Analyzing Muscle Synergies for Finger Movement Recognition using sEMG Signals

The study focuses on extracting muscle synergy from electromyographic (EMG) signals for recognition of finger movement. The process involves using a non-negative matrix factorization (NNMF) technique to decompose the EMG signal matrix into muscle synergies and their corresponding activation coefficients. The synergy obtained from this process reveals the underlying patterns in the EMG data that help in the classification of finger movements. Other steps include hyperparameter tuning using grid search, feature selection using sequential forward selection (SFS), and classification using support vector machine (SVM). The proposed method shows the effectiveness of NNMF in detecting muscle synergy for accurate classification of finger movements. A classification accuracy of 97.5±1.67% is obtained.