KNN Classifier and K-Means Clustering for Robust Classification of Epilepsy from EEG Signals. A Detailed Analysis
©2017
Textbook
54 Pages
Summary
Epilepsy is a chronic disorder, the hallmark of which is recurrent, unprovoked seizures. Many people with epilepsy have more than one type of seizures and may have other symptoms of neurological problems as well. Epilepsy is caused due to sudden recurrent firing of the neurons in the brain. The symptoms are convulsions, dizziness and confusion. One out of every hundred persons experiences a seizure at some time in their lives. It may be confused with other events like strokes or migraines. Unfortunately, the occurrence of an epileptic seizure seems unpredictable and its process still is hardly understood.
In India, the number of persons suffering from epilepsy is increasing every year. The complexity involved in the diagnosis and therapy has to be cost effective. In this project, the authors applied an algorithm which is used for a classification of the risk level of epilepsy in epileptic patients from Electroencephalogram (EEG) signals. Dimensionality reduction is done on the EEG dataset by applying Power Spectral density. The KNN Classifier and K-Means clustering is implemented on these spectral values to epilepsy risk level detection. The Performance Index (PI) and Quality Value (QV) are calculated for the above methods. A group of twenty patients with known epilepsy findings are used in this study.
In India, the number of persons suffering from epilepsy is increasing every year. The complexity involved in the diagnosis and therapy has to be cost effective. In this project, the authors applied an algorithm which is used for a classification of the risk level of epilepsy in epileptic patients from Electroencephalogram (EEG) signals. Dimensionality reduction is done on the EEG dataset by applying Power Spectral density. The KNN Classifier and K-Means clustering is implemented on these spectral values to epilepsy risk level detection. The Performance Index (PI) and Quality Value (QV) are calculated for the above methods. A group of twenty patients with known epilepsy findings are used in this study.
Excerpt
Table Of Contents
4
LIST OF TABLES
Table 2.1 Target Values for Groups ... 23
Table 2.2 Features of various windowing techniques ... 28
Table 3.1 Performance parameters of KNN Classifier ... 37
Table 4.1 Performance parameters of K-Means clustering ... 42
Table 5.1 Performance Comparisons of KNN Classifier and K-Means Clustering ... 46
5
LIST OF FIGURES
Figure 1.1 EEG Waveform ... 11
Figure 2.1 EEG Recording by 10-20 system ... 20
Figure 2.2 Epileptic EEG Signal Waveform of Patient ... 21
Figure 2.3 Sample 2-second epoch ... 22
Figure 2.4 Flow diagram of the Epilepsy Risk Level Classification System ... 24
Figure 2.5 Power spectral density of a signal ... 26
Figure 3.1 KNN Classifier ... 33
Figure 4.1 K-Means clustering working ... 40
Figure 5.1 Sensitivity and Specificity measures of KNN classifier and
K- means clustering ... 46
Figure 5.2 Quality factor and Time delay measures of KNN classifier and
K- means clustering ... 47
Figure 5.3 Average detection and Quality factor measures of KNN classifier and
K- means clustering ... 47
6
LIST OF SYMBOLS
Alpha
Sigma
Gamma
Theta
Omega
K
Kilo
Ohm
7
LIST OF ABBREVIATIONS
EEG
Encephalographer
EEGer
Electroencephalogram
ECOG
Electrocorticogram
Hz
Hertz
GA
Genetic Algorithm
NDS
Nonlinear Dynamic System
PSD
Power Spectral Density
KNN
K- Nearest Neighborhood
PI
Performance Index
PC
Perfect Classification
Qv
Quality value
ROC
Receiver Operating Characteristics
9
CHAPTER 1
INTRODUCTION
Epilepsy, from which approximate 1% of the people in the world suffer, is a group of
brain disorders characterized by the recurrent paroxysmal electrical discharges of the
cerebral cortex, that result in irregular disturbances of the brain functions, which are
associated with the significant changes of the EEG signal Electroencephalograms (EEGs)
are recordings of the electrical potentials produced by the brain[1]. Analysis of EEG
activity has been achieved principally in clinical settings to identify pathologies and
epilepsies since Hans Berger's recording of rhythmic electrical activity from the human
scalp. In the past, interpretation of the EEG was limited to visual inspection by a
neurophysiologist, an individual trained to qualitatively make a distinction between
normal EEG activity and abnormalities contained within EEG records. It is known that
biological neurons can be modeled by a set of nonlinear differential equations. The
minimal embedding dimension gives the upper number of nonlinear dynamic system
(NDS) freedom degrees and the minimal number of differential equations demanded for
mathematical modeling of NDS. Therefore, the change of the structure of brain NDS
during seizure can be shown by the change of embedding dimension of EEG signals if
the human brain is considered as a nonlinear dynamic system. A common form of
recording used for this purpose is an ambulatory recording that contains EEG data for a
very long duration of even up to one week. It involves an expert's efforts in analyzing the
entire length of the EEG recordings to detect traces of epilepsy [2]. Because seizures, in
general, occur frequently and unpredictably, automatic detection of seizures during long
term EEG monitoring sessions is highly useful and needed.
