Main Article Content
Coronary inflammation plays a vital role in causing the myocardial infarction commonly known as heart attack. Therefore it is very difficult to predict and classify the coronary inflammation to prevent the heart attack as early as possible. Fat attenuation Index (FAI) is an imaging biomarker quantifies the inflamed coronary artery and is clinically obtained using non-invasive Coronary Computed Tomography Angiography (CCTA) test. In this work, a Deep Learning (DL) based network is being used to diagnose the prognostic value of FAI using anatomical information obtained from CCTA of the coronary arteries. The Recurrent Convolutional Neural Network (RCNN) is proposed for train the model on a CCTA categorical and anatomical image datasets of coronary artery to classify the cardiac mortality by diagnosing the inflamed coronary of the heart. The trained model uses Deeper Convolutional Neural Network’s (CNNs) Residual Network (ResNet) as image reconstruction method for denoising the cardiac artifacts. The most adverse cardiac event occurs in the proximal side of Right Coronary Artery. So, the network utilizes a Multi scale Coronary Response Dynamic Balloon Tracking Method (MSCAR-DBT) method for heart region enhancement and also to segment the RCA vessel from the arteries. Later, the features that are extracted are aggregated by Recurrent Convolutional Neural Network (RCNN) which performs single task classification. The system trained and tested using UCI repository datasets. To validate the networks performance Mann-Whitney U test is employed and the evaluated result is visualized in Receiver Operating Characteristics (ROC) curve. The experimental results demonstrate that automatic analysis of the coronary inflammation in a single task RCNN can produce better sensitivity, specificity and accuracy rate. This might helps the medical practitioner to diagnose the future heart attack and prevent patients from taking further non-invasive tests unnecessarily.