Pattern recognition of head movement based on mechanomyography and its application

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Signal Processing

Background:

• Head movements are crucial for human interaction and mobility.
• Developing non-invasive methods for controlling assistive devices is essential for individuals with disabilities.
• Mechanomyography (MMG) offers a potential avenue for capturing muscle activity related to head movements.

Purpose of the Study:

• To investigate the feasibility of recognizing distinct head movements using MMG signals.
• To develop a system for controlling a wheelchair model based on recognized head movements.
• To evaluate the accuracy and success rate of the developed pattern recognition and control system.

Main Methods:

• Collected four-channel MMG signals from neck muscles (SCM and SPL) during various head movements.
• Processed MMG signals through filtering, normalization, and segmentation.
• Extracted features using wavelet packet coefficients and bispectrum analysis.
• Reduced feature dimensions with Fisher linear discriminant analysis (FLDA).
• Classified head movements using a support vector machine (SVM) classifier.

Main Results:

• Achieved a high recognition rate of 95.92% for distinct head movements.
• Successfully demonstrated head movement control of a simulated car model, achieving an 85.74% success rate.
• Validated the effectiveness of MMG signal processing and machine learning for head movement pattern recognition.

Conclusions:

• MMG-based pattern recognition is a viable method for identifying head movements.
• This technology holds promise for developing intuitive and effective control systems for wheelchairs and other assistive devices.
• Further research can optimize the system for real-world clinical applications.