A Novel Sliding Mode Differentiator-Based Feature for EMG-Based Hand Gesture Characterization.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Motor intent (MI)-based muscle computer interfaces (MCIs) are crucial for prosthetic control.
• Current MCIs face challenges with feature robustness, noise, and low electromyography (EMG) spatial resolution, limiting movement characterization.
• These limitations decrease the performance of prosthetic control systems.

Purpose of the Study:

• To develop a novel feature extraction technique for improved EMG signal analysis.
• To enhance the accuracy and robustness of hand gesture characterization for MCIs.
• To address the limitations of existing feature extraction methods in EMG-based control.

Main Methods:

• Introduced a novel feature extraction technique combining Sliding Mode Differentiator (SMD) and symmetric positive definite (SPD) matrices.
• Utilized SMD to extract unique patterns from EMG signals.
• Employed SPD matrices to leverage spatial-temporal properties of EMG data.

Main Results:

• Achieved high average classification accuracy: 98.7±3.0% for able-bodied subjects and 97.9±5.2% for amputees across 13 hand gestures.
• Demonstrated that channel reduction by 75% (24 to 6 channels) did not compromise performance.
• Showcased superior performance compared to state-of-the-art methods, especially in noisy conditions.

Conclusions:

• The proposed SMD-SPD technique significantly improves hand gesture characterization accuracy and robustness for MCIs.
• The method is effective in both high-density and sparse-density EMG electrode configurations.
• This advancement holds potential for enhancing prosthetic control, assistive robots, and gesture-based applications.