Immediate Download Computational Intelligence in Biomedical Engineering 1st Edition Rezaul Begg Ebooks 2024
Immediate Download Computational Intelligence in Biomedical Engineering 1st Edition Rezaul Begg Ebooks 2024
Immediate Download Computational Intelligence in Biomedical Engineering 1st Edition Rezaul Begg Ebooks 2024
com
https://ebookname.com/product/computational-intelligence-in-
biomedical-engineering-1st-edition-rezaul-begg/
OR CLICK BUTTON
DOWNLOAD EBOOK
https://ebookname.com/product/computational-imaging-and-
analytics-in-biomedical-engineering-1st-edition-t-r-ganesh-babu/
https://ebookname.com/product/advances-in-computational-
intelligence-fei/
https://ebookname.com/product/pinch-of-nom-100-slimming-home-
style-recipes-kay-featherstone/
Teddy Bear Teddy Bear 1st Edition Alice Schertle
https://ebookname.com/product/teddy-bear-teddy-bear-1st-edition-
alice-schertle/
https://ebookname.com/product/everywhere-taksim-sowing-the-seeds-
for-a-new-turkey-at-gezi-1st-edition-isabel-david/
https://ebookname.com/product/feldenkrais-the-busy-person-s-
guide-to-easier-movement-3rd-edition-frank-wildman/
https://ebookname.com/product/odin-s-ways-a-guide-to-the-pagan-
god-in-medieval-literature-1st-edition-annette-lassen/
https://ebookname.com/product/an-outline-of-the-history-of-
economic-thought-2nd-edition-ernesto-screpanti/
Nutritional Anthropology Biocultural Perspectives on
Food and Nutrition 2nd Edition Darna L. Dufour
https://ebookname.com/product/nutritional-anthropology-
biocultural-perspectives-on-food-and-nutrition-2nd-edition-darna-
l-dufour/
CRC-4080–fm.tex 2/11/2007 10: 18 Page i
Computational
Intelligence in
Biomedical
Engineering
CRC-4080–fm.tex 2/11/2007 10: 18 Page ii
CRC-4080–fm.tex 2/11/2007 10: 18 Page iii
Computational
Intelligence in
Biomedical
Engineering
Rezaul Begg
Daniel T. H. Lai
Marimuthu Palaniswami
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2008 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
This book contains information obtained from authentic and highly regarded sources. Reprinted
material is quoted with permission, and sources are indicated. A wide variety of references are
listed. Reasonable efforts have been made to publish reliable data and information, but the author
and the publisher cannot assume responsibility for the validity of all materials or for the conse-
quences of their use.
No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any
electronic, mechanical, or other means, now known or hereafter invented, including photocopying,
microfilming, and recording, or in any information storage or retrieval system, without written
permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.
copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC)
222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that
provides licenses and registration for a variety of users. For organizations that have been granted a
photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Begg, Rezaul.
Computational intelligence in biomedical engineering / Rezaul Begg, Daniel
T.H. Lai, and Marimuthu Palaniswami.
p. ; cm.
“A CRC title.”
Includes bibliographical references and index.
ISBN 978-0-8493-4080-2 (hardcover : alk. paper) 1. Artificial
intelligence--Medical applications. 2. Biomedical engineering--Computer
simulation. I. Lai, Daniel T. H. II. Palaniswami, Marimuthu. III. Title.
