Fast extraction of mismatch negativity based on independent component analysis

Mismatch negativity (MMN refers to auditory evoked potentials (AEP responding to changes in sensory stimuli. MMN detection and extraction are very difficult due to the extremely poor signal-to-noise ratio (SNR. This paper describes an extraction method based on the multi-decomposition of multi-channel auditory evoked potentials by independent component analysis (ICA. The signal characteristics and the physical generation mechanism of MMN were used to design independent component selection principles for MMN extraction. The simulation result shows that the method greatly improves the SNR. In actual EEG data set processing, the method can extract the MMN component in about 20% of the traditional experimental time. The method will promote the application of MMN both in cognitive neural science and clinical practice.