Signal processing methods for EEG data classification
Andreas Varnavas · Spiral (Imperial College London) · 2008
The scope of this thesis is to determine appropriate features of a person's electroencephalographic (EEG) data and the way in which they can be used to predict their performance in an "oddball" experiment.We classify a person's performance in one of the following classes: "success" or "failure", depending on the reaction time related with it.Predicting a person's performance means finding the correct class where the latter belongs to, using the person's EEG data corresponding to a time period before the reaction takes place.The problem is addressed in various ways as far as the feature construction process is concerned, whereas a Gaussian classifier is used in all cases.First, the raw time signals and the magnitude of their Fourier Transform are used as features.Then the number of these features is reduced, using various feature selection methods in combination with Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA) and Non Negative Matrix Factorization (NMF).Subspace methods, using PCA and NMF to construct different spaces for the two classes, are also used to perform the desired classification.Features are also constructed using a Time-Frequency representation of the EEG signals.In this case we propose two novel algorithms which analyze the magnitude of the Time-Frequency representation using NMF, in a single or multi-trial basis, and the coefficients of selected NMF components are used as features.Finally, a novel algorithm performing the desired classification based on the construction of signals characterising each of the classes is proposed.These characteristic signals are constructed linearly combining the EEG signals of the various channels, minimising the variance of the time samples over the trials belonging to the same class.A novel algorithm is also proposed for selecting the appropriate channels to be used in the construction of the characteristic signals.This algorithm is based on the identification of Contents 7 5.3.1 A universal set of channels used 5.3.2Channel selection applied on a single subject basis 5.4 Classification results using the whole characteristic signals with squared weights that sum up to 1 5.4.1 A universal set of channels used 5.4.2Channel selection applied on a single subject basis 5.5 Classification results constructing features weights that sum up to 1 from characteristic signals with 133 5.6 Classification results constructing features from characteristic signals with squared weights that sum up to 1 5.7 Comparison of the proposed methods Chapter 6.A comparison with methods from the field of Human Performance Monitoring 6.1 Methods from the Human Performance Monitoring Field 6.1.1Kernel PCA 6.1.2Support vector classification 6.1.3Classification Results 6.2 Comparison of various methods Chapter 7. Conclusions and future perspectives Bibliography Appendix A. Averages of EEG signals Appendix B. Computation of the confidence interval Appendix C. Significance test List of Figures 1.1 The Electrode nomenclature according to the International Federation of Clinical Neurophysiology's 10-20 system 28 1.2The ERP signal as acquired by averaging 40 EEG signals corresponding to different trials of an oddball experiment, (taken from [51]).29 1.3 Histograms of the reaction time for two red, vertical lines indicate the "medium" and "failure" subjects: S1, S2, S3, S4, S5, S6.The boundaries among classes "success", 38 1.4 Histograms of the reaction time for subjects: S8, S10, S11, S13, S14.The two red, vertical lines indicate the boundaries among classes "success", "medium" and "failure" 39 1.5 Subject 1, The blue and red lines are the average signals over the EEG signals of valid trials for classes "success" (blue) and "failure" (red).The area between the average signal plus and minus one standard deviation is marked with blue ("success") and red ("failure").The vertical line indicates the time of stimulus onset.The signals are truncated at the time point of the subject's quickest reaction.40 1.6 Subject 1.The blue and red lines are the averages of the magnitude of the spectrum over the EEG signals of valid trials for classes "success" (blue) and "failure" (red).The area between the average signal plus and minus one standard deviation is marked with blue ("success") and red ("failure").The spectrum plotted is for EEG signals truncated between the stimulus onset and the time point of the subject's quickest reaction.41 2.1 The adaptive filtering scheme 47 2.2 The parametric model for the observed EEG 49 AR