Forecasting how 3D medical scans evolve over time is important for disease progression, treatment planning, and developmental assessment. Yet existing models either rely on a single prior scan, fixed grid times, or target global labels, which limits voxel-level forecasting under irregular sampling. We present CRONOS, a unified framework for many-to-one prediction from multiple past scans that supports both discrete (grid-based) and continuous (real-valued) timestamps in one model, to the best of our knowledge the first to achieve continuous sequence-to-image forecasting for 3D medical data. CRONOS learns a spatio-temporal velocity field that transports context volumes toward a target volume at an arbitrary time, while operating directly in 3D voxel space. Across three public datasets spanning Cine-MRI, perfusion CT, and longitudinal MRI, CRONOS outperforms other baselines, while remaining computationally competitive. We will release code and evaluation protocols to enable reproducible, multi-dataset benchmarking of multi-context, continuous-time forecasting.
@article{arxiv.2512.16577,
title = {CRONOS: Continuous Time Reconstruction for 4D Medical Longitudinal Series},
author = {Nico Albert Disch and Saikat Roy and Constantin Ulrich and Yannick Kirchhoff and Maximilian Rokuss and Robin Peretzke and David Zimmerer and Klaus Maier-Hein},
journal= {arXiv preprint arXiv:2512.16577},
year = {2025}
}