EMG-based Real Time Facial Gesture Recognition for Stress Monitoring.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Facial electromyogram (EMG) signals offer insights into neuromuscular activity.
• Real-time analysis of EMG signals is crucial for developing effective biofeedback systems.
• Existing systems often face challenges with power consumption and computational complexity.

Purpose of the Study:

• To develop a low-power EMG signal acquisition and real-time classification system for facial gestures.
• To enable accurate identification of jaw movements for various clinical applications.
• To reduce computational load for embedded implementations.

Main Methods:

• A custom-designed, low-power integrated circuit (IC) for EMG signal acquisition.
• Real-time feature extraction using discrete wavelet transforms (DWT).
• Classification of jaw movements using a support vector machine (SVM).

Main Results:

• The system achieved real-time classification of jaw clenching, opening, and resting states.
• Feature dimensionality reduction using DWT without compromising classification accuracy.
• Demonstrated suitability for biofeedback applications with low power consumption (3.8nW).

Conclusions:

• The developed EMG system provides real-time, accurate facial gesture classification with minimal power.
• The system is well-suited for continuous monitoring and biofeedback in stress, bruxism, and TMJ disorder management.
• The approach offers a viable solution for embedded biofeedback applications.