Continuous Wavelet Transform for Muscle Activity Detection in Surface EMG Signals During Swallowing

Abstract

The surface electromyography (sEMG) has been used to characterize normal and abnormal behavior of the swallowing related muscles. One important activity in the analysis of the electromyographic recordings, is the detection of bursts, indicators of muscle activations but problematic in muscles with low signal-to-noise ratio (SNR). Most of methods for burst detection are based on amplitude measures which are signal-conditions dependent. We proposed a method to detect bursts based on the continuous wavelet transform and thresholding over the scalogram but not over amplitude. sEMG signals from 38 healthy subjects were recorded during swallowing tasks. We compared the proposed method to the visual method as a reference, and a previous method based on the Teager-Kaiser energy operator (TKEO). The proposed method avoids detection of false negatives better than TKEO, and it is suitable to apply in problems of burst detection in sEMG signals with low SNR.