Localisation of coronary artery blockages using ECG signal analysis

Abstract

Coronary artery disease (CAD) is a type of cardiovascular disease that is one of the top three leading causes of death globally. Caused by the buildup of plaque in arteries that supply blood to the heart, CAD worsens gradually. Thus, early diagnosis and detection are crucial in reducing fatality and mortality. A method to detect CAD or occluded coronary arteries is using electrocardiograms (ECG), which is a non-invasive and non-radioactive procedure to get heartbeats output as a wavelength signal. Since the development and improvement in technology and the field of machine learning, deep learning techniques have been implemented in the medical field. Some uses include the detection and classification of abnormalities such as CAD in ECG signals. A commonly used method is convolutional neural network (CNN), which is well suited for processing data in a grid-like format such as ECG. This project aims to develop a hybrid model involving Discrete Wavelet Transform (DWT), CNN and Support Vector Machine (SVM) for feature extraction and classification of ECG signals to localise coronary artery blockages by implementing the approach of classifying different MI localisation classes associated with different occluded coronary arteries. This project uses PhysioNet’s PTB Diagnostic ECG Dataset to train and test the proposed hybrid model. The proposed hybrid model from this project yielded promising results of an average model accuracy of 0.877, specificity of 0.808, sensitivity of 0.832 and an F1 score of 0.880.