Abstract
The EEG frequency bands are brain rhythms that indicate the activity level of the brain. This paper investigates the effects of the sub-band frequency on the classification of motor imagery of hand movements. Ten sub-bands of MHz width between 0 to 100 H_ are chosen. Band power features of the sub-bands are classified using a neural classifier. Motor imagery signals recorded from the C3 and C4 channels for four tasks are used in the analysis. Classification rates of 89.23% - 94.47% were achieved for sub-band frequencies ranging from 21HZ to 40 H_ for motor imagery signals. Results show that apart from mu and beta, low gamma frequencies are also better suited.for motor imagery classification
