Medical image model predicts anatomy changes with uncertainty
MUMINS: Metadata-conditioned Uncertainty-aware Medical Image Next-state Synthesis
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Medical images can show how parts of the body might change over time, like a tumor growing or brain shrinking. This is hard to predict because changes are small, vary a lot between people, and have some randomness. The authors created a new computer method called MUMINS that makes these predictions better by guessing the next scan from a starting one and showing where it might be wrong, all in one go. It works on different body parts without needing special adjustments and beats or matches other top methods in tests.
What this means in practice
- •For medical imaging teams: Generate patient-specific future scans that include error maps to improve diagnosis and treatment planning.
- •For medical device developers: Integrate uncertainty-aware image prediction into devices to better monitor disease progression in lung CT or brain MRI.$Commercial implications: Enables development of diagnostic tools that forecast disease changes with uncertainty indicators, improving clinical decision-making.
Authors
Anna Oliveras, Roger Marí, Rafael Redondo, Oriol Guardià, Cynthia Ifeyinwa Ugwu, Ana Tost, Bhalaji Nagarajan, Carolina Migliorelli, Vicent Ribas, Petia Radeva
Abstract
Forecasting anatomical changes such as tumor growth and neurodegeneration is a challenging generative vision task. Morphological evolution is subtle relative to static anatomy, highly patient-specific, and inherently stochastic. Existing methods struggle with several issues: deterministic networks ignore biological stochasticity, while standard diffusion models require computationally prohibitive multi-pass sampling to quantify uncertainty. We propose MUMINS (Metadata-conditioned Uncertainty-aware Medical Image Next-state Synthesis), an efficient diffusion framework that jointly diffuses a baseline scan and its follow-up residual, summed to synthesize the follow-up scan, while concurrently predicting a spatial uncertainty map, in a single reverse diffusion process. Conditioned on the time interval and relevant metadata, it preserves fine-grained anatomy by dynamically re-injecting the baseline as a soft anchor at every denoising step, and a negative-log-likelihood head learns the uncertainty map to explicitly flag error-prone regions. Designed without organ-specific heuristics, the same architecture is reused across anatomies via separate, dataset-specific retraining. Extensive evaluations demonstrate that dataset-specific retraining of MUMINS matches or outperforms dedicated, domain-specific state-of-the-art methods on lung CT (PNG) and brain MRI (OASIS-3). Project page: https://github.com/aolivtous/MUMINS.