Developing a Deep-Learning-Based Coronary Artery Disease Detection Technique Using Computer Tomography Images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cardiology

Background:

• Coronary artery disease (CAD) is a leading global cause of mortality.
• Deep learning, particularly Convolutional Neural Networks (CNNs), shows promise for CAD detection from CT images.
• Existing CAD detection models often require substantial computational resources and large datasets.

Purpose of the Study:

• To develop an efficient CNN-based model for detecting coronary artery disease (CAD).
• To reduce the computational cost and data requirements of CAD detection models.
• To enhance the performance of CAD detection using image enhancement and optimized feature extraction.

Main Methods:

• An image enhancement technique was applied to improve CT image quality.
• You Only Look Once (YOLO) V7 was utilized for effective feature extraction.
• Aquila optimization was employed to fine-tune the hyperparameters of the UNet++ model for CAD prediction.

Main Results:

• The proposed model achieved high performance metrics, including accuracy (up to 99.5%), recall (up to 98.95%), and precision (up to 98.95%) on two datasets.
• The model demonstrated superior performance over recent techniques, with Area Under the ROC Curve (AUC) of 0.97 and 0.96.
• The optimized approach resulted in reduced computational costs and improved model stability, indicated by narrow confidence intervals.

Conclusions:

• The developed CNN-based model offers an efficient and accurate solution for detecting coronary artery disease (CAD).
• The proposed method effectively reduces computational demands, making it suitable for resource-limited settings.
• This AI-driven approach can significantly support clinicians in the early identification and management of CAD.