Performance Evaluation of Empirical Mode Decomposition and Discrete Wavelet Transform for Computerized Hypoxia Detection and Prediction

This study proposes a new model relying on Empirical Mode Decomposition (EMD) and Discrete Wavelet Transform (DWT) in order to detect fetal hypoxia by using Cardiotocography (CTG) signals. We processed one well known open access intrapartum CTU-UHB database to find if our model could outperform the state-of-the art models. The model consists of three key stages: (1) Preprocessing, (2) Features extraction using EMD and DWT, (3) Classification with Support Vector Machine (SVM). Also, we present a comparative experimental study to measure the performance of SVM classifier depending on feature extraction methods. As a result, EMD and DWT have been found as useful methods for fetal hypoxia detection. Also, SVM classifier utilizing a combination of DWT and morphological features achieved the highest performance. Furthermore, DWT features produced more successful results than EMD features in terms of the classification success. Consequently, the proposed model ensured sensitivity of 57.42% and specificity of 70.11%.