A database of high-density surface electromyogram signals comprising 65 isometric hand gestures

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering
• Neuroprosthetics

Background:

• Contemporary prosthetic hand control relies on electromyographic (EMG) signals from residual forearm muscles.
• Current myoelectric control offers limited functionality for dexterous prosthetic hands, despite mimicking natural muscle-joint relationships.

Purpose of the Study:

• To develop an annotated database of high-density surface EMG signals.
• To facilitate the design of robust and versatile EMG control interfaces for prosthetic hands.

Main Methods:

• Recorded 128-channel EMG signals from 20 able-bodied volunteers using dual electrode grids on the forearms.
• Participants performed 65 timed isometric hand gestures, with synchronous recording of EMG and hand joint forces.
• Assessed signal quality through frequency analysis, crosstalk evaluation, and skin-electrode contact detection.

Main Results:

• Generated a comprehensive, annotated database of high-density EMG signals.
• Quantitatively assessed signal integrity, including frequency content and channel crosstalk.
• Validated signal quality through objective metrics relevant to prosthetic control.

Conclusions:

• The annotated EMG database is a valuable resource for advancing prosthetic hand control research.
• Improved EMG signal acquisition and analysis are crucial for developing more intuitive and functional myoelectric prostheses.
• This work supports the development of next-generation prosthetic hand interfaces.