Prediction of vocalization using time-frequency information of mu rhythm

Vocal control is one of the most important motor function in human. When the function is impaired due to lesions, it may be useful way to communicate with others through Brain Computer Interface (BCI). In this study, we investigated a possibility of BCI for speech communication using rhythmic brain activity in brain. We measured event related de-synchronization (ERD) through magnetoencephalography (MEG) when the subject performed either vocalization, or observation of video of vocalization. After transforming the measured signal to time-frequency representation (TFR), we tested whether the TFR could predict the subject´s actual behavior, that is, vocalization or just looking at video of vocalization. Preliminary result showed that the TFR could distinguish the subject´s tasks at a certain degree of accuracy. This result implies that the BCI for communication has possible applications for people who are impaired vocal functions.