Extraction of the Mismatch Negativity on Two Paradigms Using Independent Component Analysis

We compare the efficiency of the independent component analysis (ICA) decomposition procedure against the difference wave (DW) procedure in the analysis of the mismatch negativity (MMN). The comparison was made using two different experimental paradigms under a passive oddball protocol. The event related potential (ERP) recordings were decomposed into the MMN-like and non-MMN-like independent components (ICs) through the FastICA algorithm. Algorithmic reliability of estimated ICs was assessed by the clustering and visualizing tool from the ICASSO software package. The ICA decomposition procedure allowed to extract a cleaner MMN compared to the DW procedure. Mean improvement of the single trial signal-to-noise ratio (SNR) of the MMN was 9.71 and 6.28 dB in the two examined paradigms. Due to this improvement, the ICA decomposition procedure might allow to shorten the recording session of the MMN.