Bhattacharyya bound based channel selection for classification of motor imageries in EEG signals

In EEG-based brain computer interfaces (BCIs), channel selection is important for the classification of mental task, such as motor imagery. In this paper, a channel selection method is presented for motor imagery. The Bhattacharyya bound of common spatial pattern (CSP) features is used as the optimal index, and a fast sequential forward search is applied to find the optimal combination of channels. The data analysis results show the improvement of classification accuracy.