Impact of Motor Unit Action Potential Components on the Motor Unit Identification from Dynamic High-Density Surface Electromyograms

We assessed the impact of different motor unit action potential (MUAP) components in dynamic muscle contractions on decomposition of high-density surface electromyograms (hdEMG). In particular, hypothesis that nontravelling MUAP components, originating from the tendon regions, are less sensitive to changes in geometry of fusiform muscles than travelling MUAP components has been tested on synthetic monopolar hdEMG signals. The latter have been decomposed by previously introduced Convolution Kernel Compensation (CKC) method, using five different sections of simulated MUAPs for motor unit identification. Accuracy of decomposition results increased significantly when motor units were identified from the nontravelling MUAP components, compared to the results obtained from travelling components. Average motor unit identification sensitivity increased from 67.4%+/- 15.7% to 81.3%+/- 11.3% and false alarm rate decreased from 0.75%+/- 1.21% to 0.20%+/- 0.24%. Results confirmed that non-travelling MUAP components are discriminative enough to reliably identify motor units from hdEMG and less sensitive to geometric changes of fusiform muscles during dynamic muscle contractions than travelling MUAP components.