Electroencephalography (EEG) is an important clinical tool, monitoring,
diagnosing and managing neurological disorders related to epilepsy. In comparison with
other methods such as Electrocorticogram (ECOG), EEG is a clean and safe technique for
monitoring the brain activity [11]. In spite of available dietary, drug and surgical
treatment options, currently nearly one out of three epilepsy patients cannot be treated.
They are completely subject to the sudden and unforeseen seizures which have a great
10
effect on their daily life, with temporary impairments of perception, speech, motor
control, memory and/or consciousness [3]. Many new therapies are being investigated
and among them the most promising are implantable devices that deliver direct electrical
stimulation to affected areas of the brain. These treatments will greatly depend on robust
algorithms for seizure detection to perform effectively. Because the onset of the seizures
cannot be predicted in a short period, a continuous re-cording of the EEG is required to
detect epilepsy. How-ever, analysis by visual inspection of long recordings of EEG, in
order to find traces of epilepsy, is tedious, time- consuming and high-cost. Therefore,
automated detection of epilepsy has been a goal of many researchers for a long time.
With the advent of technology, the digital EEG data can be input to an automated seizure
detection system, allowing physicians to treat more patients in a given time because the
time taken to review the EEG data is greatly reduced by automation.
1.1
FUNDAMENTALS OF EEG
The human Electroencephalogram (EEG) is usually recorded from electrodes attached to
the scalp using high amplifiers, which are usually coupled to the scalp electrodes. The
amplified signals are written out on paper via a polygraph, which contains typically 8 to
16 channels. Normal subjects usually exhibit alpha, beta, theta and delta activities, while
abnormal activity may be manifested by a slowing and decrease in amplitude of EEG,
increase in the EEG frequency, and the presence of sudden EEG discharges (paroxysmal
activity) different from the background either in frequency content or amplitude or
pattern.
The EEG is a powerful tool for the diagnosis of neurological disorders. Since its
discovery, the EEG has been used for the diagnosis of epilepsy, for trauma assessment,
for sleep research, and for the analysis of higher brain functions. The EEG is highly
dependent upon the availability of high quality instrumentation, and almost from the
beginning, automated methods of signal qualification have been applied. One of the
primary goals is to help the Encephalographer (EEGer) in the time consuming task of
quantifying signal that appears to the eye as a low information content background
intermixed with either bursts of rhythmic activity with different frequencies (the EEG
11
rhythms) or short transients of clinical significance (such as spikes). In spite of years of
research to produce universal automated detection methods, success has been achieved
only in specific areas. Accomplishments include automatically sleep staging with a high
degree of accuracy; counting spikes and wave complexes, and monitoring in intensive
care units. However clinicians still rely on visual analysis for clinical applications.
Figure 1.1 EEG Waveform
The human eye-brain can be trained to recognize ostensibly defined patterns in multi-
channel EEG recordings. However, ostensive definitions are not readily disseminated.
A description of a mental image by words is normally poor and lengthy. What is
needed in a clinical practice is a way of exploring the great pattern recognition of a
human visual system and enhancing the efficiency of the visual data communication.
12
Computers can bring quantification to EEG analysis in the form of precise
Measurements (micro volt and millisecond precision), but at this time they cannot
always use the measured data to identify clinically significant features. All these
aspects lead us to approach the use of computers in EEG research from a slightly
different angle. We are also researching the design of computer-based environments
that will help the doctor in the visual clinical assessment of multi-channel EEG
recordings, and the engineer in the design of better detectors.
It is widely accepted that the information available to the physician about his
patient and about medical relationships in general is inherently uncertain. Nevertheless,
the physician is still quite capable of drawing conclusions, though approximate, from this
information. The novel attempt in this project is to provide a formal model of this process
using a mathematical approach in implementing the model in the form of a computerized
diagnostic system. The two contrasting and complementary approach include onset
indication by aggregation analysis and optimized classification of the risk level of
epilepsy patients.
1.2
EPILEPSY TYPES AND SYMPTOMS
While many types of repetitive behavior may represent a neurological problem, a doctor
needs to establish whether or not they are seizures[4].
1.2.1 GENERALIZED SEIZURES
· All areas of the brain (the cortex) are involved in a generalized seizure.
Sometimes these are referred to as grand mal seizures.
· The person experiencing such a seizure may cry out or make some sound,
stiffen for several seconds to a minute and then have rhythmic movements
of the arms and legs. Often the rhythmic movements slow before stopping.
· Eyes are generally open.
13
· The person may appear to not be breathing and actually turn blue. This
may be followed by a period of deep, noisy breathes.
· The return to consciousness is gradual and the person may be confused for
quite some time -- minutes to hours.
· Loss of urine is common.
· The person will frequently be confused after a generalized seizure.
1.2.2 PARTIAL OR FOCAL SEIZURES
· Only part of the brain is involved, so only part of the body is affected.
Depending on the part of the brain having abnormal electrical activity,
symptoms may vary[5].
