The possibilities of elimination of physiologic artifacts from the composition of electromyography (EMG) potentials, taking into consideration non-stationarity of signals, by means of their Wavelet-package decomposition are considered.
Among the methods of artifacts elimination, the Wavelet cleaning of signals, taking into consideration non-stationarity of electrophysiologic potentials, have a big preference consisted in the optimal time-frequency localization of researched signal.
Choosing the optimal quantity of decomposition levels is based on the properties of orthogonal Wavelet by means of that the composition of initial signal has been carried out.
The possibilities of application of processing and analysis method, using the Wavelet packages at the registration of stimulative EMG of the m.deltoideus muscle of patients with normal state of muscular system and muscular syndrome (carpal tunnel syndrome, cubital tunnel syndrome and demyelinating polyneuropathy) are considered.
The examples of the various levels of Wavelet decomposition, received from the electromyograph “VIASYS MEDELEC”, taking into consideration the peak amplitude of motor response, latent period, speed of signal performance and distance between electrodes, for the EMG signals of considered family of diseases are given.
As the mother Wavelet function, the Dobeshi functions (dB1), well localized at time and frequency are used.
In case of Wavelet-decomposition there is a possibility to receive one more numerical characteristic – value of entropy of coefficients in the nodes of package tree. In order to determine an optimal quantity of decomposition, the classic criteria, based on minimum of entropy are applied.
For the considered family of muscular diseases of researched muscle, selection of optimal quantity of decomposition levels of orthogonal Wavelet-transformation, taking into account the resolvability of researched signals and minimum of entropy of resolution levels is carried out. The difference between values of entropy of package Wavelet coefficients is even more significant than for the statistic characteristics, therefore it can have the diagnostic meaning as well.