An Ocular Artifacts Removal Method Based on Canonical Correlation Analysis and Two-channel EEG recordings

Abstract

In order to more effectively apply an artifact removal method in an online Brain-Computer Interface (BCI) system, a new method based on Canonical Correlation Analysis (CCA) and two-channel Electroencephalography (EEG) recordings to quickly remove ocular artifacts (OA) is proposed in this paper. Considering both the formation of EEG signals contaminated by OA and the spread of OA, vertical Electrooculography (VEOG) was appropriately introduced in CCA, and the Blind Source Separation (BSS) method based on CCA was used in a new way during the OA removal process. Both experimental and comparison with ICA and SOBI results show that the new method with simple calculation and fast processing speed can effectively separate and remove OA using only two-channel EEG recordings, with retaining useful EEG signals. Hence, this method used in an online BCI system will be more effective.