Research map / Time series / Causal and generative time series
CRONOS: Continuous time reconstruction for 4D medical longitudinal series
Nico Disch, Saikat Roy, Constantin Ulrich, Yannick Kirchhoff, Maximilian Rouven Rokuss, Robin Peretzke, David Zimmerer, Klaus Maier-Hein
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add last-observed-carry-forward (LOCF) (ablation) | discrete CRONOS · ACDC validation | PSNR: 27.30 -> 32.30 | Improved |
| add LOCF on top of masking (ablation) | discrete CRONOS · ACDC validation | PSNR: 31.20 -> 32.30 | Improved |
| add attention to the U-Net (ablation) | discrete CRONOS · ACDC validation | PSNR: 31.88 -> 32.30 | Improved |
| aggregate by mean instead of last image at inference (ablation) | discrete CRONOS · ACDC validation | PSNR: 31.87 -> 31.88 | No clear effect |
| add explicit timestamp conditioning (continuous) (ablation) | CRONOS · ACDC (continuous, 2 of 4 frames) | PSNR: 33.534 -> 36.007 | Improved |
| add explicit timestamp conditioning (continuous) (ablation) | CRONOS · ACDC (continuous setting) | SSIM: 93.27 -> 93.86 | Improved |
| use continuous instead of discrete variant (ablation) | CRONOS · ACDC (discrete, many-to-one) | PSNR: 30.510 -> 29.750 | Hurt |
| use continuous instead of discrete variant | CRONOS · Lumiere (discrete, many-to-one) | PSNR: 22.427 -> 22.551 | Improved |
| increase feature size from 8 to 64 (ablation) | CRONOS · ACDC validation | PSNR: 32.07 -> 32.35 | Improved |
| increase NFE from 1 to 5 (ablation) | CRONOS · ACDC validation | PSNR: 31.47 -> 32.47 | Improved |
| increase NFE from 10 to 200 (ablation) | CRONOS · ACDC validation | PSNR: 32.24 -> 32.47 | No clear effect |
| add training noise 0.3 (ablation) | CRONOS · ACDC validation | SSIM: 96.06 -> 95.78 | No clear effect |
Closest papers by meaning
- Progression as Latent Drift: Generative Forecasting of Slow-Evolving Pathologies (ECCV 2026 · similarity 0.87)
- Autoregressive Sequence Modeling for 3D Medical Image Representation (AAAI 2025 · similarity 0.87)
- Time-to-Event Pretraining for 3D Medical Imaging (ICLR 2025 · similarity 0.87)
- Knowledge Transfer Scaling Laws for 3D Medical Imaging (NeurIPS 2026 · similarity 0.86)
- Multimodal Disease Progression Modeling via Spatiotemporal Disentanglement and Multiscale Alignment (NeurIPS 2025 · similarity 0.86)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
