Decoding speed of hand movement execution using temporal features of EEG

Electroencephalography (EEG) processing methods mostly focus on extracting its spectral or spatial features, which are proven to discriminate bilateral hand movement, hand movement directions and speed. The focus of current study is to explore EEG time-domain features that represent neural correlates of hand movement execution speed. In this paper, we propose autocorrelation analysis of EEG and features derived from it that utilizes difference in execution time of fast v/s slow tasks. The variation in decay constant of autocorrelation of EEG over execution time is studied, and its application as a potential feature to discriminate movement speed is explored. The proposed analysis method has been validated on EEG data recorded from 7 subjects performing right hand movement at two different speeds. An average classification accuracy of 75.71% and 85.16% is obtained, using features derived from significant time segments in the data.