Evaluating a motor unit potential train using cluster validation methods
Authors: Hossein Parsaei and Daniel Stashuk.
Abstract:
Assessing the validity of motor unit potential trains (MUPTs) obtained by decomposing a needle-detected electromyographic (EMG) signal is a crucial step in using these trains for quantitative EMG analysis. In general, for MUPT validation a train is assessed using the shapes of its motor unit potentials (MUPs) and the motor unit firing pattern it represents. Here, two methods to assess the validity of a given MUPT using its MUP shape information are presented. These methods are based on the gap statistic and jump algorithms presented for estimating the number of clusters in a dataset. They evaluate the shapes of the MUPs of a MUPT to determine whether it represents the activity of a single MU (i.e. it is a valid MUPT) or not. Evaluation results using both simulated and real data show the gap statistic method is more accurate than the jump method in correctly categorizing a train. The accuracy of the gap static method was 92.3% for simulated data and 93.8% for real data while accuracy for the jump method was 88.3% and 91.0%, respectively. The results are encouraging and suggest that using these methods can improve EMG signal decomposition results, and can facilitate automatic validation of a MUPT.