Optimal imaging of cortico-muscular coherence through a novel regression technique based on multi-channel EEG and un-rectified EMG
Cortico-muscular coherence (CMC) reflects interactions between muscular and cortical activities as detected with EMG and EEG recordings, respectively. Most previous studies utilized EMG rectification for CMC calculation. Yet, recent modeling studies predicted that EMG rectification might have disadvantages for CMC evaluation. In addition, previously the effect of rectification on CMC was estimated with single-channel EEG which might be suboptimal for detection of CMC. In order to optimally detect CMC with un-rectified EMG and resolve the issue of EMG rectification for CMC estimation, we introduce a novel method, Regression CMC (R-CMC), which maximizes the coherence between EEG and EMG. The core idea is to use multiple regression where narrowly filtered EEG signals serve as predictors and EMG is the dependent variable. We investigated CMC during isometric contraction of the abductor pollicis brevis muscle. In order to facilitate the comparison with previous studies, we estimated the effect of rectification with frequently used Laplacian filtering and C3/C4 vs. linked earlobes. For all three types of analysis, we detected CMC in the beta frequency range above the contralateral sensorimotor areas. The R-CMC approach was validated with simulations and real data and was found capable of recovering CMC even in case of high levels of background noise. When using single channel data, there were no changes in the strength of CMC estimated with rectified or un-rectified EMG — in agreement with the previous findings. Critically, for both Laplacian and R-CMC analyses EMG rectification resulted in significantly smaller CMC values compared to un-rectified EMG. Thus, the present results provide empirical evidence for the predictions from the earlier modeling studies that rectification of EMG can reduce CMC.