Researchers have developed a novel deep learning framework called Fusion-YOLOv11-L for detecting coronary artery stenosis in X-ray angiography. This hybrid model integrates a YOLOv11-L detector with a UNet++ segmentation network, leveraging channel attention gating to fuse semantic and structural features. The UNet++ component is pre-trained on the ARCADE vessel segmentation dataset to enhance anatomical understanding. Experiments on the ARCADE* dataset demonstrated improved performance, achieving an AP@50 of 0.585 and an F1-score of 0.596, outperforming the standalone YOLOv11-L. The framework also showed improved generalizability on the OxAMI Study dataset through test-time adaptation and augmentation techniques. AI
IMPACT This model could improve the accuracy and efficiency of diagnosing coronary artery disease, potentially leading to earlier interventions.
RANK_REASON Academic paper detailing a new model and its performance on specific datasets. [lever_c_demoted from research: ic=1 ai=1.0]
- ARCADE*
- arXiv
- Fusion-YOLOv11-L
- Hugging Face
- OxAMI Study
- UNet++: A Nested U-Net Architecture for Medical Image Segmentation
- YOLOv11-L
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
