Classification of multi muscle activation patterns of osteoarthritis patients during level walking

Abstract 

The study compares the timing and frequency changes of surface EMGs recorded from osteoarthritis patients with previous traumatic ankle injury and normal subjects during level walking. EMG intensity (power) was obtained by a wavelet analysis. There were intensity values for each frequency characterized by the wavelets for every time point. The intensities were compounded into Multi Muscle Patterns (MMP) simultaneously showing the time and spectral aspects of the lower leg muscle activity.

The aim of the study was to test the hypothesis that the differences between the group of the MMPs from the affected leg (AFL) and the not affected leg (NAL) allow detecting whether a newly measured MMP results from an AFL or NAL. This hypothesis was tested by a spherical classification procedure yielding the correctly classified MMPs thus indicating the significance of the differences between the MMPs of the AFL and NAL. The hypothesis was supported (not falsified) by the results. Thus there were common features of muscle activity in the AFL of most osteoarthritis patients that allowed detecting whether the MMP of a new patient was of the kind seen in most other osteoarthritis patients. The spectral, timing and intensity factors in the MMP that allowed this classification were visualized in the mean MMPs of the patients and the control group. The comparison revealed where on average the relative timing and spectral differences of the muscle activation of osteoarthritis patients and control subjects occurred.

Keywords: Wavelet analysis, Pattern recognition, Muscle activity, Electromyography