A time-frequency analysis of event-related desynchronization/synchronization based on Gabor filter

An alternative method is proposed based on Gabor filter to make a time-frequency analysis of the event-related (de)synchronization (ERD/ERS). To develop an alternative time-frequency method for better extracting the feature information underlying the ERD/ERS during motor imagery, the Gabor filter is used in the algorithm to extract the amplitude information of the EEG in each frequency band. Then the inter-trail variance method is used to quantify the ERD/ERS. At last, the time-frequency map is shown to make a comprehensive analysis of hand motor imagery. The result shows a contralateral dominant alpha and beta ERD, and a beta ERS bilaterally. The present method is tested with EEG data from BCI competition III dataset IIIa. The feasibility of this method is proved and a compare with band-pass filter-based method is presented.