EUVIMEDEUVIMED entdecken
Crossref

Prototype-Guided Test-Time Adaptation for Real-Time EMG-Based Gesture Recognition

Yingtong Ma iD, Yalin Wang iD, Chenyun Dai iD, Yao Guo iD

International Journal of Neural Systems · 2027-04 · 37 (04)

DOI: 10.1142/s0129065727500249 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Neural Systems, online erschienen am 10.10.2026. Verfasst von Yingtong Ma und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Information Science and Technology, Fudan University, Shanghai 200240, P. R. China. Hauptthemen laut MEDLINE: Electromyography, Gestures, Pattern Recognition, Automated, Signal Processing, Computer-Assisted. Die Arbeit nennt 45 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Gansu Province Science and Technology Program.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Surface electromyography (sEMG) signals facilitate intuitive human–machine interaction and are widely employed in applications such as prosthetics and rehabilitation. However, sEMG signal distributions are highly sensitive to variations in electrode placement, user physiology, and session conditions, which significantly hinder model generalization in real-world settings. To address this, we propose EMG-based structure-aware test-time adaptive refinement (EMG-STAR), a test-time adaptation (TTA) framework tailored for online deployment without access to target data during training. EMG-STAR improves model stability by fusing historical and batch-wise normalization statistics from both source and target domains, effectively alleviating domain shifts. Furthermore, it refines predictions using a prototype-assisted feature clustering method that enforces consistency among semantically similar target samples. To further enhance reliability, the framework jointly leverages prediction confidence and distributional alignment to filter out noisy pseudo-labels during adaptation. Comprehensive evaluations conducted on two public datasets, Hyser, and NinaPro-DB2, demonstrate that EMG-STAR consistently outperforms ten existing TTA methods. Interpretability analysis using Grad-CAM reveals that EMG-STAR predominantly attends to physiologically meaningful muscle activation regions, suggesting that the model captures neuromuscular structures rather than transient signal noise.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ElectromyographyGesturesPattern Recognition, AutomatedSignal Processing, Computer-AssistedHumansAdaptive AlgorithmsMuscle, Skeletal

Stichwörter der Autor:innen

Surface electromyographybatch normalizationdomain generalizationhuman–machine interactionprototype learningpseudo-labelingtest-time adaptationwearable sensing

Autor:innen und Institutionen

  • School of Information Science and Technology, Fudan University, Shanghai 200240, P. R. China.
  • School of Information Science and Engineering, Lanzhou University, Lanzhou 730000, P. R. China.
  • School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai 200240, P. R. China.
  • School of Rehabilitation Sciences and Engineering, University of Health and Rehabilitation Sciences, Qingdao, Shandong 266113, P. R. China.

ORCID-Profile: Yingtong Ma, Yalin Wang, Chenyun Dai, Yao Guo

Förderung

  • National Natural Science Foundation of China
  • Gansu Province Science and Technology Program

Zitieren

APA

Ma, Y., Wang, Y., Dai, C., & Guo, Y. (2027). Prototype-Guided Test-Time Adaptation for Real-Time EMG-Based Gesture Recognition. International Journal of Neural Systems, 37(04). https://doi.org/10.1142/s0129065727500249

Vancouver

Ma Y, Wang Y, Dai C, Guo Y. Prototype-Guided Test-Time Adaptation for Real-Time EMG-Based Gesture Recognition. International Journal of Neural Systems. 2027;37(04). doi:10.1142/s0129065727500249

BibTeX
@article{Ma2027,
  author = {Ma, Yingtong and Wang, Yalin and Dai, Chenyun and Guo, Yao},
  title = {Prototype-Guided Test-Time Adaptation for Real-Time EMG-Based Gesture Recognition},
  journal = {International Journal of Neural Systems},
  year = {2027},
  volume = {37},
  number = {04},
  doi = {10.1142/s0129065727500249}
}
Kostenlos starten →