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

2025-11-14

AI Summary

Coronary artery calcification (CAC) screening via CT is costly, while real chest X-rays lack reliable annotations for large-scale deployment. Method: This work proposes a novel synthetic X-ray detection paradigm based on digitally reconstructed radiographs (DRRs), systematically validating DRRs as an effective surrogate training domain for CAC detection. The approach integrates super-resolution reconstruction, contrast enhancement, a lightweight CNN, and curriculum learning to enable stable weakly supervised training. Results: Evaluated on the COCA dataset, the method achieves a mean AUC of 0.754—matching or surpassing state-of-the-art methods using real chest X-rays. Contribution: This study establishes the feasibility of DRR-based synthetic data for CAC screening, introduces the first DRR generation–modeling co-design framework tailored for CAC detection, and provides a scalable, low-cost pathway for early cardiovascular risk assessment.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Calibration & Uncertainty Quantification
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Coronary artery calcification (CAC) is a strong predictor of cardiovascular events, with 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 and assess model capacity, super-resolution fidelity enhancement, preprocessing, and training strategies. Lightweight CNNs trained from scratch outperform large pretrained networks; pairing super-resolution with contrast enhancement yields significant gains; and curriculum learning stabilises training under weak supervision. Our best configuration achieves a mean 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.

Problem

• Detecting coronary artery calcification using synthetic chest X-rays
• Overcoming lack of reliable labels in chest X-rays for deep learning
• Evaluating digitally reconstructed radiographs as scalable training alternative

Innovation

• Using synthetic DRRs from CT scans for training
• Applying super-resolution and contrast enhancement techniques
• Employing curriculum learning under weak supervision