Related Experiment Video
Updated: Feb 28, 2026
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
1.3K
Utilizing Data Quality Indices for Strategic Sensor Channel Selection to Enhance Performance of Hand Gesture
Shen Zhang1, Hao Zhou1, Rayane Tchantchane1
1Applied Mechatronics and Biomedical Engineering Research (AMBER) Group, School of Engineering, University of Wollongong, Wollongong, NSW 2522, Australia.
Sensors (Basel, Switzerland)
|February 27, 2026
Summary
This study introduces a data quality method to select optimal channels for wearable Human-Machine Interface (HMI) systems, significantly improving hand gesture recognition performance using surface electromyography (sEMG) and myography (pFMG) signals.
Area of Science:
- Biomedical Engineering
- Human-Computer Interaction
- Machine Learning
Background:
- Wearable Human-Machine Interface (HMI) systems rely on accurate signal processing for effective control.
- Multi-channel sensor systems, like those using surface electromyography (sEMG) and pressure-based force myography (pFMG), generate complex data.
- Optimizing channel selection is crucial for enhancing the performance and efficiency of these systems, particularly in applications like prosthetic hand control.
Purpose of the Study:
- To develop and evaluate a data quality-driven methodology for selecting optimal sensor channels in multi-channel wearable HMI systems.
- To investigate the relationship between data quality indices and hand gesture recognition accuracy.
- To improve the performance of hand gesture recognition for prosthetic hand applications.
Main Methods:
- Calculated five data quality indices for sEMG and pFMG signals.
- Established correlations between data quality indices and gesture recognition accuracy.
- Employed machine learning (ML)-based and correlation-based methods to select three optimal channel/pair configurations from an eight-channel system.
Main Results:
- The proposed data quality-driven channel selection significantly outperformed random selection on UOW and Ninapro DB2 datasets.
- The ML-based approach achieved the highest accuracy: 76.36% (sEMG), 71.59% (pFMG), and 88.2% (fused sEMG-pFMG) on the UOW dataset.
- Three selected sEMG-pFMG channels achieved 88.2% accuracy, comparable to an eight-channel sEMG system (88.38%), demonstrating high efficiency.
Conclusions:
- Data quality indices are valuable for optimizing sensor channel selection in wearable HMI systems.
- The proposed methodology enhances hand gesture recognition performance and efficiency.
- This work provides a foundation for developing more effective wearable HMI systems for prosthetic control and other applications.
