Considerations on EEG-based BCIs Applied in the Control of Intelligent Prostheses
G.M. Ungureanu and R. Strungaru
In this study, we investigate the representation, analysis and classification of the EEG data recorded from four subjects during motor imagery. Four classes of imagery are recorded: left hand, right hand, the tongue and one foot, with 40 trials each. Knowing the locations of the electrodes and the brain regions where the activation related with this four motor imagery tasks occurs, we apply the time-frequency analysis for the corresponding EEG signals. For separating the source signals, the Independent Component Analysis (ICA) based on JADE algorithm is applied. Then we represent the extracted components with the time-frequency analysis, using the spectrogram and the wavelet transform. Finally, we compare the results obtained for these methods with that obtained from the Event Related Spectral Perturbation (ERSP), with the EEGLAB application, to identify the Event Related Synchronization/Desynchronization (ERS/ERD) of the miu band when the imagination of movement starts.
Keywords: Motor imagery, EEG, event related synchronization/desynchronization, signal processing, diagnosis, time-frequency analysis, ICA.