Study on EEG signal processing for visual P300-speller

In recent years, the patients with motor dysfunction have become a special social group with increasing number that needs more concerns and cares. Aimed at designing a new brain-computer interface for such patients, this paper proposed a visual P300-speller system with a matrix containing letters and numbers and developed its EEG signal processing algorithm. ABSVM-RFE method was used for channel selection to shorten time of calculations. The proposed system has been validated through offline and online experiments. The highest accuracy of the offline experiments was 99%. In the online tests, EEG data were analysed using only one channel with three subjects and the system achieved a data transfer rate of 7.27 bits/min with a highest accuracy of 80%. It may provide technique support for the effective exploration of voluntary information output channels for those severe disables.