Machine-Learning Based Detection of Coronary Artery Calcification Using Synthetic Chest X-Rays

Abstract

Coronary artery calcification (CAC) is a strong predictor of cardiovascular events, with computed tomography (CT)-based Agatston scoring widely regarded as the clinical gold standard. However, CT is costly and impractical for large-scale screening, while chest X-rays (CXRs) are inexpensive but lack reliable ground truth labels, constraining deep learning development. Digitally reconstructed radiographs (DRRs) offer a scalable alternative by projecting CT volumes into CXR-like images while inheriting precise labels. In this work, we provide the first systematic evaluation of DRRs as a surrogate training domain for CAC detection. Using 667 CT scans from the COCA dataset, we generate synthetic DRRs (posterior–anterior and lateral views per scan) and assess model capacity, super-resolution (SR) fidelity enhancement, preprocessing, and training strategies. Lightweight convolutional neural networks (CNNs) trained from scratch outperform large pretrained networks (DenseNet121, ResNet18); pairing super-resolution with contrast enhancement yields significant gains; and curriculum learning stabilises training under weak supervision. Our best configuration achieves a mean area under the receiver operating characteristic curve (AUC) of 0.754, comparable to or exceeding prior CXR-based studies. These results establish DRRs as a scalable, label-rich foundation for CAC detection, while laying the foundation for future transfer learning and domain adaptation to real CXRs.