Beyond Augmentation: Cross-Modal Transformer Fusion with Bi-directional Attention for Low-Data Aneurysm Screening

TL;DR

This paper introduces CMTF-Net, a cross-modal transformer model that improves low-data aneurysm screening by encoding vascular anatomy and providing interpretable, localized activation.

Contribution

It presents a novel anatomically structured reasoning framework that enhances aneurysm detection accuracy and interpretability in low-data scenarios.

Findings

Achieves near-perfect AUC-ROC in aneurysm screening

Maintains high precision despite class imbalance

Provides spatially localized activation maps for interpretability

Abstract

Intracranial aneurysm rupture causes subarachnoid hemorrhage with mortality near 50%, making early detection critical. Although CTA enables rapid screening, detecting small aneurysms within the complex three-dimensional branching of the Circle of Willis remains expertise-dependent. Existing automated systems are constrained by class imbalance, skull-base artifacts that mimic vascular contrast, and reliance on global binary classification without structured localization, limiting surgical relevance and interpretability. We propose CMTF-Net, a cross-modal target fusion framework that reframes aneurysm screening as anatomically structured reasoning. By supervising 14 vascular territories independently, the network encodes Circle of Willis geometry while allowing multi-segment activation, aligning model design with clinical workflow. CMTF-Net achieves near-perfect AUC-ROC with narrow confidence intervals and sustained precision under imbalance. Grad-CAM and causal maps show spatially localized activation along major arteries, supporting interpretable, anatomically grounded screening in low-data settings.