DT-CWT Based Preterm Birth Detection with Electro Hysterogram (EHG) Signals

Abstract

The current study's goal is to analyze uterine electrohysterogram (EHG) signals in order to accurately diagnose preterm birth, or birth before a full 37 weeks of gestation, with the hope of preventing the preterm birth of 15 million babies annually and lowering the preterm infant mortality rate. These signals give doctors vital information about uterine contractions. The algorithm used for the current probe was initially created by applying dual-tree complex wavelet transform (DT-CWT) to the uterine EMG signals, which results in multiple subbands. Next, using cumulants of the DT-CWT coefficients the features are calculated. The features are then classified using a Hybrid RBF network and it is noteworthy that the accuracy rate becomes 98.9% when the top 10 features were used.