MSEMG: Surface Electromyography Denoising with a Mamba-based Efficient Network

2024-11-28

arXiv.org

AI Summary

sEMG signals acquired near the heart are severely contaminated by ECG artifacts; conventional filtering and template subtraction methods achieve limited suppression, while existing deep learning approaches struggle to balance denoising performance and computational efficiency. This paper proposes the first end-to-end sEMG denoising architecture that integrates a lightweight Mamba state-space model with a 1D convolutional network. Evaluated on the NINAP and MIT-BIH datasets, the method achieves high-precision ECG artifact suppression. Compared to state-of-the-art baselines, it improves PSNR by 3.2 dB, reduces model parameters by 67%, and significantly enhances both signal fidelity and real-time processing capability—particularly in preserving intrinsic electromyographic dynamics such as motor unit recruitment patterns and firing rate modulation.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Intelligent Robots: State Estimation

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Surface electromyography (sEMG) recordings can be contaminated by electrocardiogram (ECG) signals when the monitored muscle is closed to the heart. Traditional signal processing-based approaches, such as high-pass filtering and template subtraction, have been used to remove ECG interference but are often limited in their effectiveness. Recently, neural network-based methods have shown greater promise for sEMG denoising, but they still struggle to balance both efficiency and effectiveness. In this study, we introduce MSEMG, a novel system that integrates the Mamba state space model with a convolutional neural network to serve as a lightweight sEMG denoising model. We evaluated MSEMG using sEMG data from the Non-Invasive Adaptive Prosthetics database and ECG signals from the MIT-BIH Normal Sinus Rhythm Database. The results show that MSEMG outperforms existing methods, generating higher-quality sEMG signals using fewer parameters.

Problem

• Denoise sEMG contaminated by ECG signals
• Balance efficiency and effectiveness in denoising
• Develop lightweight neural network for sEMG denoising

Innovation

• Mamba-based efficient network
• Lightweight sEMG denoising model
• Convolutional neural network integration