Adaptive Processing of Brain Signals

by Sanei, Saeid

• ISBN13: 9781118622162
• ISBN10: 1118622162
• Edition: 1st
• Format: eBook
• Copyright: 2013-06-04
• Publisher: Wiley

Summary

In this book, the field of adaptive learning and processing is extended to arguably one of its most important contexts which is the understanding and analysis of brain signals. No attempt is made to comment on physiological aspects of brain activity; instead, signal processing methods are developed and used to assist clinical findings. Recent developments in detection, estimation and separation of diagnostic cues from different modality neuroimaging systems are discussed.

These include constrained nonlinear signal processing techniques which incorporate sparsity, nonstationarity, multimodal data, and multiway techniques.

Key features:

• Covers advanced and adaptive signal processing techniques for the processing of electroencephalography (EEG) and magneto-encephalography (MEG) signals, and their correlation to the corresponding functional magnetic resonance imaging (fMRI)
• Provides advanced tools for the detection, monitoring, separation, localising and understanding of functional, anatomical, and physiological abnormalities of the brain
• Puts a major emphasis on brain dynamics and how this can be evaluated for the assessment of brain activity in various states such as for brain-computer interfacing emotions and mental fatigue analysis
• Focuses on multimodal and multiway adaptive processing of brain signals, the new direction of brain signal research

Author Biography

Dr Saeid Sanei, Reader in Biomedical Signal Processing and Deputy Head of Computing Department, Faculty of Engineering and Physical Sciences, University of Surrey, Guildford, Surrey, United Kingdom.

Dr Sanei received his PhD from Imperial College of Science, Technology and Medicine, London, in Biomedical Signal and Image Processing in 1991. He has made a major contribution to Electroencephalogram (EEG) analysis; blind source separation, sparse component analysis and compressive sensing; parallel factor analysis and tensor factorization; particle filtering; chaos and time series analysis; support vector machines; hidden Markov models; and brain computer interfacing (BCI).He has published over 180 papers in refereed journals and conference proceedings, and a book on EEG Signal Processing. He has served as an editor, member of the technical committee, and reviewer for a number of journals and conferences, and has recently been selected as the Biomedical Signal Processing Track Chair for the IEEE Engineering in Medicine and Biology Conference 2009. His international collaborations involve both educational and industrial organizations, including the RIKEN Brain Science Research Institute in Japan. He also teaches extensively at both undergraduate and postgraduate level.

