Coronary Artery Segmentation and Vessel-Type Classification in X-Ray Angiography

TL;DR

This work tackles robust segmentation of coronary arteries in X-ray angiography and vessel-type labeling (LAD/LCX/RCA) by combining a curated expert-annotated dataset with a dual-path analysis: learning-to-tune classical vesselness filters and evaluating high-resolution CNN/Transformer segmentation. It shows that per-image hyperparameter prediction improves classical methods, while high-resolution FPN-based models with merged coronary+catheter supervision achieve state-of-the-art internal Dice and reasonable external transfer with limited adaptation. A second stage assigns vessel identity within the segmented tree, enabling vessel-specific analytics. External validation on DCA1 reveals domain shift but demonstrates substantial gains with modest fine-tuning, underscoring the approach’s clinical potential for robust, vessel-localized coronary analytics in routine practice.

Abstract

X-ray coronary angiography (XCA) is the clinical reference standard for assessing coronary artery disease, yet quantitative analysis is limited by the difficulty of robust vessel segmentation in routine data. Low contrast, motion, foreshortening, overlap, and catheter confounding degrade segmentation and contribute to domain shift across centers. Reliable segmentation, together with vessel-type labeling, enables vessel-specific coronary analytics and downstream measurements that depend on anatomical localization. From 670 cine sequences (407 subjects), we select a best frame near peak opacification using a low-intensity histogram criterion and apply joint super-resolution and enhancement. We benchmark classical Meijering, Frangi, and Sato vesselness filters under per-image oracle tuning, a single global mean setting, and per-image parameter prediction via Support Vector Regression (SVR). Neural baselines include U-Net, FPN, and a Swin Transformer, trained with coronary-only and merged coronary+catheter supervision. A second stage assigns vessel identity (LAD, LCX, RCA). External evaluation uses the public DCA1 cohort. SVR per-image tuning improves Dice over global means for all classical filters (e.g., Frangi: 0.759 vs. 0.741). Among deep models, FPN attains 0.914+/-0.007 Dice (coronary-only), and merged coronary+catheter labels further improve to 0.931+/-0.006. On DCA1 as a strict external test, Dice drops to 0.798 (coronary-only) and 0.814 (merged), while light in-domain fine-tuning recovers to 0.881+/-0.014 and 0.882+/-0.015. Vessel-type labeling achieves 98.5% accuracy (Dice 0.844) for RCA, 95.4% (0.786) for LAD, and 96.2% (0.794) for LCX. Learned per-image tuning strengthens classical pipelines, while high-resolution FPN models and merged-label supervision improve stability and external transfer with modest adaptation.

Figures (6)

• Figure 1: Overview of the proposed XCA analysis workflow. From each cine, a best frame is selected near peak opacification and enhanced, followed by first-stage segmentation to obtain a global foreground mask. A second stage assigns vessel identity within the coronary tree, and outputs are reviewed by an interventional cardiologist for clinical consistency.
• Figure 2: Dataset and labeling summary. Best frames are extracted from cine angiograms and annotated to produce coronary-artery and catheter masks, with additional vessel-level labels for the major coronaries when visible. All masks were reviewed and corrected as needed by a senior interventional cardiologist prior to final approval.
• Figure 3: Representative qualitative outputs of classical vesselness-based pipelines on XCA frames. The examples illustrate differences in sensitivity to small branches and background artifacts, and the role of thresholding and morphological postprocessing in producing the final binary masks.
• Figure 4: Example prediction on the external DCA1 cohort compared with the provided ground-truth vessel mask. This figure highlights typical cross-dataset differences in appearance and labeling conventions that contribute to performance drops under strict external testing.
• Figure 5: Qualitative vessel-type labeling within the segmented coronary tree. Pixels in the first-stage foreground are assigned to one of the main coronary classes by the second-stage model; colors indicate the predicted vessel identity.