Physics-based reaction-diffusion model predicts stroke infarct growth from routine MRI
maier_ak · x · 2026-09-11
Researchers from the University of Zurich and GIK Institute published a MICCAI Satellite Events paper, "An Imaging-Informed Reaction-Diffusion Model of Infarct Growth," exploring whether a classic physics-based PDE driven directly by routine MRI can predict infarct and penumbra evolution after stroke.
- Clinical context: thrombectomy decisions hinge on predicting how much brain tissue will die; current ML algorithms are accurate but opaque, while electrophysiological models need dozens of parameters not measurable from clinical MRI.
- The study tests whether the reaction-diffusion equation, driven by already-acquired imaging data, captures enough infarct dynamics to be useful.
- Author Andreas Maier adds caveats: Dice 0.46 is only an upper bound since reaction rate and threshold were fitted to each patient's 90-day outcome; the prospective model would lack those labels, the cohort is just 29 patients, no recanalization modeling, and no comparison to deep-learning segmenters.
Related event: Physics-Based Reaction-Diffusion Model Improves Stroke Lesion Prediction(2 posts)→
More from Research
- HABIT-Bench: Best agent memory system scores just 33.61% at inferring user habits — SJTU · 2026-10-28
- MatAnyone-E brings 7.4x faster 4K video matting via sparse memory propagation — mmlab-ntu · 2026-10-28
- CodeCap-Bench: executable code beats natural language as video captions — Jiafu Tang · 2026-10-28
- NVIDIA's ThinkWorld learns robot skills from unlabeled human videos via chain-of-dynamics reasoning — nvidia · 2026-10-28
- Omni-InteractScope benchmarks proactive interaction in continuous omnimodal streams — Qwen · 2026-10-28
- Memory Tax study: dedicated agent memory systems often lose to naive RAG — UIUC-CS · 2026-10-28