GeoSAE: Geometric Prior-Guided Layer-Wise Sparse Autoencoder Annotation of Brain MRI Foundation Models

• URL: http://arxiv.org/abs/2605.01829v1
• Date: Sun, 03 May 2026 11:42:25 GMT
• Title: GeoSAE: Geometric Prior-Guided Layer-Wise Sparse Autoencoder Annotation of Brain MRI Foundation Models
• Abstract summary: GeoSAE identifies a compact, fully interpretable feature set that predicts mild cognitive impairment (MCI)-to-AD conversion.<n>This shows that geometry-guided SAEs can extract interpretable, biomarkers from frozen brain MRI foundation models.
• Score: 8.51943483922984
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Brain MRI foundation models learn rich representations of anatomy, but interpreting what clinical information they encode remains an open problem. Standard sparse autoencoders (SAEs) suffer from severe feature collapse in deep transformer layers, and in Alzheimer's disease (AD) research, aging confounds nearly every clinical variable, making naive annotation unreliable. We propose GeoSAE, a geometry-guided SAE framework that uses the foundation model's learned manifold structure to prevent feature collapse and annotates each surviving feature via age-deconfounded partial correlations. Applied to ~14k T1-weighted MRI scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Australian Imaging biomarkers and Lifestyle (AIBL) datasets, GeoSAE identifies a compact, fully interpretable feature set that predicts mild cognitive impairment (MCI)-to-AD conversion (AUC 0.746) using only 2% of the embedding dimensions, while comorbidity-annotated features achieve only chance-level performance. The identified features replicate across cohorts without retraining (r=0.97) and localize to neuroanatomically distinct regions consistent with Braak staging. This shows that geometry-guided SAEs can extract interpretable, biomarkers from frozen brain MRI foundation models.

Related papers

• BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining [4.651750987298774]
  We introduce BrainNorm, a normative foundation model, trained and tested on 66,000 T1-weighted structural MRI (T1w sMRI) scans.<n>BrainNorm learns a Semantic Atlas Latent space (SAL), where each scan is represented as a set of atlas-parcel embeddings.<n>The resulting deviation patterns in SAL space enable zero-shot tasks for disease prediction using parcel-wise abnormalities.
  arXiv  Detail & Related papers  (2026-08-18T08:45:02Z)
• MIRAGE: Knowledge Graph-Guided Cross-Cohort MRI Synthesis for Alzheimer's Disease Prediction [15.543131466384658]
  We introduce MIRAGE, a novel framework that reframes the missing-MRI problem as an anatomy-guided cross-modal latent distillation task.<n>We employ a frozen pre-trained 3D U-Net decoder strictly as an auxiliary regularization engine.<n>Experiments demonstrate that our framework successfully bridges the missing-modality gap, improving the AD classification rate by 13%.
  arXiv  Detail & Related papers  (2026-03-02T22:17:37Z)
• Four-Stage Alzheimer's Disease Classification from MRI Using Topological Feature Extraction, Feature Selection, and Ensemble Learning [0.8374077003751697]
  TDA-Alz is a novel framework for four-stage Alzheimer's disease severity classification using topological data analysis and ensemble learning.<n>It achieves an accuracy of 98.19% and an AUC of 99.75%, outperforming or matching state-of-the-art deep learning-based methods reported on datasets.
  arXiv  Detail & Related papers  (2026-01-01T07:44:51Z)
• R-GenIMA: Integrating Neuroimaging and Genetics with Interpretable Multimodal AI for Alzheimer's Disease Progression [63.97617759805451]
  Early detection of Alzheimer's disease requires models capable of integrating macro-scale neuroanatomical alterations with micro-scale genetic susceptibility.<n>We introduce R-GenIMA, an interpretable multimodal large language model that couples a novel ROI-wise vision transformer with genetic prompting.<n>R-GenIMA achieves state-of-the-art performance in four-way classification across normal cognition, subjective memory concerns, mild cognitive impairment, and AD.
  arXiv  Detail & Related papers  (2025-12-22T02:54:10Z)
• A Semantically Enhanced Generative Foundation Model Improves Pathological Image Synthesis [82.01597026329158]
  We introduce a Correlation-Regulated Alignment Framework for Tissue Synthesis (CRAFTS) for pathology-specific text-to-image synthesis.<n>CRAFTS incorporates a novel alignment mechanism that suppresses semantic drift to ensure biological accuracy.<n>This model generates diverse pathological images spanning 30 cancer types, with quality rigorously validated by objective metrics and pathologist evaluations.
  arXiv  Detail & Related papers  (2025-12-15T10:22:43Z)
• Enhancing Alzheimer's Diagnosis: Leveraging Anatomical Landmarks in Graph Convolutional Neural Networks on Tetrahedral Meshes [16.100668392804476]
  Alzheimer's disease (AD) is a major neurodegenerative condition that affects millions around the world.<n>Brain structural magnetic resonance imaging (sMRI) may provide a safer and more convenient solution for the AD diagnosis.<n>Recent advances in geometric deep learning have facilitated sMRI analysis and early diagnosis of AD.
  arXiv  Detail & Related papers  (2025-03-06T23:02:18Z)
• Knowledge-Guided Prompt Learning for Lifespan Brain MR Image Segmentation [53.70131202548981]
  We present a two-step segmentation framework employing Knowledge-Guided Prompt Learning (KGPL) for brain MRI. Specifically, we first pre-train segmentation models on large-scale datasets with sub-optimal labels. The introduction of knowledge-wise prompts captures semantic relationships between anatomical variability and biological processes.
  arXiv  Detail & Related papers  (2024-07-31T04:32:43Z)
• SNeurodCNN: Structure-focused Neurodegeneration Convolutional Neural Network for Modelling and Classification of Alzheimer's Disease [0.0]
  Alzheimer's disease (AD), the predominant form of dementia, is a growing global challenge. Current clinical diagnoses rely on radiologist expert interpretation, which is prone to human error. This paper proposes a deep learning framework that includes a novel structure-focused neurodegeneration CNN architecture named SNeurodCNN.
  arXiv  Detail & Related papers  (2024-01-08T14:33:57Z)
• Patched Diffusion Models for Unsupervised Anomaly Detection in Brain MRI [55.78588835407174]
  We propose a method that reformulates the generation task of diffusion models as a patch-based estimation of healthy brain anatomy. We evaluate our approach on data of tumors and multiple sclerosis lesions and demonstrate a relative improvement of 25.1% compared to existing baselines.
  arXiv  Detail & Related papers  (2023-03-07T09:40:22Z)
• G-MIND: An End-to-End Multimodal Imaging-Genetics Framework for Biomarker Identification and Disease Classification [49.53651166356737]
  We propose a novel deep neural network architecture to integrate imaging and genetics data, as guided by diagnosis, that provides interpretable biomarkers. We have evaluated our model on a population study of schizophrenia that includes two functional MRI (fMRI) paradigms and Single Nucleotide Polymorphism (SNP) data.
  arXiv  Detail & Related papers  (2021-01-27T19:28:04Z)
• A Graph Gaussian Embedding Method for Predicting Alzheimer's Disease Progression with MEG Brain Networks [59.15734147867412]
  Characterizing the subtle changes of functional brain networks associated with Alzheimer's disease (AD) is important for early diagnosis and prediction of disease progression. We developed a new deep learning method, termed multiple graph Gaussian embedding model (MG2G) We used MG2G to detect the intrinsic latent dimensionality of MEG brain networks, predict the progression of patients with mild cognitive impairment (MCI) to AD, and identify brain regions with network alterations related to MCI.
  arXiv  Detail & Related papers  (2020-05-08T02:29:24Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.