How Good are Foundation Models in Longitudinal MRI Disease Progression Reasoning?

2026-08-13

AI Summary

This work addresses the limited spatiotemporal reasoning and change localization capabilities of current vision-language models in longitudinal, multi-view MRI for tracking disease progression. The authors introduce the first time-aware, multi-view MRI benchmark specifically designed for monitoring brain disease evolution, integrating multi-planar anatomical images, longitudinal temporal sequences, and structured spatial grounding annotations. The benchmark encompasses seven disease categories, 890 patients, over 3,200 timepoints, and 3,920 expert-annotated question-answer pairs, supporting open-ended, multiple-choice, and binary classification tasks. Experiments across 16 models reveal that while existing architectures exhibit rudimentary temporal alignment, they systematically fail in identifying change directionality and quantifying volumetric alterations. Although multi-view inputs enhance spatial localization, they degrade temporal reasoning in compact architectures, underscoring the critical need for clinically grounded evaluation of temporal reasoning in medical vision-language systems.

Abstract

Magnetic Resonance Imaging (MRI) interpretation is fundamental to clinical decision-making, requiring radiologists to integrate multi-view anatomical planes across sequential timepoints while precisely localizing interval changes. However, existing vision-language benchmarks remain confined to single-timepoint, single-view interpretation, failing to capture the temporal-spatial reasoning essential to radiologic practice. We introduce the Time-Aware Multi-View MRI Benchmark, an evaluation framework unifying multi-view anatomical input, temporal reasoning across longitudinal scans, and structured localization guidance. The benchmark comprises 3,920 expert-verified question-answer pairs derived from 890 patients across over 3,200 longitudinal MRI timepoints, drawn from seven clinical cohorts covering glioblastoma, neurodegeneration, vestibular schwannoma, and brain metastases, in open-ended, multiple-choice, and binary formats, requiring models to identify anatomical regions of maximal change, characterize progression across sequences and views, and provide structured guidance specifying boundaries, imaging features, and confounders. Experiments across 16 vision-language models reveal moderate temporal alignment but systematic failure on change direction recognition and volumetric quantification, while multi-view inputs improve spatial localization yet degrade temporal reasoning in compact architectures. Our benchmark provides a systematic framework for evaluating progression tracking, interval change localization, and temporal ordering, which are essential for clinical deployment. Code, evaluation splits, and the dataset are available at: https://github.com/wafaAlghallabi/Time-Aware-MRI.

Problem

• longitudinal MRI
• disease progression
• temporal-spatial reasoning
• vision-language models
• interval change localization

Innovation

• longitudinal MRI
• temporal-spatial reasoning
• multi-view integration
• disease progression benchmark
• vision-language models