The Influence of Force Level and Motor Unit Coherence on Nonlinear Surface EMG Features Examined Using Model Simulation

Nonlinear features extracted from surface EMG signals have been previously used to infer information on coherent or synchronous activity in the underlying motor unit discharges. However, it has not yet been assessed how these features are affected by the density of the surface EMG signal, and whether changes in the level of muscle activation can influence the effective detection of correlated motor unit firing. To examine this, a motoneuron pool model receiving a beta-band modulated cortical input was used to generate correlated motor unit firing trains. These firing trains were convolved with motor unit action potentials generated from an anatomically accurate electrophysiological model of the first dorsal interosseous muscle. The sample entropy (SampEn) and percentage determinism (%DET) of recurrence quantification analysis were calculated from the composite surface EMG signals, for signals comprised of both correlated and uncorrelated motor unit firing trains. The results show that although both SampEn and %DET are influenced by motor unit coherence, they are differentially affected by muscle activation and motor unit distribution. The results also suggest that sample entropy may provide a more accurate assessment of the underlying motor unit coherence than percentage determinism, as it is less sensitive to factors unrelated to motor unit synchrony.