Frequency-wise optimal duty-cycle selection in steady state visual evoked potentials: A pilot study

Abstract

We propose an optimal duty-cycle selection method for steady state visual evoked potentials (SSVEP) based brain computer interface (BCI) system. In SSVEP based BCI, visual stimuli are usually selected in a low frequency range. However, previous studies showed that a higher frequency range is more visually comfortable than the low frequency range for SSVEP based BCI. In addition, frequencies of 15-25Hz ranges are provocative to cause photosensitivity epileptic seizure problem. We designed an experiment to select an optimal duty cycle visual flicker for each frequency in a range of 27.5Hz to 33.5Hz. We utilized canonical correlation analysis (CCA) in order to extract the frequency component features from multi-channel EEG data. The goal of this study was to select optimal visual flicker within minimum visual fatigue. Visual flickers with the largest amplitude among the duty-cycle higher than 50% were selected as optimum. The performance of the proposed system showed an average accuracy of 90%. In conclusion, we proposed a comfortable SSVEP based BCI system by reducing visual fatigue for the user (or patient), with encouraging results.