Forearm High-Density Electromyography Data Visualization and Classification with Machine Learning for Hand Prosthesis...

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Machine Learning

Background:

• Myoelectric prostheses offer intuitive control for amputees but face low user acceptance due to complexity.
• Existing systems often lack ease-of-use and intuitive control, hindering widespread adoption.
• High-density surface electromyography (HD-EMG) presents a potential solution for more nuanced muscle signal interpretation.

Purpose of the Study:

• To develop a low-cost, easy-to-install wearable electrode system for myoelectric prostheses.
• To enhance the intuitiveness and ease-of-use of prosthetic control through advanced signal processing.
• To achieve high accuracy in classifying forearm muscle contractions for hand gesture recognition.

Main Methods:

• Fabrication of a flexible electrode array using standard printed circuit board manufacturing for cost-effectiveness.
• Utilizing high-density surface electromyography (HD-EMG) to capture detailed forearm muscle activity.
• Application of deep learning techniques for classifying muscle contraction patterns.
• Visualization of HD-EMG data using t-distributed Stochastic Neighbor Embedding (t-SNE).
• Validation of the system's performance on able-bodied subjects performing various hand gestures.

Main Results:

• A median recognition accuracy of 98.61% was achieved for an 8-gesture set using a majority vote on 5 successive inferences.
• For a 6-gesture set representing common prosthesis positions, the median accuracy reached 99.57% with the majority vote.
• The proposed flexible electrode array allows for quick design-to-production cycles and low-cost manufacturing.

Conclusions:

• The developed HD-EMG system with deep learning offers a highly accurate and intuitive control method for myoelectric hand prostheses.
• The low-cost, flexible electrode array design facilitates rapid development and adoption of advanced prosthetic technology.
• This approach addresses key limitations in current myoelectric prostheses, paving the way for improved user acceptance and functionality.