Predicting coronary artery disease using learning rate based snow leopard optimization with soft swish scaling based bidirectional long short‐term memory

Abstract

AbstractThe most common kind of heart disease is coronary artery disease (CAD), which impacts millions globally. For CAD detection, the computed tomography (CT) image is very helpful. CT aids in the quick visualization of the heart and coronary arteries with a higher spatial resolution. So, this research methodology uses the CT image for CAD detection with the risk assessment by the proposed SS‐B‐LSTM (Soft Swish Scaling based Bidirectional Long Short‐Term Memory) algorithm. Moreover, the proposed methodology considers the vital features from different regions of the heart encompassing calcium tissue, leaflet tissue, and blood pool. Thus, CAD can be detected efficiently, and the risk assessment is done precisely. The proposed research mainly consists of seven steps: preprocessing, heart segmentation, clustering, feature extraction, feature selection, disease detection, and risk assessment. The proposed technique detects CAD with an accuracy of 96.66%. Furthermore, the computational time of the proposed framework is 0.3948 s. After experimental evaluation, the proposed technique is found to be more efficient in detecting and classifying CAD. Moreover, the complexity is low compared to the existing works.