Automatic extracting event-related potentials within several trials using Infomax ICA algorithm

Abstract

This study presents a new method for automatic extracting event-related potentials (ERP) within several trials based on independent component analysis (ICA). After mixed data is decomposed by Infomax ICA, independent component (IC) of ERP is automatically selected according to the standard deviation of fixed temporal pattern of IC, and applied in ERP reconstruction. Visual evoked potential extraction, used to confirm effectiveness of algorithm, can be obtained automatically after 6 trials on experimental data, and result of its Pearson correlation coefficient (PCC) within the average of 205 trials (standard signal) is 0.9106. However, PCC of average result of 6 trials within standard signal is only 0.3066, demonstrating practical applicability of proposed method. This algorithm enhances objectivity of ERP extraction within several trials.