Spatial variation and inconsistency between estimates of onset of muscle activation from EMG and ultrasound

Abstract

Delayed onset of muscle activation can be a descriptor of impaired motor control. Activation onset can be estimated from electromyography (EMG)-registered muscle excitation and from ultrasound-registered muscle motion, which enables non-invasive measurements in deep muscles. However, in voluntary activation, EMG- and ultrasound-detected activation onsets may not correspond. To evaluate this, ten healthy men performed isometric elbow flexion at 20% to 70% of their maximal force. Utilising a multi-channel electrode transparent to ultrasound, EMG and M(otion)-mode ultrasound were recorded simultaneously over the biceps brachii muscle. The time intervals between automated and visually estimated activation onsets were correlated with the regional variation of EMG and muscle motion onset, contraction level and speed. Automated and visual onsets indicated variable time intervals between EMG- and motion onset, median (interquartile range) 96 (121) ms and 48 (72) ms, respectively. In 17% (computed analysis) or 23% (visual analysis) of trials, motion onset was detected before local EMG onset. Multi-channel EMG and M-mode ultrasound revealed regional differences in activation onset, which decreased with higher contraction speed (Spearman ρ ≥ 0.45, P < 0.001). In voluntary activation the heterogeneous motor unit recruitment together with immediate motion transmission may explain the high variation of the time intervals between local EMG- and ultrasound-detected activation onset.

Differences in muscle contraction onset as determined by ultrasound and electromyography

“…Electrical stimulation produces small electromechanical delays around 10 ms and will activate all motor units within the region at once, thus eliminating the hierarchical activation of low‐threshold to high threshold motor units . However, in voluntary contractions, sEMG onset is regionally variable and contraction dependent . Previous studies comparing sEMG and ultrasound often report smaller differences with narrower confidence intervals than our current study.…”

“…As a follow‐up to their previous study, Dieterich et al demonstrated regional differences in sEMG onset using a transparent multi‐channel surface electrode array, suggesting that movement registered by ultrasound may not necessarily equate to electrical activation of the muscle in the sampling region. The authors also noted that ultrasound registered initial motion in the deeper region of the muscle.…”

“…First, synchronization biases between ultrasound and EMG data streams should be addressed. Differences in hardware delays and signal processing can introduce systematic latencies up to 90 ms between system data streams . For the current experimental set‐up, we transmitted a 1‐V analog impulse to the EMG system and to the M‐mode trace through the ECG port of the ultrasound system.…”

“…Immediately before each contraction, a 1‐V square‐wave pulse was transmitted to the ultrasound unit through the ECG port and to the EMG A/D board, enabling off‐line temporal synchronization of the two systems. Excessive synchronization latencies associated with ECG channel filtering have been reported in other commercial ultrasound systems . The current system is set to align all signal paths to have the same temporal alignment with a minimal acceptable temporal variance performance of 50 ms between all real‐time data streams.…”

“…Excessive synchronization latencies associated with ECG channel filtering have been reported in other commercial ultrasound systems. 19,23 The current system is set to align all signal paths to have the same temporal alignment with a minimal acceptable temporal variance performance of 50 ms between all real-time data streams. This is monitored internally with an injected pulse into both the transducer signal path and ECG signal path for both Doppler and M-mode.…”