Enhancement of Performance of an EEG-based Brain-Computer Interface by means of a Time-Frequency Approach

We have developed a novel time-frequency approach of classification of motor imagery (MI) tasks for brain-computer interface (BCI) applications. Through off-line data analysis on data collected during a "cursor control" experiment, we evaluated the capability of our proposed method in revealing the major features of the EEG control and enhancing MI classification accuracy. The pilot results in two human subjects are promising, with a mean accuracy rate of 87.9%, suggesting the feasibility of defining a new and more reliable EEG-based BCI