Latent Representation Learning from 3D Brain MRI for Interpretable Prediction in Multiple Sclerosis

2025-09-28

AI Summary

To address the limited interpretability of deep learning models and insufficient representational capacity of conventional methods in predicting cognitive decline in multiple sclerosis (MS), this paper proposes InfoVAE-Med3D—a novel information-maximizing variational autoencoder tailored for 3D brain MRI. By explicitly optimizing mutual information between latent variables and input images, the model learns compact, structured, and clinically meaningful embeddings. Evaluated on brain age estimation and Symbol Digit Modalities Test (SDMT) score prediction, InfoVAE-Med3D outperforms mainstream VAE variants. Its latent space exhibits clear, semantically coherent clustering aligned with neuroanatomical and clinical knowledge, enabling interpretable biomarker discovery. This work represents the first systematic integration of mutual information–driven representation learning into MS neuroimaging analysis, achieving state-of-the-art predictive accuracy while substantially enhancing model transparency and clinical interpretability.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We present InfoVAE-Med3D, a latent-representation learning approach for 3D brain MRI that targets interpretable biomarkers of cognitive decline. Standard statistical models and shallow machine learning often lack power, while most deep learning methods behave as black boxes. Our method extends InfoVAE to explicitly maximize mutual information between images and latent variables, producing compact, structured embeddings that retain clinically meaningful content. We evaluate on two cohorts: a large healthy-control dataset (n=6527) with chronological age, and a clinical multiple sclerosis dataset from Charles University in Prague (n=904) with age and Symbol Digit Modalities Test (SDMT) scores. The learned latents support accurate brain-age and SDMT regression, preserve key medical attributes, and form intuitive clusters that aid interpretation. Across reconstruction and downstream prediction tasks, InfoVAE-Med3D consistently outperforms other VAE variants, indicating stronger information capture in the embedding space. By uniting predictive performance with interpretability, InfoVAE-Med3D offers a practical path toward MRI-based biomarkers and more transparent analysis of cognitive deterioration in neurological disease.

Problem

• Predict cognitive decline using 3D brain MRI latent representations
• Overcome black-box limitations in deep learning for medical imaging
• Identify interpretable biomarkers for multiple sclerosis and brain aging

Innovation

• Extends InfoVAE to maximize image-latent mutual information
• Produces compact structured embeddings with clinical relevance
• Outperforms other VAE variants in reconstruction and prediction tasks