Calibrating Independent Component Analysis with Laplacian Reference for Real-Time EEG Artifact Removal

Independent Component Analysis (ICA) has emerged as a necessary preprocessing step when analyzing Electroencephalographic (EEG) data. While many studies reported on the use of ICA for EEG, most of these studies rely on visual inspection of the signal to detect those components that need to be removed from the signal. Little has been done on how to process EEG data in real-time, autonomously, and independent of a human expert inspecting the data. A few attempts have been made in the literature to design standard procedures on the processing of EEG data in real-time environments. To enable standardization to occur, the work and discussion of this paper focus on understanding the impact of different preprocessing steps on the performance of ICA. A proposed cut-off threshold for ICA is demonstrated to produce reliable and sound processing when compared to a Laplacian reference system. A methodology for real-time processing that is simple and efficient is being suggested.