[DNLM: 1. Artificial Intelligence. 2. Biomedical Engineering. 3. Medical
Informatics Applications. W 26.55.A7 B416c 2008]
R859.7.A78B44 2008
610.285’63--dc22 2007032849
Dedication
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Biomedical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Medical Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Medical Instrumentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
1.1.3 Medical Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
1.2 Bioelectric Signals and Electrode Theory . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Signal Processing and Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Computational Intelligence Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4.1 Supervised Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.2 Fuzzy Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
1.4.3 Evolutionary Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.5 Chapter Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6 Book Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
vii
CRC-4080–fm.tex 2/11/2007 10: 18 Page viii
viii Contents
2.4.2
Nonparametric Estimation Models. . . . . . . . . . . . . . . . . . . . . . . .36
2.4.3
Parametric Estimation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.4.3.1 Autoregressive Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.4.3.2 Moving Average Model . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4.3.3 Autoregressive Moving Average Model . . . . . . . . . . . 40
2.5 Analog Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5.1 Basic Analog Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5.1.1 RC Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5.1.2 RL Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.5.2 Butterworth Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.5.3 Chebyshev Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.6 Digital Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.1 Infinite Impulse Response Filters . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.1.1 Direct IIR Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.1.2 Canonical IIR Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.6.1.3 Cascaded IIR Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.6.2 Design of IIR Digital Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.6.2.1 Impulse-Invariant Transformation . . . . . . . . . . . . . . . . 53
2.6.2.2 Bilinear Transformation . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.6.3 Finite Impulse Response Filters . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.6.3.1 Direct FIR Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.6.3.2 Cascaded FIR Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.6.4 Design of FIR Digital Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.6.4.1 The Window Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.6.4.2 Frequency Sampling Method . . . . . . . . . . . . . . . . . . . . . 59
2.6.5 Integer Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.6.5.1 Design of Integer Filters . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.7 Adaptive Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.7.1 General Adaptive Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.7.2 The LMS Algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64
2.7.3 The RLS Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.8 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Contents ix
x Contents
Contents xi
6 Computational Intelligence in
Electroencephalogram Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
6.1.1 The Brain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
6.1.2 Electroencephalography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
6.1.3 Uses of EEG Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
6.2 Computational Intelligence in EEG Analysis . . . . . . . . . . . . . . . . . . . 279
6.2.1 Computational Intelligence in the Diagnosis
of Neurological Disorders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
6.2.1.1 Epilepsy Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
6.2.1.2 Huntington’s Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
6.2.1.3 Alzheimer’s Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
6.2.2 Evoked Potentials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .284
6.3 Brain–Computer Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
6.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
xii Contents
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
CRC-4080–fm.tex 2/11/2007 10: 18 Page xiii
Preface
xiii
CRC-4080–fm.tex 2/11/2007 10: 18 Page xiv
xiv Preface
every new CI technique has been tested on processed ECG data. Chapter 5
focuses on CI applications in electromyography (EMG), pattern recognition,
and the diagnosis of neuromuscular pathologies that affect EMG-related sig-
nal characteristics. More recently, CI has also been applied to the control of
myoelectric prostheses and exoskeletons. Of interest here is the combination of
signal-processing and classifier systems to detect the user’s intention to move
the prosthesis. In Chapter 6, CI applications to bioelectric potentials rep-
resenting brain activity are outlined based on electroencephalogram (EEG)
recordings. CI approaches play a major role in EEG signal processing because
of their effectiveness as pattern classifiers. Here we concentrated on several
applications including the identification of abnormal EEG activity in patients
with neurological diseases (e.g., epilepsy) and in the control of external devices
using EEG waveforms, known as brain–computer interfaces (BCI). The BCI
has many potential applications in rehabilitation, such as assisting individuals
with disabilities to independently operate appliances. Chapter 7 provides an
overview of CI applications for the detection and classification of gait types
from their kinematic, kinetic, and EMG features. Gait analysis is routinely
used for detecting abnormality in lower-limb function and also for evaluating
the progress of treatment. Various studies in this area are discussed with a
particular focus on CI’s potential as a tool for gait diagnostics. Finally, Chap-
ter 8 discusses progress in biomedical engineering, biomedicine, and human
health areas with suggestions for future applications.
This book should be of considerable help to a broad readership, including
researchers, professionals, academics, and graduate students from a wide range
of disciplines who are beginning to look for applications in health care. The
text provides a comprehensive account of recent research in this emerging
field and we anticipate that the concepts presented here will generate further
research in this multidisciplinary field.
