Abstract
In this study, we target end-users who are disabled persons and healthy persons whose hands are occupied. We select a steady-state visually evoked potential (SSVEP) based interface that requires no user training. The user interface menu has six stimuli with different frequencies (20-30 Hz), and the intensities are modulated sinusoidally and presented on a liquid crystal display. To reduce the risk of photosensitive epileptic seizure, we select the middle-band range of frequencies. We introduce a commercial headset with 14 channels, which decreases the time needed to attach electrodes. To extract and remove ocular-movement related components in the EEG data, we used a previously obtained un-mixing matrix of independent component analysis. Further, we use the components to detect the eye-fixation points, where the decision making was considered to have been complete, and start the modulation of the stimuli. We calculate the mean SSVEP amplitudes obtained from two occipital regions by averaging EEG data using triggers that correspond to six frequencies. To determine the gazed stimuli using the Z-scored mean amplitudes, we propose a simple algorithm that requires no learning.
