Unleashing Vision Foundation Models for Coronary Artery Segmentation: Parallel ViT-CNN Encoding and Variational Fusion

2025-07-17

AI Summary

Coronary artery segmentation is highly challenging due to fine vessel structures, complex morphologies, and low tissue contrast. To address these issues, we propose a parallel ViT-CNN encoder architecture with variational fusion. First, we design a cross-branch variational fusion module that models the latent distribution of features and dynamically learns modality-specific weights. Second, we introduce an evidence-based uncertainty refinement module to enhance robustness in ambiguous boundary segmentation. Third, we integrate attention-guided enhancement with multi-scale feature aggregation to achieve synergistic global-local representation learning. Evaluated on multiple public and private datasets, our method consistently outperforms state-of-the-art approaches, achieving average Dice score improvements of 2.3–4.1%. Moreover, it demonstrates superior cross-domain generalization capability.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

Accurate coronary artery segmentation is critical for computeraided diagnosis of coronary artery disease (CAD), yet it remains challenging due to the small size, complex morphology, and low contrast with surrounding tissues. To address these challenges, we propose a novel segmentation framework that leverages the power of vision foundation models (VFMs) through a parallel encoding architecture. Specifically, a vision transformer (ViT) encoder within the VFM captures global structural features, enhanced by the activation of the final two ViT blocks and the integration of an attention-guided enhancement (AGE) module, while a convolutional neural network (CNN) encoder extracts local details. These complementary features are adaptively fused using a cross-branch variational fusion (CVF) module, which models latent distributions and applies variational attention to assign modality-specific weights. Additionally, we introduce an evidential-learning uncertainty refinement (EUR) module, which quantifies uncertainty using evidence theory and refines uncertain regions by incorporating multi-scale feature aggregation and attention mechanisms, further enhancing segmentation accuracy. Extensive evaluations on one in-house and two public datasets demonstrate that the proposed framework significantly outperforms state-of-the-art methods, achieving superior performance in accurate coronary artery segmentation and showcasing strong generalization across multiple datasets. The code is available at https://github.com/d1c2x3/CAseg.

Problem

• Accurate coronary artery segmentation for CAD diagnosis
• Handling small size and low contrast in arteries
• Improving segmentation with global and local feature fusion

Innovation

• Parallel ViT-CNN encoding for global and local features
• Cross-branch variational fusion for adaptive feature integration
• Evidential-learning uncertainty refinement for accurate segmentation