Inter-scale local maxima-a new technique for wavelet analysis of EMG signals

A new method, called inter-scale local maxima (ISLM) has been developed for estimation of the local frequency content of a signal and applied to the analysis of EMG interference pattern signals. The algorithm, based on the propagation of local maxima across scales, is used to classify signal singularities in a specific frequency band. EMG signals were decomposed at several scales with the redundant dyadic wavelet transform and ISLM values derived. The ISLM represents the most prominent local maximum, across the scales, associated with a singularity. A statistical analysis of the ISLM histograms at each scale showed that subjects from normal, neuropathic and myopathic groups could be classified by differences in standard deviation, kurtosis and standard error of the mean. Although the sample size was small and further evaluation is required, ISLM represents a potentially useful technique in the diagnosis of patients with neuromuscular disorders