Detection of uterine MMG contractions using a multiple change-point estimator and K-means cluster algorithm

We propose a single-channel two-stage detector of uterine magnetomyogram (MMG) contractions during pregnancy. In the first stage, we assume that the measurements are modeled by a zero-mean Gaussian random variable with time-varying piecewise constant variance. Therefore, we apply a model-based segmentation procedure which detects multiple change points in the variance values using the Schwarz information criterion (SIC) and a binary search approach. Then, in the second stage, we apply the K-means cluster algorithm to classify each time segment using the root-mean square (RMS) as a feature. We apply our algorithm to real MMG records obtained from five patients having contractions with gestational ages between 31 and 40 weeks.