Muscle activity mapping by single peak localization from HDsEMG.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Electromyography (EMG) is crucial for human-machine interfaces (HMIs) in prosthetics and rehabilitation.
• Decoding muscle activation patterns is key for advanced HMI control.
• Current methods face challenges in precisely localizing muscle activity in compact anatomies.

Purpose of the Study:

• To develop and validate a new approach for assessing and decoding muscle activity using high-density surface EMG.
• To localize the origin of individual temporal peaks in EMG signals from forearm muscles during finger extensions.
• To estimate distinct muscle volumes for each finger and classify EMG peaks based on these volumes.

Main Methods:

• Utilized high-density surface EMG recordings from the dorsal forearm during low-force finger extensions.
• Applied a spatial domain surface Gaussian fit to localize EMG peak origins.
• Estimated muscle volumes for individual finger extensions across 10 subjects.
• Classified individual EMG peaks into corresponding finger actions based on estimated volumes.

Main Results:

• Demonstrated high consistency in estimated muscle volumes across subjects, suggesting distinct muscle regions for each finger action.
• Achieved high classification accuracy for EMG peaks: 79% (index), 84% (middle), 76% (ring), and 79% (little finger).
• Indicated potential structural differences in muscle fibers between digits.

Conclusions:

• The volume analysis method effectively assesses spatial activation patterns in compact muscle groups.
• The single peak classification approach enables near-instantaneous identification of muscle activations.
• This technique offers a promising advancement for EMG-based HMIs, improving prosthetic control and assistive technologies.