AGA3DNet: Anatomy-Guided Gaussian Priors with Multi-view xLSTM for 3D Brain MRI Subtype Classification

2026-05-07

AI Summary

This work addresses the challenge of effectively integrating local anatomical details with long-range contextual information in 3D brain MRI subtype classification by proposing an interpretable, densely voxel-annotated-free approach. The method first extracts anatomical phrases from radiology reports and maps them to standardized atlas regions, then generates Gaussian-weighted spatial priors via signed distance transforms. These priors are fused with image features extracted by a lightweight 3D CNN, and a multi-view xLSTM module is introduced to model global contextual dependencies. Notably, this is the first framework to embed report-guided anatomical priors as continuous spatial maps within a 3D classification architecture. Evaluated on an institutional retrospective dataset, the approach achieves more balanced classification performance and enables clinically interpretable lesion localization through the prior channels.

Abstract

Accurate 3D brain MRI subtype classification benefits from both localized anatomical cues and long-range contextual reasoning. We present AGA3DNet, a report-grounded framework that incorporates brief anatomical phrases extracted from radiology reports as a soft anatomical prior channel and fuses it with a lightweight 3D CNN and multi-view xLSTM aggregation. Specifically, extracted anatomical phrases are mapped to atlas-defined regions and converted into smooth spatial priors using a signed-distance transform followed by Gaussian weighting, providing interpretable, anatomy-grounded guidance without requiring dense voxel annotations. We evaluate AGA3DNet on a retrospective institutional brain MRI cohort for abnormal subtype discrimination and compare against reproducible 3D classification baselines. AGA3DNet achieves improved overall balance across performance metrics and supports clinically interpretable localization through the prior channel. We discuss limitations related to single-cohort evaluation and the lack of large-scale public brain MRI datasets paired with radiology reports under broadly usable terms.

Problem

• 3D brain MRI
• subtype classification
• anatomical priors
• multi-view learning
• clinical interpretability

Innovation

• anatomy-guided prior
• Gaussian spatial prior
• multi-view xLSTM
• 3D brain MRI classification
• radiology report grounding