A Wearable Ultra-Low-Power sEMG-Triggered Ultrasound System for Long-Term Muscle Activity Monitoring

TL;DR

This paper presents a wearable, ultra-low-power system combining sEMG and ultrasound for long-term muscle activity monitoring, using an EMG-driven wake-up to save energy and improve measurement accuracy.

Contribution

It introduces an integrated, energy-efficient wearable system that synchronizes sEMG and ultrasound modalities with an EMG-triggered US activation for enhanced muscle monitoring.

Findings

Achieves over 59% energy savings compared to continuous operation.

Maintains accurate muscle activity detection with synchronized sEMG and US.

Operates with a full-system average power of 12.2 mW.

Abstract

Surface electromyography (sEMG) is a well-established approach to monitor muscular activity on wearable and resource-constrained devices. However, when measuring deeper muscles, its low signal-to-noise ratio (SNR), high signal attenuation, and crosstalk degrade sensing performance. Ultrasound (US) complements sEMG effectively with its higher SNR at high penetration depths. In fact, combining US and sEMG improves the accuracy of muscle dynamic assessment, compared to using only one modality. However, the power envelope of US hardware is considerably higher than that of sEMG, thus inflating energy consumption and reducing the battery life. This work proposes a wearable solution that integrates both modalities and utilizes an EMG-driven wake-up approach to achieve ultra-low power consumption as needed for wearable long-term monitoring. We integrate two wearable state-of-the-art (SoA) US and ExG biosignal acquisition devices to acquire time-synchronized measurements of the short head of the biceps. To minimize power consumption, the US probe is kept in a sleep state when there is no muscle activity. sEMG data are processed on the probe (filtering, envelope extraction and thresholding) to identify muscle activity and generate a trigger to wake-up the US counterpart. The US acquisition starts before muscle fascicles displacement thanks to a triggering time faster than the electromechanical delay (30-100 ms) between the neuromuscular junction stimulation and the muscle contraction. Assuming a muscle contraction of 200 ms at a contraction rate of 1 Hz, the proposed approach enables more than 59% energy saving (with a full-system average power consumption of 12.2 mW) as compared to operating both sEMG and US continuously.