Table of Contents

Preface xiii

1 Brain Signals, Their Generation, Acquisition and Properties 1

1.1 Introduction 1

1.2 Historical Review of the Brain 1

1.3 Neural Activities 5

1.4 Action Potentials 5

1.5 EEG Generation 8

1.6 Brain Rhythms 10

1.7 EEG Recording and Measurement 14

1.7.1 Conventional EEG Electrode Positioning 16

1.7.2 Conditioning the Signals 18

1.8 Abnormal EEG Patterns 19

1.9 Aging 22

1.10 Mental Disorders 23

1.10.1 Dementia 23

1.10.2 Epileptic Seizure and Nonepileptic Attacks 24

1.10.3 Psychiatric Disorders 28

1.10.4 External Effects 29

1.11 Memory and Content Retrieval 30

1.12 MEG Signals and Their Generation 32

1.13 Conclusions 32

References 33

2 Fundamentals of EEG Signal Processing 37

2.1 Introduction 37

2.2 Nonlinearity of the Medium 38

2.3 Nonstationarity 39

2.4 Signal Segmentation 40

2.5 Other Properties of Brain Signals 43

2.6 Conclusions 44

References 44

3 EEG Signal Modelling 45

3.1 Physiological Modelling of EEG Generation 45

3.1.1 Integrate-and-Fire Models 45

3.1.2 Phase-Coupled Models 46

3.1.3 Hodgkin and Huxley Model 48

3.1.4 Morris–Lecar Model 52

3.2 Mathematical Models 54

3.2.1 Linear Models 54

3.2.2 Nonlinear Modelling 57

3.2.3 Gaussian Mixture Model 59

3.3 Generating EEG Signals Based on Modelling the Neuronal Activities 61

3.4 Electronic Models 64

3.4.1 Models Describing the Function of the Membrane 64

3.4.2 Models Describing the Function of Neurons 65

3.4.3 A Model Describing the Propagation of an Action Pulse in an Axon 67

3.4.4 Integrated Circuit Realizations 68

3.5 Dynamic Modelling of the Neuron Action Potential Threshold 68

3.6 Conclusions 68

References 68

4 Signal Transforms and Joint Time–Frequency Analysis 72

4.1 Introduction 72

4.2 Parametric Spectrum Estimation and Z-Transform 73

4.3 Time–Frequency Domain Transforms 74

4.3.1 Short-Time Fourier Transform 74

4.3.2 Wavelet Transform 75

4.3.3 Multiresolution Analysis 78

4.4 Ambiguity Function and the Wigner–Ville Distribution 82

4.5 Hermite Transform 85

4.6 Conclusions 88

References 88

5 Chaos and Dynamical Analysis 90

5.1 Entropy 91

5.2 Kolmogorov Entropy 91

5.3 Lyapunov Exponents 92

5.4 Plotting the Attractor Dimensions from Time Series 93

5.5 Estimation of Lyapunov Exponents from Time Series 94

5.5.1 Optimum Time Delay 96

5.5.2 Optimum Embedding Dimension 97

5.6 Approximate Entropy 98

5.7 Using Prediction Order 98

5.8 Conclusions 99

References 100

6 Classification and Clustering of Brain Signals 101

6.1 Introduction 101

6.2 Linear Discriminant Analysis 102

6.3 Support Vector Machines 103

6.4 k-Means Algorithm 109

6.5 Common Spatial Patterns 112

6.6 Conclusions 115

References 116

7 Blind and Semi-Blind Source Separation 118

7.1 Introduction 118

7.2 Singular Spectrum Analysis 119

7.2.1 Decomposition 119

7.2.2 Reconstruction 120

7.3 Independent Component Analysis 121

7.4 Instantaneous BSS 125

7.5 Convolutive BSS 130

7.5.1 General Applications 130

7.5.2 Application of Convolutive BSS to EEG 132

7.6 Sparse Component Analysis 133

7.7 Nonlinear BSS 134

7.8 Constrained BSS 135

7.9 Application of Constrained BSS; Example 136

7.10 Nonstationary BSS 137

7.10.1 Tensor Factorization for BSS 140

7.10.2 Solving BSS of Nonstationary Sources Using Tensor Factorization 144

7.11 Tensor Factorization for Underdetermined Source Separation 151

7.12 Tensor Factorization for Separation of Convolutive Mixtures in the Time Domain 153

7.13 Separation of Correlated Sources via Tensor Factorization 153

7.14 Conclusions 154

References 154

8 Connectivity of Brain Regions 159

8.1 Introduction 159

8.2 Connectivity Through Coherency 161

8.3 Phase-Slope Index 163

8.4 Multivariate Directionality Estimation 163

8.4.1 Directed Transfer Function 164

8.5 Modelling the Connectivity by Structural Equation Modelling 166

8.6 EEG Hyper-Scanning and Inter-Subject Connectivity 168

8.6.1 Objectives 168

8.6.2 Technological Relevance 169

8.7 State-Space Model for Estimation of Cortical Interactions 173

8.8 Application of Adaptive Filters 175

8.8.1 Use of Kalman Filter 176

8.8.2 Task-Related Adaptive Connectivity 178

8.8.3 Diffusion Adaptation 179

8.8.4 Application of Diffusion Adaptation to Brain Connectivity 179

8.9 Tensor Factorization Approach 182

8.10 Conclusions 184

References 185

9 Detection and Tracking of Event-Related Potentials 188

9.1 ERP Generation and Types 188

9.1.1 P300 and Its Subcomponents 191

9.2 Detection, Separation, and Classification of P300 Signals 192

9.2.1 Using ICA 193

9.2.2 Estimation of Single Trial Brain Responses by Modelling the ERP Waveforms 195

9.2.3 ERP Source Tracking in Time 197

9.2.4 Time–Frequency Domain Analysis 200

9.2.5 Application of Kalman Filter 203

9.2.6 Particle Filtering and Its Application to ERP Tracking 206

9.2.7 Variational Bayes Method 209

9.2.8 Prony’s Approach for Detection of P300 Signals 211

