Statistics of inter-spike intervals as a routine measure of accuracy in automatic decomposition of surface electromyogram

Automated motor unit (MU) decomposition algorithms of surface electromyogram (EMG) have been developed recently. However, a routine estimate of the decomposition accuracy is still lacking. The objective of this preliminary study was to examine the statistics of the inter-spike intervals (ISIs) of the identified MUs as a measure of the decomposition accuracy, such that the ISI analysis can be used as a routine procedure to assess the accuracy of the surface identified MU spike timings. A surface EMG recording and decomposition system was used to record EMG signals and extract single MU activities from the first dorsal interosseous muscle of three healthy individuals. The estimated ISI statistics were cross-validated with decomposed MUs from simultaneous intramuscular EMG recordings. Our preliminary results reveal that the distribution of the ISIs, specifically the deviation from the Gaussian distribution as represented by secondary peaks at the short or long ISIs, can provide information regarding the spurious errors and missed firing errors in the decomposition. In addition, the variability (coefficient of variation) of the ISIs also correlated inversely with the decomposition accuracy. These findings show that the ISI statistics can be used to assess the spike timing accuracy of the identified MUs from surface EMG decomposition algorithms.