Abstract

Globally, coronary atherosclerosis in the context of Chronic Heart Failure (CHF) represents a significant cause of mortality. It is therefore imperative that timely intervention is undertaken in order to reduce the mortality rate associated with coronary atherosclerosis. Coronary atherosclerosis is commonly assessed through echocardiography (ECHO), although the utilisation of this technique remains limited in some healthcare facilities. Accordingly, this study proposes an automated technique to enhance the accuracy of ECG interpretation through the advancement of Electrocardiography (ECG) images using a wave translation algorithm with a Continuous Wavelet Transform (CWT). In this study, the CWT algoritm is employed for the characterisation of the dynamics of Electrocardiogram (ECG) signals. The CWT algoritm is initially derived through partial derivatives, and subsequently, a new formula, Continuous Wavelet Multiscale (CWMs), is introduced. This new formula is in accordance with the standards set forth by the European Society of Cardiology (ESC) and incorporates expert-provided information for enhanced diagnostic precision. Moreover, a Convolutional Neural Network (CNN) was subsequently employed for the classification of ECG data from patients into three categories: atherosclerosis, normal, and anomaly. The CWMs technique thus demonstrates a significant improvement in ECG interpretation, with an accuracy rate of 99.6% for the detection of coronary atherosclerosis. These findings highlight the potential of this automated technique to serve as a reliable diagnostic tool, particularly in healthcare settings where Echocardiography (ECHO) equipment is not readily accessible. This advancement could facilitate more widespread and accurate early detection of coronary atherosclerosis, ultimately contributing to improved patient outcomes and reduced mortality rates associated with congestive heart failure.