9.2.9 Adaptive Time–Frequency Methods 214

9.3 Brain Activity Assessment Using ERP 216

9.4 Application of P300 to BCI 217

9.5 Conclusions 218

References 219

10 Mental Fatigue 223

10.1 Introduction 223

10.2 Measurement of Brain Synchronization and Coherency 224

10.2.1 Linear Measure of Synchronization 224

10.2.2 Nonlinear Measure of Synchronization 226

10.3 Evaluation of ERP for Mental Fatigue 227

10.4 Separation of P3a and P3b 234

10.5 A Hybrid EEG-ERP-Based Method for Fatigue Analysis Using an Auditory Paradigm 238

10.6 Conclusions 243

References 243

11 Emotion Encoding, Regulation and Control 245

11.1 Theories and Emotion Classification 246

11.2 The Effects of Emotions 248

11.3 Psychology and Psychophysiology of Emotion 251

11.4 Emotion Regulation 252

11.5 Emotion-Provoking Stimuli 257

11.6 Change in the ERP and Normal Brain Rhythms 259

11.6.1 ERP and Emotion 259

11.6.2 Changes in Normal Brain Waves with Emotion 261

11.7 Perception of Odours and Emotion: Why Are They Related? 262

11.8 Emotion-Related Brain Signal Processing 263

11.9 Other Neuroimaging Modalities Used for Emotion Study 264

11.10 Applications 267

11.11 Conclusions 268

References 268

12 Sleep and Sleep Apnoea 274

12.1 Introduction 274

12.2 Stages of Sleep 275

12.2.1 NREM Sleep 275

12.2.2 REM Sleep 277

12.3 The Influence of Circadian Rhythms 278

12.4 Sleep Deprivation 279

12.5 Psychological Effects 280

12.6 Detection and Monitoring of Brain Abnormalities During Sleep by EEG Analysis 281

12.6.1 Analysis of Sleep Apnoea 281

12.6.2 Detection of the Rhythmic Waveforms and Spindles Employing Blind Source Separation 282

12.6.3 Application of Matching Pursuit 282

12.6.4 Detection of Normal Rhythms and Spindles Using Higher Order Statistics 285

12.6.5 Application of Neural Networks 287

12.6.6 Model-Based Analysis 288

12.6.7 Hybrid Methods 290

12.7 EEG and Fibromyalgia Syndrome 290

12.8 Sleep Disorders of Neonates 291

12.9 Dreams and Nightmares 291

12.10 Conclusions 292

References 292

13 Brain–Computer Interfacing 295

13.1 Introduction 295

13.2 State of the Art in BCI 296

13.3 BCI-Related EEG Features 300

13.3.1 Readiness Potential and Its Detection 300

13.3.2 ERD and ERS 300

13.3.3 Transient Beta Activity after the Movement 302

13.3.4 Gamma Band Oscillations 302

13.3.5 Long Delta Activity 303

13.4 Major Problems in BCI 303

13.4.1 Pre-Processing of the EEGs 304

13.5 Multidimensional EEG Decomposition 306

13.5.1 Space–Time–Frequency Method 308

13.5.2 Parallel Factor Analysis 309

13.6 Detection and Separation of ERP Signals 310

13.7 Estimation of Cortical Connectivity 311

13.8 Application of Common Spatial Patterns 314

13.9 Multiclass Brain–Computer Interfacing 316

13.10 Cell-Cultured BCI 318

13.11 Conclusions 319

References 320

14 EEG and MEG Source Localization 325

14.1 Introduction 325

14.2 General Approaches to Source Localization 326

14.2.1 Dipole Assumption 327

14.3 Most Popular Brain Source Localization Approaches 329

14.3.1 ICA Method 329

14.3.2 MUSIC Algorithm 329

14.3.3 LORETA Algorithm 333

14.3.4 FOCUSS Algorithm 335

14.3.5 Standardised LORETA 335

14.3.6 Other Weighted Minimum Norm Solutions 336

14.3.7 Evaluation Indices 338

14.3.8 Joint ICA-LORETA Approach 338

14.3.9 Partially Constrained BSS Method 340

14.3.10 Constrained Least-Squares Method for Localization of P3a and P3b 341

14.3.11 Spatial Notch Filtering Approach 342

14.3.12 Deflation Beamforming Approach for EEG/MEG Multiple Source Localization 347

14.3.13 Hybrid Beamforming – Particle Filtering 351

14.4 Determination of the Number of Sources from the EEG/MEG Signals 353

14.5 Conclusions 355

References 356

15 Seizure and Epilepsy 360

15.1 Introduction 360

15.2 Types of Epilepsy 362

15.3 Seizure Detection 365

15.3.1 Adult Seizure Detection 365

15.3.2 Detection of Neonate Seizure 371

15.4 Chaotic Behaviour of EEG Sources 376

15.5 Predictability of Seizure from the EEGs 378

15.6 Fusion of EEG – fMRI Data for Seizure Detection and Prediction 391

15.7 Conclusions 391

References 392

16 Joint Analysis of EEG and fMRI 397

16.1 Fundamental Concepts 397

16.1.1 Blood Oxygenation Level Dependent 399

16.1.2 Popular fMRI Data Formats 400

16.1.3 Preprocessing of fMRI Data 401

16.1.4 Relation between EEG and fMRI 401

16.2 Model-Based Method for BOLD Detection 403

16.3 Simultaneous EEG-fMRI Recording: Artefact Removal from EEG 405

16.3.1 Gradient Artefact Removal 405

16.3.2 Ballistocardiogram Artefact Removal 406

16.4 BOLD Detection in fMRI 413

16.4.1 Implementation of Different NMF Algorithms for BOLD Detection 414

16.4.2 BOLD Detection Experiments 416

16.5 Fusion of EEG and fMRI 419

16.5.1 Extraction of fMRI Time-Course from EEG 419

16.5.2 Fusion of EEG and fMRI, Blind Approach 421

16.5.3 Fusion of EEG and fMRI, Model-Based Approach 425

16.6 Application to Seizure Detection 425

16.7 Conclusions 427

References 427

Index 433