Classification of uterine EMG signals by using Normalized Wavelet Packet Energy

The aim of this work is to detect labor by classifying uterine electromyogram (EMG) signals into 2 classes: pregnancy and labor. Based on the fact that the energy distribution of the uterine EMG signals varies throughout pregnancy, the recorded signals are first decomposed into a 3-level wavelet packet tree then the Normalized Wavelet Packets Energies (NWPEs) are calculated and used to classify the signals. The high percentage of correctly classified events indicates that the use of NWPE can be a suitable choice for solving classification problems of electrophysiological signals and proves the efficiency of this approach in detecting labor.