Brain-Computer Interfaces (eBook)

Preface 6
Contents 8
Contributors 10
List of Abbreviations 14
BrainComputer Interfaces: A Gentle Introduction 16
1 What is a BCI? 17
2 How Do BCIs Work? 20
2.1 Measuring Brain Activity (Without Surgery) 21
2.2 Measuring Brain Activity (With Surgery) 22
2.3 Mental Strategies and Brain Patterns 24
2.3.1 Selective Attention 25
2.3.2 Motor Imagery 26
2.4 Signal Processing 28
3 BCI Performance 29
4 Applications 31
5 Summary 37
References 39
Brain Signals for BrainComputer Interfaces 43
1 Introduction 43
1.1 The Need for BCIs 43
1.2 Key Principles 43
1.3 The Origin of Brain Signals Used in BCIs 44
2 Brain Signals for BCIs and Their Neurophysiological Origins 45
2.1 Brain Signal Features Measured Noninvasively 46
2.1.1 Event-related Potentials (ERPs) 46
2.1.2 Cortical Oscillations 49
2.2 Brain Signal Features Measured from the Cortical Surface 51
2.3 Brain Signal Features Measured Within the Cortex 51
2.3.1 Local Field Potentials (LFPs) in the Time Domain 52
2.3.2 Local Field Potentials in the Frequency Domain 52
2.3.3 Single-Neuron Activity 52
3 Requirements for Continued Progress 53
References 54
Dynamics of Sensorimotor Oscillations in a Motor Task 61
1 Introduction 61
2 EventRelated Potentials Versus ERD/ERS 62
3 Mu and Beta ERD in a Motor Task 62
4 Interpretation of ERD and ERS 65
5 Focal ERD/Surround ERS 66
6 Induced Beta Oscillations after Termination of a Motor Task 67
7 Short-Lived Brain States 69
8 Observation of Movement and Sensorimotor Rhythms 71
9 Conclusion 73
References 73
Neurofeedback Training for BCI Control 79
1 Introduction 79
2 Principles of Neurofeedback 80
2.1 Training of Sensorimotor Rhythms 81
2.2 How Neurofeedback Works 82
3 Training Paradigms for BCI Control 82
3.1 Training with the Graz-BCI 83
3.2 Impact of Feedback Stimuli 85
4 Final Considerations 87
References 89
The Graz Brain-Computer Interface 93
1 Introduction 93
2 The Graz BCI 93
3 Motor Imagery as Mental Strategy 95
3.1 Induced Oscillations in Non-attended Cortical Body Part Areas 96
3.2 Induced Beta Oscillations in Attended Cortical Body Part Areas 97
3.3 The Beta Rebound (ERS) and its Importance for BCI 98
4 Feature Extraction and Selection 99
5 Frequency Band and Electrode Selection 101
6 Special Applications of the Graz BCI 102
6.1 Self-Paced Exploration of the Austrian National Library 102
6.2 Simulation of Self-Paced Wheel Chair Movement in a Virtual Environment 103
6.3 Control of Google Earth 105
7 Future Aspects 106
References 107
BCIs in the Laboratory and at Home: The WadsworthResearch Program 111
1 Introduction 111
2 Sensorimotor Rhythm-Based Cursor Control 112
3 P300-Based Item Selection 116
4 A BCI System for Home Use 120
5 SMR-Based Versus P300-Based BCIs 121
References 123
Detecting Mental States by Machine Learning Techniques: The Berlin BrainComputer Interface 126
1 Introduction 126
1.1 The Machine Learning Approach 126
1.2 Neurophysiological Features 127
1.2.1 Readiness Potential 128
1.2.2 Sensorimotor Rhythms 129
2 Processing and Machine Learning Techniques 129
2.1 Common Spatial Patterns Analysis 130
2.2 Regularized Linear Classification 131
2.2.1 Mathematical Part 131
3 BBCI Control Using Motor Paradigms 133
3.1 High Information Transfer Rates 133
3.2 Good Performance Without Subject Training 135
3.3 BCI Illiteracy 136
4 Applications of BBCI Technology 138
4.1 Prosthetic Control 138
4.2 Time-Critical Applications: Prediction of Upcoming Movements 139
4.3 Neuro Usability 140
4.4 Mental State Monitoring 141
4.4.1 Experimental Setup for Attention Monitoring 142
4.4.2 Results 143
5 Conclusion 143
