A Generative Approach to EEG Source Separation, Classification and Artifact Correction

Helle Henriksen

AbstractThis thesis deals with the detection of right and left hand-pull stimuli in EEG data for five healthy subjects. This paradigm give rise to activation of motor cortex contra-lateral to stimuli side.
ICA components obtained from a Kalman filter based algorithm have been applied as features in the classification task and compared with time series features and Infomax ICA features. The Kalman ICA components have proven to be well-suited for separating the two classes in this thesis, and the Kalman features accomplished the lowest error rates when classifying left and right stimuli. Different classifiers have been tested on the three feature types, and the advanced SVM classifier performed best in all cases. The percentage of significant different features between the two classes showed to be strongly correlated to the classification performance. For the purpose of stimuli detection a visual inspection of the ICA components has been made. The visible distinction is not as pronounced as the difference in classification performance for the two ICA features.
TypeMaster's thesis [Academic thesis]
PublisherTechnical University of Denmark, DTU Informatics, E-mail: reception@imm.dtu.dk
AddressAsmussens Alle, Building 305, DK-2800 Kgs. Lyngby, Denmark
NoteSupervised by Associate Professor Ole Winther, owi@imm.dtu.dk, DTU Informatics
Electronic version(s)[pdf]
Publication linkhttp://www.imm.dtu.dk/English.aspx
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing

Back  ::  IMM Publications