Signal Processing Methods for SSVEP-Based BCIs

Abstract

Frequency coded SSVEP-based BCIs have been increasingly studied in recent years. They have shown potential as useful tools for disabled people to restore fundamental skills of communication and control. Distinguishing target frequency components from weak and noisy SSVEPs with high accuracy using a minimum of recording electrodes is one of the key issues for a practical SSVEP-based BCI. The most challenging task is to effectively eliminate the artefacts whose frequency spectra usually overlap with those of the target signals. In this chapter, a new signal processing method based on the adjacent narrow band filter (ANBF) is proposed for the purpose of artefact reduction and frequency recognition in a 12-class SSVEP-based BCI. The proposed ANBF method effectively suppresses irrelevant artefacts whose frequency spectra overlap with those of the targets, and successfully estimates the noise-free energy of the target frequency bands. The proposed ANBF is compared with the widely used Canonical Correlation Analysis (CCA) and verified online with two channel EEG data from nine healthy participants. This study was done without preventing participants’ normal eye blinks and high performance can be achieved with no more than two electrodes, the proposed ANBF provides a new approach for SSVEP-based BCIs for real-life use.