Rest-to-work transfer of spatial filters for a motor imagery based brain computer interface

An approach based on independent component analysis (ICA) is described and tested. ICA was used for separating Mu and Beta rhythms generated in both hemispheres and for constructing spatial filters in preprocessing the electroencephalographic (EEG) data in brain computer interface (BCI) research. It was proposed a rest-to-work translation of spatial filters for EEG based BCI. Three different ICA algorithms were exploited in order to obtain independent components which computed the feature vector. The classification was performed with linear discriminant classifier and with quadratic classifier. The proposed method is robust, efficient and subject training can be eliminated.