Rethinking Intracranial Aneurysm Vessel Segmentation: A Perspective from Computational Fluid Dynamics Applications

ICLR 2026 Conference Submission17205 Authors

Keywords: Intracranial Aneurysm Segmentation, Vessel Segmentation, Magnetic Resonance Angiography, Computational Fluid Dynamics

Abstract:

The precise segmentation of intracranial aneurysms and their parent vessels (IA-Vessel) is a critical step for hemodynamic analyses, which mainly depends on computational fluid dynamics (CFD). However, current segmentation methods predominantly focus on image-based evaluation metrics, often neglecting their practical effectiveness in subsequent CFD applications. To address this deficiency, we collect and annotate the **I**ntracranial **A**neurysm **V**essel **S**egmentation (IAVS) dataset, a comprehensive, multi-center collection comprising 641 3D MRA images with 587 annotations of aneurysms and IA-Vessels.
In addition to image-mask pairs, the dataset includes detailed hemodynamic analysis outcomes, addressing the limitations of existing datasets that neglect topological integrity and CFD applicability.
To facilitate the development and evaluation of clinically relevant techniques, we construct two evaluation benchmarks including global localization of aneurysms (Stage I) and fine-grained segmentation of IA-Vessel (Stage II) and develop a simple and effective two-stage framework, which can be used as a out-of-the-box method and strong baseline.
The first stage utilizes a detection network with dynamic queries to globally locate aneurysms. The second stage implements a topology-aware segmentation network for localized IA-Vessel delineation, designed to minimize geometric inaccuracies.
For comprehensive evaluation, we establish a standardized CFD applicability evaluation system that enables the automated and consistent conversion of segmentation masks into CFD models, offering an applicability-focused assessment of segmentation outcomes.
The data, code, and model will be made publicly available upon acceptance.ns of aneurysms and IA-Vessels. This dataset incorporates detailed hemodynamic analysis outcomes, overcoming the shortcomings of prior datasets that overlook topological integrity and CFD applicability. Utilizing this dataset, we developed a two-stage framework for the segmentation task. The first stage utilizes a detection network with dynamic queries to globally locate aneurysms. The second stage implements a topology-aware segmentation network for localized IA-Vessel delineation, designed to minimize geometric inaccuracies. Moreover, we established a standardized CFD applicability evaluation system. This system facilitates the automated, uniform transformation of segmentation masks into CFD models and provides an applicability-focused assessment of segmentation outcomes. Extensive experiments demonstrate that our proposed framework achieves state-of-the-art segmentation performance and a superior CFD applicability score compared to existing segmentation approaches. This study offers a reproducible benchmark for the development and evaluation of clinically relevant segmentation techniques. The data, code, and model will be made publicly available upon acceptance.

Primary Area: datasets and benchmarks

Submission Number: 17205