RegionFM: Interpretable Region-Based Brain MRI Classification Using Foundation Model Embeddings

2026-07-15

AI Summary

This work addresses the limited interpretability of existing foundation models for brain MRI, whose explanations are typically confined to voxel- or patch-level outputs without alignment to clinically meaningful anatomical regions. The authors propose a region-additive prediction framework that first segments brain MRI scans into anatomical regions, then leverages a frozen pretrained foundation model to generate embeddings for each region. These region-specific embeddings are combined via an additive logistic regression model, enabling each brain region’s diagnostic contribution to be explicitly quantified as a scalar weight. This approach uniquely integrates anatomical segmentation with foundation model embeddings, achieving classification performance on par with fine-tuned models in cognitive impairment tasks while offering anatomically grounded interpretability. Ablation studies on the embeddings further confirm their semantic validity.

Abstract

Foundation models provide powerful representations for brain MRI analysis, but their predictions remain difficult to interpret in anatomically meaningful terms. Clinical assessment of brain MRI is commonly organized around anatomically defined structures and regional abnormalities, whereas conventional explanation methods typically produce voxel- or patch-level importance maps that do not explicitly quantify the contributions of individual brain regions. To address this mismatch, we propose RegionFM, an interpretable framework that integrates anatomical segmentation with brain MRI foundation-model embeddings. RegionFM first divides each MRI scan into anatomical regions and constructs a separate MRI volume for each region. A frozen foundation model then encodes each region into an embedding, and a region-additive logistic model combines these embeddings such that every anatomical region contributes an explicit scalar term to the final prediction. This formulation supports both subject-level and cohort-level analyses of regional contributions. We evaluate RegionFM on cognitive-impairment classification using embeddings from multiple pretrained brain MRI foundation models. The results show that RegionFM maintains performance comparable to less interpretable fine-tuning approaches while providing anatomically grounded explanations. Randomized embedding ablations yield near-chance performance, indicating that the predictions rely on meaningful structure captured by the foundation-model embeddings rather than simple feature statistics. Overall, RegionFM better aligns model explanations with anatomy-based clinical reasoning while maintaining competitive predictive performance.

Problem

• interpretable AI
• brain MRI classification
• anatomical regions
• foundation models
• model explainability

Innovation

• interpretable AI
• foundation models
• region-based analysis
• brain MRI classification
• anatomical segmentation