Removal of the ocular artifacts from EEG data using a cascaded spatio-temporal processing

Eye movements and blinks may produce unusual voltage changes in human electroencephalogram (EEG). These effects may spread across scalp and mask brain signals. In this paper, a cascaded spatio-temporal processing procedure (CAST) is presented to remove artifact electrooculogram (EOG). Firstly a discrete equivalent distributed source on the cortical surface is reconstructed from the contaminated scalp recordings by a linear minimum norm estimation (i.e. a spatial analysis step). Then, the equivalent sources of EOG are identified by principal component analysis (PCA) of the equivalent distributed source time series (i.e. a temporal analysis step). Finally, the EOG-corrected scalp EEG is reconstructed from the equivalent distributed source where EOG components have been removed. The effectiveness of CAST is confirmed by the application to actual scalp data and a detailed comparative study.