Semantically Conditioned Diffusion Models for Cerebral DSA Synthesis

TL;DR

This paper introduces a semantically conditioned latent diffusion model that synthesizes realistic cerebral DSA images with explicit control over anatomy and acquisition parameters, aiding research and training.

Contribution

We developed a novel conditional latent diffusion model trained on a large DSA dataset, enabling controlled synthesis of cerebral angiography images with high clinical realism.

Findings

Generated images received high Likert scores (3.1-3.3) from experts.

High inter-rater reliability (ICC 0.80-0.87) in image quality assessment.

Low FID score (15.27) indicating distributional similarity to real data.

Abstract

Digital subtraction angiography (DSA) plays a central role in the diagnosis and treatment of cerebrovascular disease, yet its invasive nature and high acquisition cost severely limit large-scale data collection and public data sharing. Therefore, we developed a semantically conditioned latent diffusion model (LDM) that synthesizes arterial-phase cerebral DSA frames under explicit control of anatomical circulation (anterior vs.\ posterior) and canonical C-arm positions. We curated a large single-centre DSA dataset of 99,349 frames and trained a conditional LDM using text embeddings that encoded anatomy and acquisition geometry. To assess clinical realism, four medical experts, including two neuroradiologists, one neurosurgeon, and one internal medicine expert, systematically rated 400 synthetic DSA images using a 5-grade Likert scale for evaluating proximal large, medium, and small peripheral vessels. The generated images achieved image-wise overall Likert scores ranging from 3.1 to 3.3, with high inter-rater reliability (ICC(2,k) = 0.80--0.87). Distributional similarity to real DSA frames was supported by a low median Fr\'echet inception distance (FID) of 15.27. Our results indicate that semantically controlled LDMs can produce realistic synthetic DSAs suitable for downstream algorithm development, research, and training.