Coronary Artery Disease Detection Using Pre-Trained ResNet and VGG-16

Coronary Artery Disease (CAD) affects coronary arteries, responsible for supplying oxygen and vital nutrients to the heart muscle. This is because of the development of atherosclerotic plaques on the walls of coronary arteries. The early detection of coronary atherosclerosis is a challenging task and can increase the financial burden on patients. In this research, an ensemble neural network is developed for the detection and classification of coronary artery disease. The ensemble neural network has integrated two models, VGG-16 and ResNet 50. The datasets used in the research are the coronary artery disease dataset, which has images and the MIT-BIH Arrhythmia dataset. The min-max normalization approach is employed in the pre-processing stage and an ensemble neural network is utilized for both the detection and classification of coronary artery diseases. The proposed ensemble neural network attained an accuracy of 99.45%, precision of 98.82%, recall of 98.85%, and f1-score of 98.80%, comparatively higher than conventional methods like Convolutional Neural Network (CNN), ResNet and VGG-16.