Study of A Brain-Controlled Switch during Motor Imagery

摘要

Most brain-computer interface (BCI) systems use the synchronization paradigm to detect specific brain activities to control external devices. However, for further asynchronous control application it is necessary to provide users with a switch for the system on and off based on spontaneous brain activities. EEG data during motor imagery of right hand movement were collected by 64 electrodes from 4 healthy subjects. After pre-processing the feature related with motion were extracted by common spacial pattern and then by a linear discriminant classifier, the recognition rate of system on/off was about 90% for the offline analysis. Additionally by the maximum redundancy minimum correlation analysis, the most relevant channel was obtained. In the future, the brain-controlled switch may play an important role in brain-computer interface capacities in practical applications.