• Kersin Witte
  • Peter Blaser
Keywords: surface electromyography, non-stationary methods


Commercially-used stationary analysis techniques for quantification of surface EMG signals yield in general only small amounts of information about motor unit recruitments and their frequency behavior. Therefore, new methods for time-dependent analysis of EMG signals are of great interest. In this study two analysis techniques developed for non-stationary biological signals are verified. First, adaptive power estimation proved that the time characteristic of muscle activity is very variable under constant external conditions. Secondly, a bivariate ARMA model was used for a time-dependent frequency analysis of EMG signals showing specific similarities in the curves as well as the high variability of the time courses of median frequency. The application of the methods presented here to EMG signals gives a new understanding of the inner dynamics of the muscle functioning.