of a stained cortical slice in which many neural cells are visible. These are known as pyramidal neurons and are the most common nerve cells found in the cortex. A structural decomposition of a typical pyramidal neuron is shown in (b). Neurons come in many different shapes and sizes, but they are composed of four basic structures — dendrites, soma or cell body, axon and synaptic terminals.

The in...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• List of Figures
• Preface
• 1 Introduction
• 2 EEG Generation and Measurement
• 3 Signal Processing in EEG Analysis
• 4 Classifying the EEG
• 5 Seizure Detection
• 6 Modeling for Epilepsy
• 7 On the Predictability of Seizure
• 8 Concluding Remark
• Glossary
• Bibliography
• Index