CAS-IQA: Teaching Vision-Language Models for Synthetic Angiography Quality Assessment

2025-05-23

AI Summary

Evaluating the quality of synthetically generated X-ray angiography images lacks reference-free, clinically fine-grained metrics tailored to vascular interventional procedures. Method: We propose the first explainable, task-driven quality assessment framework specifically designed for vascular intervention. It comprises: (1) CAS-3K, a large-scale, clinically realistic synthetic dataset containing 3,565 samples; (2) three clinically relevant, task-specific metrics—guidewire visibility, branch discernibility, and artifact interference; (3) MUST, a multi-path feature fusion and routing module that adaptively allocates visual tokens to respective metric branches; and (4) tight coupling with a vision-language model (VLM) to enable semantic alignment and interpretable scoring. Contribution/Results: Our framework significantly outperforms existing no-reference image quality assessment (IQA) methods on CAS-3K, establishing the first clinically ready, generalizable VLM-based solution for synthetic angiography evaluation.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Synthetic X-ray angiographies generated by modern generative models hold great potential to reduce the use of contrast agents in vascular interventional procedures. However, low-quality synthetic angiographies can significantly increase procedural risk, underscoring the need for reliable image quality assessment (IQA) methods. Existing IQA models, however, fail to leverage auxiliary images as references during evaluation and lack fine-grained, task-specific metrics necessary for clinical relevance. To address these limitations, this paper proposes CAS-IQA, a vision-language model (VLM)-based framework that predicts fine-grained quality scores by effectively incorporating auxiliary information from related images. In the absence of angiography datasets, CAS-3K is constructed, comprising 3,565 synthetic angiographies along with score annotations. To ensure clinically meaningful assessment, three task-specific evaluation metrics are defined. Furthermore, a Multi-path featUre fuSion and rouTing (MUST) module is designed to enhance image representations by adaptively fusing and routing visual tokens to metric-specific branches. Extensive experiments on the CAS-3K dataset demonstrate that CAS-IQA significantly outperforms state-of-the-art IQA methods by a considerable margin.

Problem

• Assessing quality of synthetic X-ray angiographies for clinical safety
• Leveraging auxiliary images for fine-grained quality score prediction
• Developing task-specific metrics for clinically relevant angiography evaluation

Innovation

• VLM-based framework for fine-grained quality scores
• Constructs CAS-3K dataset with score annotations
• MUST module enhances image representations adaptively