Nonnegative matrix factorization for the identification of EMG finger movements: evaluation using matrix analysis

Area of Science:

• Biomedical Engineering
• Signal Processing
• Rehabilitation Technology

Background:

• Surface electromyography (sEMG) is crucial for assessing hand function in gesture recognition, prosthetics, and rehabilitation.
• The nonstationary nature of sEMG signals presents challenges for accurate feature selection.

Purpose of the Study:

• To develop and validate a nonnegative matrix factorization (NMF)-based feature selection method for hand gesture recognition using sEMG signals.
• To enhance the accuracy and reliability of sEMG feature extraction with minimal sensor usage.

Main Methods:

• Utilized nonnegative matrix factorization (NMF) for feature selection, exploiting signal additivity and sparsity.
• Employed artificial neural networks for classifying ten finger flexions (five simple, five complex).
• Conducted experiments using two sEMG sensors and validated results with NMF permutation matrix analysis.

Main Results:

• Achieved up to 92% accuracy in classifying ten finger flexions.
• Demonstrated high accuracy for simple (95%) and complex (87%) finger flexions.
• Confirmed the effectiveness of NMF in extracting reliable sEMG features.

Conclusions:

• The proposed NMF-based feature selection method significantly improves hand gesture recognition accuracy from sEMG signals.
• This approach offers a robust and efficient solution for applications in prosthetics and rehabilitation.
• The method's ability to use fewer sensors while maintaining high accuracy is a key advantage.