A Wavelet-CSP method to classify hand movement directions in EEG based BCI system

Neethu Robinson, A. Prasad Vinod, Cuntai Guan, Kai Keng Ang, Keng Peng Tee

Abstract: The Electroencephalogram (EEG) based Brain Computer Interface (BCI) is a non-invasive system to acquire, decode and convert brain signals into control signals for an external device. The Motor-Imagery based BCI (MI-BCI) efficiently decodes the brain signals from the imagination of movement but the performance is limited by the number of commands such as right and left hand motor imageries. However, other parameters of an actual voluntary movement, such as the direction of movement, speed and extent, are encoded in the brain signals. This paper investigates the EEG brain signals from directional changes in actual hand movement. The Wavelet-Common Spatial Pattern algorithm is proposed to extract discriminative features of the brain signals that carries the direction-related information. The experiment performed on two subjects yielded a mean classification accuracy of 87.85% in decoding two classes of the direction-related information.