Rezaul K. Begg
Daniel T. H. Lai
Marimuthu Palaniswami
CRC-4080–fm.tex 2/11/2007 10: 18 Page xv
Abbreviations
Acronym Definition
AD Alzheimer’s disease
AEP Auditory evoked potential
AIC Akaike’s information criterion
ADALINE Adaptive linear networks
AMI Acute myocardial infarctions
ANN Artificial neural networks
ANFIS Adaptive neurofuzzy system
A/P Anterior–posterior
AR Autoregressive model
ARMA Autoregressive model with moving average
ART Adaptive resonance theory
BCI Brain–computer interface
BP Backpropagation
CI Computational intelligence
CP Cerebral palsy
CV Cross-validation
CVD Cardiovascular disease
CHF Congestive heart failure
CNV Cognitive negative variation
CSSD Common spatial subspace decomposition
ECG Electrocardiogram
EMG Electromyogram
EVP Evoked potentials
FA Fuzzy ARTMAP
FDA Food and Drug Administration
FIR Finite impulse response
FFT Fast fourier transform
FPE Final prediction error
FLD Fisher’s linear discriminant
GA Genetic algorithm
GRF Ground reaction forces
HD Huntington’s disease
HMM Hidden Markov model
HOS Higher-order statistics
IIR Infinite impulse response
ICD Implantable cardioverter–defibrillators
i.i.d Independent and identically distributed
IIP Intelligent information processing
KKT Karush–Kuhn–Tucker optimality conditions
LS Least squares
LDA Linear discriminant analysis
xv
CRC-4080–fm.tex 2/11/2007 10: 18 Page xvi
xvi Abbreviations
Acknowledgments
xvii
CRC-4080–fm.tex 2/11/2007 10: 18 Page xviii
CRC-4080–fm.tex 2/11/2007 10: 18 Page xix
Authors
xix
CRC-4080–fm.tex 2/11/2007 10: 18 Page xx
xx Authors
1
Introduction
With the advent of the twenty-first century, the focus on global health and
personal well-being has increased. In 2006, the population of the Earth was
estimated to be 6.5 billion,∗ which is expected to rise steadily in the coming
years. Health has become more important with recent advances in genetic
research, development of new and improved drugs, diagnosis of disease, and
the design of more advanced systems for patient monitoring. Health now ranks
as one of the most important needs in modern society as evidenced by increas-
ing government funding around the globe.
In this opening chapter we examine the fundamental components of a
biomedical system, ranging from the medical data to the instrumentation
that can be used to diagnose a disorder. The vast amount of data collected
has resulted in increasingly extensive databases, which cannot always be effi-
ciently interpreted by clinicians or doctors, resulting in excessive waiting times
for a patient to be diagnosed. The problem is exacerbated by the shortage of
qualified medical staff causing unnecessary delays, which can be exceptionally
fatal to patients with serious disorders. Consequently, recent research issues
have focused on new and improved diagnostic methods to alleviate the risk of
life loss due to inadequate diagnostic capacity.
A new paradigm in the field of medicine is computational intelligence (CI) for
diagnostic systems. Computational intelligence is a combination of advances
in artificial learning and computers, which was introduced more than five
decades ago. These techniques are extremely valuable for interpreting large
volumes of data and have been successfully applied to pattern recognition
and function estimation problems. In short, CI techniques provide a possible
solution to the growing need for automated diagnostic systems. In the fol-
lowing chapters, we will examine the fundamentals of automated biomedical
diagnostic systems and highlight the areas in which CI techniques have been
successfully employed.
∗ http://geography.about.com/od/obtainpopulationdata/a/worldpopulation.htm
1
CRC-4080–ch001.tex 10/10/2007 15: 4 Page 2
Medical data
Physiological Alphanumeric
Image data signal data data
FIGURE 1.1
Categories of medical data that can be collected.
CRC-4080–ch001.tex 10/10/2007 15: 4 Page 3
Introduction 3
central nervous system proved extremely difficult due to the lack of under-
standing in the individual workings of the subsystems. It was realized that
newer techniques, such as classifiers in pattern recognition, could contribute
greatly toward understanding these subsystems and research began in inte-
grating CI into systems modelling.
Traditional modelling approaches had a number of weaknesses; for exam-
ple, strict algorithms were inadequate in modelling biological systems since
they required precise knowledge, which was unavailable then. Lack of under-
standing resulted in a shortage of deterministic models, and other approaches
such as knowledge-based modelling were also inefficient. This type of method
required opinions of a small number of experts to make a diagnosis. Serious
problems arose when these experts disagreed on the diagnosis. Further, it
was discovered that generalization to other cases were inadequate since they
depended on data instances and patient physiological characteristics tended
to vary from person to person.
Nevertheless, biomedical systems have been created using several other
approaches. The symbolic approach constructs a knowledge-based system
using a group of experts. The numerical approach relies on computational
techniques such as neural networks for pattern recognition whereas hybrid
systems are birthed from two or more techniques combined to solve a single
problem. A biomedical system incorporating these intelligent techniques is
depicted in Figure 1.2.
Feedback
control
Controller
Interpretation
Diagnosis
with CI
FIGURE 1.2
Flowchart diagram detailing the fundamental components in a biomedical
monitoring system.
CRC-4080–ch001.tex 10/10/2007 15: 4 Page 5
Introduction 5
TABLE 1.1
Bioelectric Measurement Techniques and the Respective Voltage and Fre-
quency Ranges
Frequency
Bioelectric Measurements Signal Range (mV) Range (Hz)
RT
EM = ln(M + ) + E0 (1.1)
nF
where R is the universal gas constant, T the absolute temperature, n the ion
valence, F the Faraday constant, and E0 some constant. An ionic solution is
sometimes referred to as an electrolyte because it is used to conduct charges
between two electrodes.