· If the part of the brain controlling movement of the hand is involved, then
only the hand may show rhythmic or jerky movements.
· If other areas of the brain are involved, symptoms might include strange
sensations like a full feeling in the stomach or small repetitive movements
such as picking at one's clothes or smacking of the lips.
· Sometimes the person with a partial seizure appears dazed or confused.
This may represent a complex partial seizure. The term complex is used by
doctors to describe a person who is between being fully alert and
unconscious.
1.2.3 ABSENCE OR PETIT MAL SEIZURES
· These are most common in childhood.
· Impairment of consciousness is present with the person often staring blankly.
· Repetitive blinking or other small movements may be present.
· Typically, these seizures are brief, lasting only seconds. Some people may
have many of these in a day.
14
1.3
MATHEMATICAL APPROACH IN MEDICAL DIAGNOSIS
Mathematics is one of the most useful and fascinating divisions of human knowledge.
The most important skills in mathematics are careful analysis and clear reasoning,
entirely based on logic. Starting with widely accepted statements, mathematics can be
used to draw logical conclusions and develop complete systems based on such
conclusions. The two forms of mathematics, namely pure and applied mathematics, do
not have a clear preset boundary between them. Where pure mathematics seeks to
advance mathematical knowledge, applied mathematics seeks to develop mathematical
techniques for use in science and other fields.
Mathematics is an essential part of any scientific study. It provides a plethora of
techniques to analyze, quantify and qualify the scientific data. The same can be said
about application of mathematical techniques in medical diagnosis. Frequency time
analysis of signals is possible with the aid of Fourier Transforms and likewise, other
analytical methods can be used in obtaining different characteristics of the various signals
originating in different parts of the body, helpful in determining defects and other
inherent pathologies.
Precision exists only through abstraction. Abstraction may be defined as the
ability of human beings to recognize and select the relevant properties of real world
phenomena and objects. This leads to the construction of conceptual models defining
abstract classes of phenomena and objects. However, in actuality every real-world
phenomenon and object is of course unique. Abstract models of real-world phenomena
and objects such as mathematical structures (circle, point, etc.), equalities (a=b + c), and
propositions (yes, no) are artificial constructs. They represent ideal structures, ideal
equalities, and ideal positions. Nevertheless, despite these caveats, abstraction forms the
basis of human thought, and human knowledge is its result.
15
1.4
EEG SIGNALS FOR EPILEPSY DETECTION
Epileptic seizures result from a temporary electrical disturbance of the brain. Sometimes
seizures may go unnoticed, depending on their presentation, and sometimes may be
confused with other events, such as a stroke, which can also cause falls or migraines.
Approximately one in every 100 persons will experience a seizure at some time in their
life. Unfortunately, the occurrence of an epileptic seizure seems unpredictable and its
process is very little understood. Since its discovery by R.Caton, the
Electroencephalogram (EEG) has been the most utilized signal to clinically assess brain
activities[7][8][9]. Twenty five percent of the world's 50 million people with epilepsy
have seizures that cannot be controlled by any available treatment. The need for new
therapies, and success of similar devices to treat cardiac arrhythmias, has spawned an
explosion of research into algorithms for use in implantable therapeutic devices for
epilepsy. Most of these algorithms focus on either detecting unequivocal EEG onset of
seizures or on quantitative methods for predicting seizures in the state space, time, or
frequency domains that may be difficult to relate to the Neuro physiology of epilepsy.
Between seizures, the EEG of a patient with epilepsy may be characterized by occasional
epileptic form transients-spikes and sharp waves. EEG patterns have shown to be
modified by a wide range of variables including biochemical, metabolic, circulatory,
hormonal, neuro electric and behavioral factors [10].
Exploring various analytical approaches, both linear and non linear methods to
process data from medical database is meaningful before deciding on the tool that will be
most useful, accurate, and relevant for practitioners. For example, assigning a new patient
to a particular outcome class is a classification problem commonly described as "pattern
recognition", "discriminant analysis", and "supervised learning". In the past, the
Encephalographer, by visual inspection was able to qualitatively distinguish normal EEG
activity from localized or generalized abnormalities contained within relatively long EEG
records. The different types of epileptic seizures are characterized by different EEG
waveform patterns. With real-time monitoring to detect epileptic seizures gaining
widespread recognition, the advent of computers has made it possible to effectively apply
a host of methods to quantify the changes occurring based on the EEG signals. One of
them is a classification of risk level of epilepsy by using Fuzzy techniques. The
Details
- Pages
- Type of Edition
- Erstausgabe
- Year
- 2017
- ISBN (PDF)
- 9783960676409
- ISBN (Softcover)
- 9783960671404
- File size
- 1.6 MB
- Language
- English
- Institution / College
- Bannari Amman Institute of Technology
- Publication date
- 2017 (March)
- Grade
- 9.5/10
- Keywords
- Electroencephalogram Performance Index Quality Value India Epileptic seizure Epilepsy detection EEG Data K-Nearest Neighbor Algorithm Seizure detection Diagnostics Diagnosis
