Power spectrum of the rectified EMG: Influence of motor unit action potential shapes

The rectification of EMG signals is a preprocessing method widely used for inferring neural connectivity by coherence analysis. The assumption for the use of this non-linear operator is that it enhances the neural information in the signal, i.e. the motor unit spike trains. However, because of non-linearity, it is difficult to predict the effect of rectification on the EMG power spectrum. In this study, we analyze the influence of the motor unit action potential properties on spectral content of the rectified EMG. The results show that changes in the action potential waveforms have a strong influence on the rectified EMG power spectrum, with an effect on estimated coherence functions. Knowledge on the properties of action potentials may be necessary for properly comparing the rectified EMG power spectrum across conditions.