Cyclostationary-based detection of steady-state visually evoked potential signals recorded from EEG

Abstract

Steady-state visual evoked potentials (SSVEP) are a class of signals obtained from the electroencephalogram (EEG) that are used in conjunction with brain-computer interfaces (BCIs). Inducing SSVEP signals requires flickering lights as stimuli, typically in the range of 5–45 Hz. However, due to low signal-to-noise ratio (SNR), SSVEP signals generated in certain frequency ranges can be difficult to detect. This paper studies cyclostationary-based detection for SSVEPs, which is a popular method for signal detection in low SNR environments, but whose application in the context of BCI systems has received only limited attention in the BCI research community. The results presented in the paper demonstrate that cyclostationary-based detection of SSVEP using spectral correlation density (SCD) performs as well as canonical correlation analysis (CCA), which is the most widely used method of SSVEP classification.