Electromyography interference pattern decomposition.

Abstract

Decomposition of an interference pattern enables examination of individual electromyography (EMG) motor units and their firing rates at more than minimal contraction forces. In this decomposition method, significant events with a constant occurrence (near motor unit action potentials) can be enhanced, and unwanted events (distant motor unit action potentials, artifacts) eliminated, by calculating the average accumulated change while sliding a fixed-width window along the digitized EMG interference pattern. Nonparametric statistical methods are then applied to these data to determine which information is significant at the .05 level. The exact duration of significant information is identified without the need for arbitrary thresholds and filters to eliminate unwanted information. Events are then classified into groups of similar events by comparing: (1) correlation coefficients, (2) point-to-point differences, (3) amplitudes, and (4) areas. The classification is further refined by the use of firing-rate information.