RESUMO
In this study EEG patterns ofsensorimotor rhythm were examined in 10 healthy subjects while perform- ing motor imagery of upper arm and hand movements. Participants received visual feedback through so called brain-computer interface (BCI) used for detection of user-specific spatio-temporal.EEG patterns associated with performed mental tasks. During the course study,.all of the subjects were able to modulate their sensorimotor EEG by performing motor imagery of shoulder and fingers movements. Patterns during imagery of shoulder movements were found to have more pronounced contralateral localization, than those during hand movements' imagery. That led to significantly better classification accuracies of the most lateralized patterns when discriminating between left and right hand (72 and 58% corresponding to shoulder and hand motor imagery). Value of difference of patterns' lateralization indexes had shown strong correlation with classification accuracy, suggests it could be used as a good ref- erence mark for.choosing optimal motor imagery tasks for BCI application.