Development and Validation of an Automated Pipeline for Motor Unit Analysis from HD-sEMG Signals.

High-density surface electromyography (HD-sEMG) enables non-invasive analysis of motor unit (MU) behavior and is crucial in rehabilitation, pathophysiological assessment, and assistive technologies. However, current pipelines still rely on expert intervention, particularly for spike-train editing and channel selection for motor unit conduction velocity (MUCV) estimation. This study developed and validated an automated pipeline for MU analysis from HD-sEMG signals acquired during isometric trapezoidal contractions. The pipeline integrates blind source separation (BSS)-based decomposition, automated source validation and spike-train selection, estimation of discharge rate (DR), recruitment threshold (RT), and automatic channel selection for MUCV. Validation was first performed on a public dataset containing manually edited spike trains from eight operators. The proposed procedure matched 106 of 135 reference MUs with an overall median rate of agreement of 0.98 among matched MUs. Sensitivity analysis supported the robustness of the predefined parameters. The complete pipeline was then evaluated on 276 HD-sEMG signals acquired from the tibialis anterior (TA) muscle of nine healthy volunteers in four sessions using a custom-built acquisition device. The resulting MU parameters were physiologically plausible, and within-session tracking showed high repeatability of MU properties even after electrode repositioning. Validation was limited to healthy volunteers, and further evaluation in clinical populations is needed. These findings support the feasibility of reducing operator dependency while maintaining the quality of HD-sEMG-based MU analysis.