The use of the wavelet transform in EMG M-wave pattern classification.

A system was previously designed to obtain estimates of the number of motor units (MUNE) in a superficial muscle and hence number of functioning motor neurons to that muscle. This method uses incremental stimulation of a motor nerve and subsequent recognition and classification of the elicited M-waves. In this earlier work we used the Fourier power coefficients as pattern classifiers. The presented work compares the Fourier transform classifier results with those obtained using a wavelet transform classifier. Data to test the two approaches were obtained from the thenar muscles of ten normal subjects. The results show that the wavelet transform is superior to the Fourier in classifying M-waves with significantly improved inter and intra-class variances.