Related Experiment Video
Updated: Jun 10, 2025
06:57
Author Spotlight: Advancing Cardiovascular Imaging - Introducing the Spatially Weighted Calcium Score for Early Disease Detection
Published on: September 22, 2023
935
HAGMN-UQ: Hyper association graph matching network with uncertainty quantification for coronary artery semantic
Chen Zhao1, Michele Esposito2, Zhihui Xu3
1Department of Computer Science, Kennesaw State University, Marietta, GA, USA.
Medical Image Analysis
|October 16, 2024
Summary
We developed a novel deep learning model, HAGMN-UQ, for accurate semantic labeling of coronary arteries in angiograms. This method improves CAD diagnosis by efficiently identifying arterial branches with high precision.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiovascular Disease Research
Background:
- Coronary artery disease (CAD) is a major global health concern.
- Accurate identification of coronary artery branches from invasive coronary angiograms (ICAs) is crucial for diagnosing CAD and detecting stenosis.
- Deep learning models for semantic segmentation of coronary arteries face challenges in accuracy and computational complexity due to similar arterial morphologies.
Purpose of the Study:
- To propose an innovative deep learning approach for accurate coronary artery semantic labeling on ICAs.
- To address the challenges of accuracy and computational efficiency in coronary artery segmentation.
- To improve the diagnosis and detection of stenosis in CAD.
Main Methods:
- Developed a hyper association graph-matching neural network with uncertainty quantification (HAGMN-UQ).
- Utilized graph matching to map arterial branches and classify unlabeled segments based on labeled ones.
- Employed hypergraphs to model higher-order associations for improved robustness and accuracy.
- Integrated uncertainty quantification to reduce comparisons and accelerate inference speed.
Main Results:
- Achieved an accuracy of 0.9211 for coronary artery semantic labeling.
- Demonstrated fast inference speed, enabling efficient predictions.
- The HAGMN-UQ model effectively addresses the challenges of morphological similarity in coronary arteries.
Conclusions:
- The HAGMN-UQ model offers an effective and efficient solution for coronary artery semantic labeling.
- This approach has the potential to enhance real-time clinical decision-making for CAD diagnosis.
- The method provides a robust and accurate tool for analyzing invasive coronary angiograms.
