Multimodal ultrasound and surface electromyography assessment of training-dependent biceps brachii fatigue during resistance exercise

Objective assessment of localized muscle fatigue may support exercise prescription, rehabilitation monitoring, and sports performance management. Surface electromyography (sEMG) reflects neuromuscular activation, whereas B-mode and Doppler ultrasonography provide complementary information on muscle morphology and vascular response. This study evaluated training-dependent biceps brachii fatigue using ultrasound and sEMG and examined whether combined dynamic recordings improved fatigue-state classification. Forty healthy men were allocated to an advanced resistance-training group ( n = 20) or a novice group ( n = 20). Participants performed seated right-arm 10-kg dumbbell curls until voluntary exhaustion. Because the same absolute load represented different relative intensities across participants and groups, training-status comparisons were interpreted accordingly. Biceps brachii muscle thickness, anterior circumflex humeral artery peak systolic velocity (PSV), sEMG root mean square (RMS), and median frequency (MDF) were measured at baseline, moderate fatigue, and exhaustion. Stage-matched ultrasound video and sEMG recordings from 35 artifact-free participants were evaluated by participant-level five-fold internal cross-validation. Advanced trainees completed more repetitions than novices (16.3 ± 2.7 vs. 13.3 ± 2.3 repetitions, p < 0.001). Baseline-to-exhaustion muscle-thickness change was smaller in advanced trainees than in novices (12.69 ± 4.67% vs. 18.14 ± 5.87%; mean difference − 5.44% points, 95% CI -8.84 to -2.04; p = 0.003). PSV increased with fatigue, but the exhaustion value did not differ between groups (mean difference − 1.35 cm/s, 95% CI -7.78 to 5.07; p = 0.672). RMS increased with fatigue, whereas MDF showed an inverted-U trajectory. Participant-level internal cross-validation yielded an exploratory three-class accuracy of 86.23% and a macro-F1 score of 86.32%. Ultrasound and sEMG captured complementary morphological, hemodynamic, and electrophysiological associations during localized biceps brachii fatigue. The exploratory internal-validation results suggest potential value for multimodal fatigue-state recognition; they do not establish deployment readiness and require confirmation with individualized loading, larger and more diverse samples, and external validation.