Enhanced Single Channel SSVEP Detection Method on Benchmark Dataset

Abstract

Steady state visual evoked potential (SSVEP) is a brain response that allows a practical and high-performance brain-computer interface (BCI) to be designed. SSVEP response is a near sinusoidal waveform at a visual stimulus frequency and is time-locked to stimulus onset. This paper presents a new single channel SSVEP detection method that takes advantage of the behaviour of SSVEP response. The proposed method defines subject-specific sinusoids at the training stage. Detection of a target stimulus frequency is achieved by a correlation value between the electroencephalography (EEG) signal and subject specific sinusoids at the test stage. The performance of the developed method was compared with the well-known power spectral density analysis (PSDA) on a benchmark dataset. Experimental results show that the developed method significantly improves the SSVEP detection accuracy (by about 23%) as well as the information transfer rate (ITR) compared to PSDA methods.