Abstract
Intracranial aneurysms are small and morphologically heterogeneous lesions embedded within complex cerebrovascular structures. Accurate aneurysm detection, anatomical localization, and vessel-specific segmentation are essential for clinical assessment and treatment planning, yet remain challenging due to the extreme imbalance between aneurysms and surrounding tissues and the strong anatomical dependence of aneurysms on their parent vessels. We propose an anatomy-aware multi-task framework for joint vessel and aneurysm analysis, including a shared encoder with task-specific decoding branches to learn complementary vascular and aneurysmal representations. To exploit their anatomical relationship, vessel-derived features are progressively incorporated into the aneurysm branch through a hierarchical guidance mechanism, encouraging the network to focus on anatomically plausible regions while preserving lesion-specific information. Multi-task supervision further promotes consistent learning across vessel segmentation, aneurysm detection, anatomical localization, and vessel-specific classification. Experiments on the TopAneu 2026 benchmark demonstrate that the proposed framework achieves a precision of 0.835 for aneurysm localization and a Dice of 0.654 for vessel-specific segmentation, highlighting the effectiveness of incorporating vascular anatomy into fine-grained intracranial aneurysm analysis.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/TopAneu_026.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=lgVVNaSC3u
BibTex
@InProceedings{HanLuy_AnatomyAware_MICCAISAT2026,
author = { Han, Luyi AND Min, Pengxiang AND Tan, Tao},
title = { { Anatomy-Aware Intracranial Aneurysm Analysis with Vessel-Guided Dual-Decoder Multi-Task Learning } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17255},
month = {pending},
page = {pending}
}