Adaptive Electrode Calibration Method Based on Muscle Core Activation Regions and Its Application in Myoelectric...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Electrode shifts significantly degrade the accuracy of myoelectric pattern recognition.
• Existing calibration methods are often sensitive to electrode displacement, limiting real-world applications.

Purpose of the Study:

• To develop an adaptive electrode calibration method for high-density surface electromyography (HD-sEMG) to mitigate the impact of electrode shifts.
• To improve the robustness and accuracy of hand gesture recognition in myoelectric control systems.

Main Methods:

• Decomposition of HD-sEMG data using fast independent component analysis (ICA) to identify core muscle activation regions.
• Unsupervised alignment of extracted core activation regions for adaptive electrode calibration.
• Utilizing a CNN+LSTM network with specific training strategies for gesture recognition.

Main Results:

• The proposed method effectively reduces the negative effects of electrode shifts on gesture recognition accuracy, increasing overall accuracy by 5.72-7.69%.
• Significant accuracy improvements were observed, particularly when using limited data (13.32-17.30% increase) or fewer repetitions (12.01-13.75% increase).

Conclusions:

• The adaptive electrode calibration method enhances the accuracy and robustness of myoelectric gesture recognition systems.
• This approach shows potential for reducing user training time and improving the usability of myoelectric control systems.