Cross architectural fusion improves coronary artery stenosis detection
Cross-Architectural Feature Fusion for Coronary Artery Stenosis Detection in X-ray Angiography
Computer Vision and Pattern Recognition
Summary
Detecting narrowing in heart arteries from X-ray images is crucial but hard for doctors because the images can be unclear. The authors combine two advanced AI models to better find these blockages by mixing detailed vessel shapes with detection skills. Their approach works better on a test set and adapts to new data more reliably, helping with more accurate and consistent diagnoses.
What this means in practice
- •For medical imaging teams: Use the proposed fusion model to improve automated detection of blocked heart arteries in angiography images with better accuracy and reliability.
- •For clinical ai developers: Integrate test-time augmentation techniques to maintain detection performance when applying models to new clinical data with domain shifts.
Authors
Haobo Zhu, Mojtaba Lashgari, Jason Chai, Giovanni Luigi De Maria, Vicente Grau, Abhirup Banerjee
Abstract
Coronary artery disease remains a leading cause of mortality, with stenosis detection from invasive X-ray Coronary Angiography (XCA) critical for diagnosis. Manual interpretation is time-consuming, subjective, and variable, especially under low contrast and vessel overlap. Existing deep learning methods often rely solely on detection or segmentation, limiting performance.We propose a hybrid framework, Fusion-YOLOv11-L, combining a YOLOv11-L detector with a pretrained UNet++ via a channel attention gating module. This enables fusion of semantic and vessel-aware structural features, improving localisation of stenotic regions. The UNet++ is pretrained on ARCADE vessel segmentation dataset to provide anatomically informed representations. Experiments on the relabeled ARCADE (ARCADE*) dataset show improved performance over YOLOv11-L, achieving AP@50 of 0.585 and F1-score of 0.596. Cross-domain evaluation on a dataset from the Oxford Acute Myocardial Infarction (OxAMI) Study reveals performance degradation due to domain shift, mitigated using test-time adaptation and augmentation. Test-time augmentation improves AP@50 from 0.434 to 0.480, indicating better generalisability. These results show that cross-architecture feature fusion and lightweight adaptation improve robustness and generalisability of automated stenosis detection in XCA.