Comparative analysis of deep learning algorithms for detection of coronary atherosclerosis

Coronary atherosclerosis is a chronic & ever-evolving condition that can present clinically as anything from symptoms to acute coronary syndrome, heart failure, or sudden cardiac death. Coronary atherosclerosis develops and progresses due to environmental or genetic causes. Unquestionably, CVD is one of the primary causes of mortality worldwide. One of the most common disorders associated with CVD is CAD. Atherosclerosis, which restricts blood flow to the heart muscle, is the major cause of CAD. The amount of gold for geometrical assessment, angiography, is now used to examine atherosclerosis, which is a difficult condition to diagnose. The diagnosis of the lesion or visual evaluation by the clinician are absolutely necessary for the angiography. In this study, DL potent feature extraction abilities provide it a significant advantage in the area of defect diagnostics. A Resnet50, Resnet101, and VGG19 were used. A ResNet-50 pre-trained on a dataset of coronary artery segments was used to substitute the backbone. Three classification systems' effectiveness in predicting heart disease is examined and contrasted with earlier research. Numerical outcomes demonstrate the increased 95.2 % accuracy of percentage of risk estimation of suggested approach Resnet50 as contrasted to other two approaches.