Proficiency in motor imagery is linked to the lateralization of focused ERD patterns and beta PDC

Background: Motor imagery based brain-computer interfaces (MI-BCIs) are systems that detect the mental rehearsal of movement from brain activity signals (EEG) for controlling devices that can potentiate motor neurorehabilitation. Considering the problem that MI proficiency requires training and it is not always achieved, EEG desirable features should be investigated to propose indicators of successful MI training. Methods: Nine healthy right-handed subjects trained with a MI-BCI for four sessions. In each session, EEG was recorded for 30 trials that consisted of a rest and a dominant-hand MI sequence, which were used for calibrating the system. Then, the subject participated in 160 trials in which a cursor was displaced on a screen by performing MI or relaxing to hit a target. The session’s accuracy was calculated. For each trial from the calibration phase of the first session, the power spectral density (PSD) and the partial directed coherence (PDC) of the rest and MI EEG segments were obtained to estimate the event-related synchronization changes (ERS) and the connectivity patterns of the θ, α, β and γ bands that are associated with high BCI control (accuracy above 70% in at least one session). Finally, t-tests and rank-sum tests (p<0.05, with Benjamini-Hochberg correction) were used to compare the ERS/ERD and PDC values of subjects with high and low accuracy, respectively. Results: Proficient users showed greater α ERD on the right-hand motor cortex (left hemisphere). Furthermore, the β PDC related to the ipsilateral motor cortex is commonly weakened during motor imagery, while the contralateral motor cortex γ PDC is enhanced. Conclusions: Motor imagery proficiency is related to the focused and lateralized event-related α desynchronization patterns and the lateralization of β and γ PDC. Future analysis of these features could allow complimenting the information for assessment of subject-specific BCI control and the prediction of the effectiveness of motor-imagery training.