References 145
Practical Designs of BrainComputer Interfaces Based on the Modulation of EEG Rhythms 149
1 Introduction 149
1.1 BCIs Based on the Modulation of Brain Rhythms 149
1.2 Challenges Confronting Practical System Designs 151
2 Modulation and Demodulation Methods for Brain Rhythms 152
2.1 Power Modulation/Demodulation of Mu Rhythm 153
2.2 Frequency Modulation/Demodulation of SSVEPs 154
2.3 Phase Modulation/Demodulation of SSVEPs 155
3 Designs of Practical BCIs 156
3.1 Designs of a Practical SSVEP-based BCI 157
3.1.1 Lead Position 157
3.1.2 Stimulation Frequency 158
3.1.3 Frequency Feature 158
3.2 Designs of a Practical Motor Imagery Based BCI 159
3.2.1 Phase Synchrony Measurement 160
3.2.2 Electrode Layout 161
4 Potential Applications 162
4.1 Communication and Control 162
4.2 Rehabilitation Training 163
4.3 Computer Games 164
5 Conclusion 164
References 165
BrainComputer Interface in Neurorehabilitation 167
1 Introduction 167
2 Basic Research 169
3 BrainComputer Interfaces for Communication in Complete Paralysis 169
4 BrainComputer Interfaces in Stroke and Spinal Cord Lesions 172
5 The Emotional BCI 175
6 Future of BCI in Neurorehabilitation 178
References 179
Non Invasive BCIs for Neuroprostheses Control of the Paralysed Hand 182
1 Introduction 182
1.1 Spinal Cord Injury 182
1.2 Neuroprostheses for the Upper Extremity 182
2 Brain-Computer Interface for Control of Grasping Neuroprostheses 185
2.1 Patients 186
2.2 EEG Recording and Signal Processing 188
2.3 Setup Procedures for BCI Control 188
2.3.1 BCI-Training of Patient TS Using a Neuroprosthesis with Surface Electrodes 189
2.3.2 BCI-Training of Patient HK Using an Implanted Neuroprosthesis 190
2.4 Interferences of Electrical Stimulation with the BCI 190
2.5 Evaluation of the Overall Performance of the BCI Controlled Neuroprostheses 191
3 Conclusion 191
References 193
BrainComputer Interfaces for Communication and Control in Locked-in Patients 196
1 Introduction 196
2 Locked-in the Body and Lock-Out of Society 197
3 BCI Applications for Locked-in Patients 199
4 Experiences of a BCI User 202
5 BCI Training with Patients 205
6 Conclusion 208
References 210
Intracortical BCIs: A Brief History of Neural Timing 213
1 Introduction 213
2 Why Penetrate the Brain? 213
3 Neurons, Electricity, and Spikes 215
4 The Road to Imperfection 217
5 A Brief History of Intracortical BCIs 219
6 The Holy Grail: Continuous Natural Movement Control 223
7 What Else Can We Get from Intracortical Microelectrodes? 226
References 228
BCIs Based on Signals from Between the Brain and Skull 230
1 Introduction 230
2 Electrocorticogram: Signals from Between the Brain and Skull 230
3 Advantages of ECoG 231
3.1 Advantages of ECoG Versus EEG 232
3.2 Advantages over Microelectrodes 233
3.3 Everything Affects the Brain 235
4 Disadvantages of ECoG 235
5 Successful ECoG-Based BCI Research 237
6 Past and Present ECoG Research for BCI 238
6.1 ECoG Animal Research 239
6.2 Human ECoG Studies 239
6.2.1 Smith-Kettlewell Eye Research Institute 239
6.2.2 The University of Michigan -- Ann Arbor (Levine and Huggins) 239
6.2.3 The University of Washington in St. Louis 242
6.2.4 University of Wisconsin -- Madison 243
6.2.5 Tuebingen, Germany 244
6.2.6 University Hospital of Utrecht 244
6.2.7 The University of Michigan -- Ann Arbor (Kipke) 244
6.2.8 University of Florida -- Gainesville 245
6.2.9 Albert-Ludwigs-University, Freiburg, Germany 245
7 Discussion 245
References 246
A Simple, Spectral-Change Based, Electrocorticographic BrainComputer Interface 249
1 Introduction 249
2 Signal Acquisition 249
3 Feature Selection 254
4 Feedback 258