Ions in the human body can be divided into majority and minority carriers.
Majority carriers are responsible for carrying most of the electrical charge
and include ions such as sodium (Na+ ), chloride (Cl− ), and potassium (K+ )
(usually at higher frequencies, such as in the muscle). These ions are present
in high concentrations in the human body. Minority carriers are other ions
present in lower concentration and do not contribute significantly to carrying
charges. Ions can form interfaces between metal electrodes, between electrodes
and electrolytes, and between electrolytes and membranes. The electrode con-
ducts charges from the ionic solution to a metal and can then be recorded and
stored.
When a metal is placed in an electrolyte, double layers are formed near the
electrode surface, a net current results, which passes from the electrode to
the electrolyte. The current consists of electrons and negative ions (anions)
moving in opposite direction to the current as well as positive ions (cations)
moving in the same direction of the current. These can be written as the
following chemical equations:
C Cn+ + ne−
A Am− − me−
Introduction 7
Core
60 years.
Hybrid
Very well Supervised 10 years.
known, almost
Still ongoing.
saturated,
Includes
difficult to
concepts such
develop
as fuzzy
something
Evolutionary classifiers,
novel
fuzzy
evolutionary
Fuzzy algorithms, and
evolutionary
supervised
algorithms
Tri-breed (new)
Not much evidence, attempts made,
deals with online data in real time world, e.g.,
evolvable neural networks with fuzzification
FIGURE 1.3
The main computational intelligence paradigms and the state of current
research.
CRC-4080–ch001.tex 10/10/2007 15: 4 Page 9
Introduction 9
swarm and ant swarm optimization are similar in concept; they use obser-
vations on how swarms work collectively to achieve a global objective. In
these techniques, a global fitness function or objective is defined along with
several local objectives. These local objectives depict members of a swarm
and are continuously updated according to certain criteria so that the swarm
eventually discovers the optimal solution. One must define the fitness func-
tion, local objectives, local update rules, and elements of the swarm. The
main drawback, however, is algorithm stability and convergence, which is
difficult to guarantee at times. In Chapter 3, we describe several CI tech-
niques in detail but have not included the topic of evolutionary algorithms
because there are very few biomedical applications that have applied them.
There is nevertheless scope for their application, and the interested reader
is referred instead to some of the more authoritative references (Fogel, 1995;
Sarker et al., 2002) for applying this CI technology to problems in biomedical
engineering.
Introduction 11
Introduction 13
References
Barnett, G., N. Justice, M. Somand, J. Adams, B. Waxman, P. Beaman,
M. Parent, F. Vandeusen, and J. Greenlie (1979). COSTAR—a computer-
based medical information system for ambulatory care. Proceedings of the
IEEE 67(9), 1226–1237.
Barricelli, N. A. (1954). Esempi numerici di processi di evoluzione. Methodos,
45–68.
Fogel, D. (1995). Evolutionary Computation: Toward a New Philosophy of
Machine Intelligence. New York: IEEE Press.
Fraser, A. (1957). Simulation of genetic systems by automatic digital comput-
ers I. Introduction. Australian Journal of Biological Sciences 10, 484–491.
Hudson, D. L. and M. E. Cohen (1999). Neural Networks and Artificial Intel-
ligence for Biomedical Engineering. IEEE Press Series in Biomedical Engi-
neering. New York: Institute of Electrical and Electronics Engineers.
Rosenblatt, F. (1958). The perceptron: a probabilistic model for information
storage and organization in the brain. Psychological Review 65, 386–408.
Sarker, R., M. Mohammadian, and X. Yao (Eds) (2002). Evolutionary Opti-
mization. Boston, MA: Kluwer Academic.
Shavlik, J. and T. Dietterich (1990). Readings in Machine Learning. The
Morgan Kaufmann series in machine learning. San Mateo, CA: Morgan
Kaufmann Publishers.
CRC-4080–ch001.tex 10/10/2007 15: 4 Page 14
Dimanche 16 novembre.
Mardi.
« Madeleine Airvault. »
Une larme avait maculé ces dernières lignes, que Raymonde frôla
pieusement des lèvres.
Il lui suffit d’ouvrir la troisième enveloppe pour qu’un fluide
brûlant parcourût ses veines. C’était seulement un petit carton,
griffonné de quelques lignes.
« Valentin. »
« Raymonde. »
VIII