Automatic validation of motor unit potential trains extracted by EMG signal decomposition

A novel method is proposed to assess the validity of motor unit potential trains (MUPTs) obtained during electromyographic (EMG) signal decomposition. The algorithm evaluates a MUPT in terms of motor unit potential (MUP) shape homogeneity, motor unit (MU) firing pattern consistency, and the estimated level of false classification errors in the MUPT. Performance metrics based on simulated EMG signals showed high accuracy for detecting invalid MUPT, similar to those expected to be created by a decomposition algorithm (>;93%), as well as valid MUPTs (98%). Compared to methods that evaluate only MUP shape consistency, the presented methods improved accuracy by 57.6% for detecting invalid MUPTs composed of MUPTs with similar-shaped MUP templates. For invalid sparse MUPTs, the accuracy of the presented method was 40% higher than that of the methods that use only MU firing pattern information.