Algorithm for the detection of muscle activation in surface electromyograms during periodic activity

A simple surface electromyography (EMG) activation detection algorithm was developed for improved numerical definition of initiation and deactivation of muscle activity during periodic motion when maximum voluntary contractions are impractical to obtain. For the encapsulation of activation/deactivation periods of a signal as percentages of normal cycle parameters, two interrelated and variable thresholds of percent amplitude and duration of a normalized cycle were the analyzed inputs into an algorithm. Outputs for statistical analysis were total percent activation per cycle, standard deviation of activity per cycle, and temporal indices of where the signal turned on and off. Percent activity per cycle had a coefficient of variance of 0.24 (0.11). After the user chose whether to consider the signal for either encompassing all non-base-line activity or peak activity only, resulting coefficients of variation for percent activity were reduced to 0.16 (0.08). The results indicated the feasibility of a mathematically simple algorithm for repeatable decomposition of EMG activity. The need for a modifiable threshold parameter to incorporate varying needs of salient activity levels was also substantiated. (C) 2002 Biomedical Engineering Society.