MotorUNet: sim-to-real learning for identifying motor unit action potentials from low-density surface EMG recordings
File(s)
Accepted version
Author(s)
Type
Conference Paper
Abstract
Surface EMG signals are the spatio-temporal summation of motor unit action potentials (MUAPs). Each MUAP reflects the biophysical properties of its originating motor unit—including fibre geometry and conduction velocity, as well as tissue impedance. MUAPs therefore enable direct observation of motor unit physiology and characterization of neuromuscular function. While signal processing techniques such as blind source separation (BSS) have been used to identify MUAPs in high-density (HD) EMG recordings (several tens to hundreds of channels), isolating MU contributions from low-density EMG signals remains a key challenge. Here, we present MotorUNet, an AI model trained on synthetic EMG data generated by a myoelectric digital twin to iteratively recover MUAPs from short windows of low-density EMG signals. On experimental
data, MotorUNet outperformed state-of-the-art BSS methods when applied to spatially downsampled HD-EMG recordings, identifying 17±1 MUAPs vs 2±1 and was able to match 4±2 MUAPs identified on the original HD-EMG recordings. The proposed approach is the first method based on training on simulated signals that reliably identifies MUAPs from low-density EMG recordings, enabling a solution for applications such as muscle activation mapping for biofeedback systems, and inferring
motor unit properties from the extracted MUAPs.
data, MotorUNet outperformed state-of-the-art BSS methods when applied to spatially downsampled HD-EMG recordings, identifying 17±1 MUAPs vs 2±1 and was able to match 4±2 MUAPs identified on the original HD-EMG recordings. The proposed approach is the first method based on training on simulated signals that reliably identifies MUAPs from low-density EMG recordings, enabling a solution for applications such as muscle activation mapping for biofeedback systems, and inferring
motor unit properties from the extracted MUAPs.
Date Acceptance
2026-04-15
Citation
Conference proceedings : ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference
ISSN
1557-170X
Publisher
IEEE
Journal / Book Title
Conference proceedings : ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference
Copyright Statement
Subject to copyright. This paper is embargoed until publication. Once published the author’s accepted manuscript will be made available under a CC-BY License in accordance with Imperial’s Research Publications Open Access policy (www.imperial.ac.uk/oa-policy).
License URL
Source
IEEE EMBC 2026
Publication Status
Accepted
Start Date
2026-07-26
Finish Date
2026-07-30
Coverage Spatial
Toronto, Canada