Reliability of Surface EMG Decomposition Across Accuracy Thresholds During Isometric and Isokinetic Contractions

Accurate decomposition of surface electromyographic (EMG) signals is critical for understanding neuromuscular function during dynamic contractions. However, the reliability of decomposition-based motor unit (MU) metrics across accuracy thresholds and contraction intensities remains unclear. PURPOSE: This study evaluated the test–retest reliability of decomposition-derived MU firing behaviors during isometric and isokinetic contractions. METHODS: Eighteen resistance-trained males (21.1 ± 1.3 yrs) completed four separate sessions of randomized isometric and isokinetic biceps brachii contractions (30% - 100% MVC) on a Biodex System 4. Surface EMG was recorded with a Delsys Trigno Galileo sensor (2 kHz) and decomposed using Precision Decomposition III (PDIII). Identified MUs were filtered using DSDC thresholds (≥0%, 80%, 90%, 91%, 92%). The MU count and the slope and y-intercept of the mean firing rate (MFR)–MU action potential (MUAP) relationships were evaluated using two-way random-effects ICCs (absolute agreement) and 95% CIs for test–retest reliability. RESULTS: Detailed ICCs (95% CIs) are provided in Table 1. Reliability varied across thresholds and contraction types. During isokinetic contractions, slope ICCs were highest at 80% (0.74 [0.69–0.98]) across intensities, while intercepts were moderate (0.55–0.81 [0.52–0.98]). MU count reliability declined at >90% (0.28–0.57 [0.19–0.98]). Isometric slopes and intercept reliabilities were moderate at 80% (slope = 0.69 ± 0.27; intercept = 0.66 ± 0.16). Stricter thresholds (≥91–92%) consistently reduced reliability and MU yield. CONCLUSION: The PDIII-based decomposition yields moderate to good reliability for MFR–MUAP slopes and intercepts across contraction types. An 80% DSDC threshold optimizes reproducibility and MU yield, while stricter criteria diminish MU yield without improving accuracy.