Multichannel EHG Segmentation for automatically identifying contractions

摘要

In a recent past, several techniques have been developed to analyze the events contained in the electrohysterogram signals (EHG). But, the most of them focused on online methods. In this study, we use online methods like Fractal Dimension, Wavelet transform technique, Dynamic Cumulative Sum (DCS). Methods were applied on synthetic signals and real labeled EHG signals database acquiring using $4 x 4$ electrodes matrix. For this purpose, three parameters affecting the Fractal dimension method, the size of analyzing window, thresholding value and the window overlapping, were first tuned in order to identify the recommended values for ruptures detection of the EHG signals by comparing results to its label. According to the obtained results, these three methods seem to be encouraging methods that could be used for automatic ruptures segmentation.