Prediction of Optimal Facial Electromyographic Sensor Configurations for Human-Machine Interface Control

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Human-Computer Interaction

Background:

• Surface electromyography (sEMG) is a key computer access method for individuals with motor impairments.
• Optimal sEMG sensor placement is complex and requires expert trial-and-error, especially for facial muscles.
• Simplifying sEMG sensor configuration is crucial for broader adoption in assistive technology.

Purpose of the Study:

• To reduce the complexity of sEMG sensor configuration for computer access.
• To predict human-machine interface (HMI) performance using quantitative signal features from a calibration task.
• To assess the impact of sensor configuration and order on HMI performance.

Main Methods:

• Developed a cursor control system using sEMG signals from facial muscles.
• Collected sEMG data across various sensor configurations during a calibration task.
• Extracted quantitative signal features (energy, complexity, inter-sensor activity).
• Utilized principal component factor analysis to predict HMI performance in a target selection task.

Main Results:

• Signal features related to EMG energy, complexity, and inter-sensor muscle activity significantly predicted HMI performance.
• The order of sensor configuration influenced performance more than the specific configurations used.
• Non-expert placement of sEMG sensors near usable muscle sites was feasible.

Conclusions:

• Quantitative sEMG signal features can predict HMI performance, simplifying sensor setup.
• Non-experts can effectively place sEMG sensors for computer access applications.
• Healthy individuals can learn to efficiently control sEMG-based HMI systems.