| Detection and classification of Coronary Artery Disease Diagnosis using AI. |
| Paper ID : 1052-ISCH |
| Authors |
|
mostafa abdelazem abdelmaksoud * faculty in science , Helwan university |
| Abstract |
| The diagnostic process for Coronary Artery Disease (CAD) often relies on complex, time-consuming analysis of clinical data and medical imagery. This study investigates the efficacy of artificial intelligence (AI) in automating the detection and classification of CAD. We propose a novel hybrid framework that synergizes explainable statistical models with high-performance deep learning. Firstly, feature engineering was performed on a dataset comprising Normal and diseased cases from 100 patients. These features were used to train and optimize multiple machine learning classifiers—such as Logistic Regression, XGBoost, and a Random Forest ensemble—to establish a robust statistical baseline for CAD prediction. Secondly, to harness the rich information within raw medical images [e.g., angiograms, echocardiograms], a deep convolutional neural network (CNN) was constructed. To address the challenge of limited annotated medical data, we employed transfer learning techniques. Models pre-trained on large-scale datasets (e.g., ImageNet), including VGG16 and EfficientNet, were fine-tuned on our specific medical imaging corpus, allowing for effective feature extraction and generalization. Our results indicate that while traditional ML models provided good interpretability and an accuracy of up to 85%, the deep transfer learning approach achieved state-of-the-art performance, attaining a classification accuracy of 92% in distinguishing between normal and abnormal cases. The fusion of these approaches offers a powerful tool for cardiologists, combining the transparency of statistical learning with the high predictive power of deep learning. This work underscores the transformative potential of AI in cardiovascular medicine and paves the way for its integration into clinical decision support systems. |
| Keywords |
| Coronary Artery Disease (The core condition) Artificial Intelligence (The overarching methodology) Deep Learning (A key technique you used) Transfer Learning (A specific, important technique you implemented) Machine Learning (The broader set of techniques) Medical Image Classification (The specific task) |
| Status: Abstract Accepted (Poster Presentation) |