A potential real-time procedure to evaluate correlation of recordings among single trials (CoRaST) for mismatch negativity (MMN) with Fourier transformation

Abstract

Objective To design a fast algorithm that evaluates the degree of correlation of recordings among single trials (CoRaST) for mismatch negativity (MMN) activity. Methods The participants were 114 children, aged 8–16 years. MMNs were elicited by two deviants in duration that occurred in an uninterrupted sound within a passive oddball paradigm, and each trial lasted 650 ms with 130 samples. CoRaST was derived from the frequency-domain MMN model through Fourier transformation. To validate the effectiveness of the proposed method, the wavelet transformation-based inter-trial coherence (ITC) was taken as a reference. Results Performances of the proposed CoRaST and ITC were similar in evaluating the correlation of MMN activity among single trials. However, the analysis of electroencephalograph (EEG) recordings comprising approximately 330 trials at one channel took approximately 0.12 s with CoRaST, whereas ITC required approximately 45 s in our workstation. Conclusions CoRaST has the potential to evaluate the correlation of MMN activity among single trials in a real-time system. Furthermore, the new method can be facilitated to study other event-related potentials (ERPs) of evoked or induced brain activity. Significance The correlation of ERP activity among single trials can be immediately inspected during the ERP data collection in a real-time system.