5 Learning 261
6 Case Study 262
7 Conclusion 264
References 264
Using BCI2000 in BCI Research 267
1 Introduction 267
1.1 Proven Components 268
1.2 Documentation 269
1.3 Adaptability 269
1.4 Access 269
1.5 Deployment 269
2 BCI2000 Design 269
2.1 System Model 270
2.2 Software Components 273
2.3 Interfacing Components 274
2.3.1 Data Formats 274
2.3.2 Data Exchange 275
2.3.3 Matlab Filter Scripts 275
2.3.4 Online Data Exchange 276
2.3.5 Operator Module Scripting 276
2.4 Important Characteristics of BCI2000 276
2.5 Getting Started with BCI2000 277
3 Research Scenarios 277
3.1 BCI Classroom 277
3.1.1 EEG Hardware 278
3.1.2 Software 278
3.1.3 Getting Acquainted 278
3.1.4 Tutorial Experiments 279
3.2 Performing Psychophysiological Experiments 279
3.3 Patient Communication System 280
3.4 Multi-Site Research 282
4 Research Trajectories 284
5 Dissemination and Availability 284
References 285
The First Commercial BrainComputer Interface Environment 288
1 Introduction 288
2 Rapid Prototyping Environment 290
2.1 Biosignal Amplifier Concepts 290
2.2 Electrode Caps 296
2.3 Programming Environment 296
2.4 BCI System Architectures 299
3 BCI Training 300
3.1 Training for a Motor Imagery BCI Approach 300
3.2 Training with a P300 Spelling Device 302
4 BCI Applications 304
4.1 IntendiX 304
4.2 Virtual Reality Smart Home Control with the BCI 305
4.3 Avatar Control 308
References 309
Digital Signal Processing and Machine Learning 311
1 Architecture of BCI systems 311
2 Preprocessing 313
2.1 Spatial Filtering 313
2.1.1 Linear Transformations 313
2.1.2 Common Average Reference (CAR) 314
2.1.3 Laplacian Reference 315
2.1.4 Principal Component Analysis (PCA) 316
2.1.5 Independent Component Analysis (ICA) 317
2.1.6 Common Spatial Patterns (CSP) 318
2.2 Temporal Filtering 319
3 Feature Extraction 319
3.1 SSVEP-based BCIs 320
3.2 The P300-based BCI 320
3.3 ERD/ERS-based BCI 321
3.3.1 Power Feature Extraction Based on Band-Pass Filter 321
3.3.2 Autoregressive Model Coefficients 322
4 Feature Selection 322
4.1 Channel Selection 323
4.2 Frequency Band Selection 323
5 Translation Methods 324
5.1 Classification Methods 324
5.1.1 Fisher Linear Discriminant 324
5.1.2 Support Vector Machine 326
5.2 Regression Method 327
6 Parameter Setting and Performance Evaluation for a BCI System 327
6.1 K--folds Cross-Validation 328
6.2 Performance Evaluation of a BCI System 329
6.2.1 Speed and Accuracy 329
6.2.2 Information Transfer Rate 329
6.2.3 ROC Curve 329
7 An Example of BCI Applications: A P300 BCI Speller 331
8 Summary 333
References 333
Adaptive Methods in BCI Research - An Introductory Tutorial 337
1 Introduction 337
1.1 Why We Need Adaptive Methods 337
1.2 Basic Adaptive Estimators 339
1.2.1 Mean Estimation 339
1.2.2 Variance Estimation 341
1.2.3 Variance-Covariance Estimation 341
1.2.4 Adaptive Inverse Covariance Matrix Estimation 342
Kalman Filtering and the State Space Model 343
1.3 Feature Extraction 345
1.3.1 Adaptive Autoregressive Modeling 345
1.4 Adaptive Classifiers 347
1.4.1 Adaptive QDA Estimator 347
1.4.2 Adaptive LDA Estimator 348
1.5 Selection of Initial Values, Update Coefficient and Model Order 350
1.6 Experiments with Adaptive QDA and LDA 352
1.7 Discussion 357
References 358
Toward Ubiquitous BCIs 362
1 Introduction 362
2 Key Factors in BCI Adoption 363
2.1 BCI Catalysts 364
2.2 Cost 367
2.3 Information Transfer Rate (ITR) 369
2.4 Utility 370
2.5 Integration 373
2.6 Appearance 378
3 Other Incipient BCI Revolutions 380
3.1 Funding 380
3.2 User Groups Today 381
3.3 User Groups Tomorrow 382
4 BCI Ethics Today and Tomorrow 384
References 388
Index 393