Characterization and Identification of Dependence in EMG Signals from Action Potentials and Random Firing Patterns

Abstract

Electromyographic (EMG) signals are biomedical signals that represent neuromuscular activities. The EMG signal is neither stationary nor periodic and exhibits complex interference patterns of several single motor unit action potentials (SMUAPs). This study aims to characterize EMG signals concerning firing patterns and other characteristics and to identify whether these MUAP firing patterns present short-range dependencies (SRD) or long-range dependencies (LRD). To do so, we characterized 208 EMG signals in terms of the number of phases, turns and combinations of phases. Then, we performed a statistical comparison of the (more efficient) Variance-time plot against the (less bias) Log-scale diagram for the estimation of the Hurst parameter and detection of LRD. Using these estimators, we managed to detect LRD in a sample taken with needle electrodes. In contrast, the tools used for the dependence identification on signals achieved with surface electrodes did not yield conclusive results on such dependence.