Trust the Prior (Or Not): Uncertainty-Aware Abdominal Aortic Aneurysm Segmentation

Jul 16, 2026

Upload by Logic

Robust segmentation of intraluminal thrombus is critical for abdominal aortic aneurysm risk assessment, yet it is hindered by heterogeneous thrombus morphology and weak contrast between thrombus and non-enhanced tissues. Domain shifts from differing CTA acquisition and reconstruction protocols further limit multi-center generalization. A patient-specific framework is introduced that combines discriminative learning with anatomically informed priors via (1) GMM-based intensity normalization and (2) uncertainty-gated anatomical attention. The method reaches state-of-the-art in-distribution results and improves out-of-distribution multi-center performance while keeping interpretability through explicit separation of visual and anatomical evidence.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages12

Reading Time30 min

Abstract

Introduction

Motivation and Challenges

Proposed Patient-Specific Framework

Validation and Contributions

TRUST THE PRIOR ( OR NOT): UNCERTAINTY-AWARE ABDOMINAL AORTIC ANEURYSM SEGMENTATION A PREPRINT arXiv :2607 .00201v1 [ cs .CV] 30 Jun 2026 Erich Robbi Department of Computer Science University of Trento Trento, Italy erich .robbi@unitn .it Daniele Ravanelli Medical Physics Unit Santa Chiara Hospital Trento, Italy [daniele.ravanelli@asuit.tn.it](daniele.ravanelli@asuit.tn.it) Andrea Passerini Department of Computer Science University of Trento Trento, Italy andrea.passerini@unitn .it July 2, 2026 ABSTRACT Robust segmentation of intraluminal thrombus is critical for risk assessment in Abdominal Aortic Aneurysm, yet it remains challenging due to heterogeneous thrombus features and low contrast with surrounding non-enhanced tissues. Domain shifts induced by different Computed Tomography Angiography (CTA) protocols further inhibit multi-center generalization of deep learning models. To address these challenges, we propose a patient-specific framework that integrates discriminative learning with anatomically informed priors. Our approach introduces two key components: (1) a patient-specific intensity normalization based on a Gaussian Mixture Model of local anatomy, and (2) an Uncertainty-Gated Anatomical Attention module that incorporates spatial priors while adaptively modulating their influence according to voxel-wise confidence. This design allows for anatomical guidance in ambiguous regions while suppressing unreliable priors. The proposed method achieves state-of-the-art performance on in-distribution test data and substantially outperforms existing alternatives in generalization to external multi-center CTA data, while remaining interpretable through an explicit separation of visual and anatomical evidence. Keywords Abdominal Aortic Aneurysm · Thrombus Segmentation · Uncertainty Quantification · OOD Generalisation 1 Introduction Accurate segmentation of intraluminal thrombus (ILT) in patients with abdominal aortic aneurysm (AAA) is crucial for risk assessment Wanhainen et al. [2024] . However, obtaining reliable ILT segmentation in pre-operative computed tomography angiography scans (CTAs) poses significant challenges Wang et al. [2022], due to insufficient contrast in Hounsfield unit (HU) values between the thrombus and adjacent abdominal structures (Figure 1), leading to indistinct boundaries and a high rate of misclassification Hwang et al. [2022] . These challenges are further aggravated by the substantial inter-and intra-patient variability Abdolmanafi et al. [2023] . Deep learning has improved automated ILT segmentation Guo et al. [2025], Xu et al. [2025] . Early methods primarily relied on computationally efficient 2D architectures, including FCN-and U-Net-based models, further enhanced through 3D reconstruction, multimodal fusion, hybrid priors, and multi-view/2.5D integration strategies López-Linares et al. [2018], Wang et al. [2018], López-Linares et al. [2017], Caradu et al. [2021], Lareyre et al. [2021], Jung et al. [2022], Abdolmanafi et al. [2023], Hwang et al. [2022], Wang et al. [2022], achieving high Dice similarity coefficients (DSC) . To enhance volumetric consistency, 3D models, especially those based on 3D U-Net and its variants, were later introduced Kongrat et al. [2022], Mu et al. [2023], Kim et al. [2024], Lyu et al. [2024], Robbi et al. [2025], Zhang et al. [2025], reporting DSC values between .804 and .987 in well-curated datasets. Furthermore, sophisticated frameworks have tackled the challenges of postoperative aortic repair imaging using Mask R-CNN and Bi-CLSTM architectures Hwang et al. [2022], Jung et al. [2022] . Nevertheless, purely data-driven models often face difficulties with out-of-distribution (OOD) data Guo et al. Figure 1: CTA in which the thrombus boundary detection is non-trivial, along with the ground truth (blue labeled) and the intuition of using calcifications as positive inductive bias to guide the network shown in axial and 3D views (r

FAQs

Why is intraluminal thrombus segmentation challenging in abdominal aortic aneurysm CTAs?

Boundaries are often indistinct because thrombus-to-adjacent-tissue contrast in HU values is low, and thrombus features vary greatly between and within patients. These issues can increase misclassification rates.

How does the proposed method reduce domain shift across different CTA protocols?

It performs patient-specific intensity normalization using a Gaussian Mixture Model of local anatomy. Subject-specific HU distributions derived from detected vascular structures help minimize inter-scan intensity variability.

What is uncertainty-gated anatomical attention and what problem does it address?

The UGAA module injects spatial anatomical priors but scales their influence using voxel-wise confidence. When prior confidence is low, the model suppresses unreliable anatomical bias and relies more on learned image features.

Jul 16, 2026

Upload by Logic