mlsanityData to 7 Oct 2026

Scaling Tumor Segmentation: Best Lessons from Real and Synthetic Data

arXiv 2510.14831 ↗First posted 16 Oct 2025cs.CVcited by 13

Not on the map. Why?

Abstract (arXiv)

AI for tumor segmentation is limited by the lack of large, voxel-wise annotated datasets, which are hard to create and require medical experts. In our proprietary JHH dataset of 3,000 annotated pancreatic tumor scans, we found that AI performance stopped improving after 1,500 scans. With synthetic data, we reached the same performance using only 500 real scans. This finding suggests that synthetic data can steepen data scaling laws, enabling more efficient model training than real data alone. Motivated by these lessons, we created AbdomenAtlas 2.0--a dataset of 10,135 CT scans with a total of 15,130 tumor instances per-voxel manually annotated in six organs (pancreas, liver, kidney, colon, esophagus, and uterus) and 5,893 control scans. Annotated by 23 expert radiologists, it is several orders of magnitude larger than existing public tumor datasets. While we continue expanding the dataset, the current version of AbdomenAtlas 2.0 already provides a strong foundation--based on lessons from the JHH dataset--for training AI to segment tumors in six organs. It achieves notable improvements over public datasets, with a +7% DSC gain on in-distribution tests and +16% on out-of-distribution tests.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 69 of its 107 references that are in the map. First posted on arXiv 16 Oct 2025.

Builds on

69 of 107 references.

References · 69

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2020Scaling Laws for Neural Language Modelscited by 6,411
  5. 2022Scalable Diffusion Models with Transformerscited by 5,931
  6. 2021Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervisioncited by 3,848
  7. 2016V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentationcited by 2,284
  8. 2021TransUNet: Transformers Make Strong Encoders for Medical Image Segmentationcited by 1,552
  9. 2018UNet++: A Nested U-Net Architecture for Medical Image Segmentationcited by 1,518
  10. 20163D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotationcited by 1,471
  11. 2021UNETR: Transformers for 3D Medical Image Segmentationcited by 954
  12. 2022Swin UNETR: Swin Transformers for Semantic Segmentation of Brain Tumors in MRI Imagescited by 823
38 more
  1. 2021The Medical Segmentation Decathloncited by 620
  2. 2019UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentationcited by 583
  3. 2021Solving Inverse Problems in Medical Imaging with Score-Based Generative Modelscited by 514
  4. 2020Scaling Laws for Autoregressive Generative Modelingcited by 510
  5. 2019The Liver Tumor Segmentation Benchmark (LiTS)cited by 504
  6. 2021Self-Supervised Pre-Training of Swin Transformers for 3D Medical Image Analysiscited by 418
  7. 2022AMOS: A Large-Scale Abdominal Multi-Organ Benchmark for Versatile Medical Image Segmentationcited by 347
  8. 20183D MRI brain tumor segmentation using autoencoder regularizationcited by 324
  9. 2024nnU-Net Revisited: A Call for Rigorous Validation in 3D Medical Image Segmentationcited by 235
  10. 2023CLIP-Driven Universal Model for Organ Segmentation and Tumor Detectioncited by 224
  11. 2020AbdomenCT-1K: Is Abdominal Organ Segmentation A Solved Problem?cited by 222
  12. 2021Diffusion Models for Implicit Image Segmentation Ensemblescited by 198
  13. 2022Unsupervised Medical Image Translation with Adversarial Diffusion Modelscited by 191
  14. 2019The KiTS19 Challenge Data: 300 Kidney Tumor Cases with Clinical Context, CT Semantic Segmentations, and Surgical Outcomescited by 168
  15. 2015DeepOrgan: Multi-level Deep Convolutional Networks for Automated Pancreas Segmentationcited by 158
  16. 2023Scaling Laws of Synthetic Images for Model Training ... for Nowcited by 107
  17. 2023The KiTS21 Challenge: Automatic segmentation of kidneys, renal tumors, and renal cysts in corticomedullary-phase CTcited by 101
  18. 2023STU-Net: Scalable and Transferable Medical Image Segmentation Models Empowered by Large-Scale Supervised Pre-trainingcited by 96
  19. 2024MAISI: Medical AI for Synthetic Imagingcited by 86
  20. 2022Conversion Between CT and MRI Images Using Diffusion and Score-Matching Modelscited by 69
  21. 2021Transferable Visual Words: Exploiting the Semantics of Anatomical Patterns for Self-supervised Learningcited by 68
  22. 2024Towards Generalizable Tumor Synthesiscited by 65
  23. 2022Measurement-conditioned Denoising Diffusion Probabilistic Model for Under-sampled Medical Image Reconstructioncited by 64
  24. 2023GenerateCT: Text-Conditional Generation of 3D Chest CT Volumescited by 64
  25. 2024AbdomenAtlas: A Large-Scale, Detailed-Annotated, & Multi-Center Dataset for Efficient Transfer Learning and Open Algorithmic Benchmarkingcited by 64
  26. 2022PET image denoising based on denoising diffusion probabilistic modelscited by 57
  27. 2023Label-Free Liver Tumor Segmentationcited by 56
  28. 2022Diffusion Adversarial Representation Learning for Self-supervised Vessel Segmentationcited by 55
  29. 2019C2FNAS: Coarse-to-Fine Neural Architecture Search for 3D Medical Image Segmentationcited by 49
  30. 2021DiNTS: Differentiable Neural Network Topology Search for 3D Medical Image Segmentationcited by 47
  31. 2019Generating large labeled data sets for laparoscopic image processing tasks using unpaired image-to-image translationcited by 38
  32. 2022Can segmentation models be trained with fully synthetically generated data?cited by 37
  33. 2025RadGPT: Constructing 3D Image-Text Tumor Datasetscited by 34
  34. 2024Universal and Extensible Language-Vision Models for Organ Segmentation and Tumor Detection from Abdominal Computed Tomographycited by 31
  35. 2025Medical World Model: Generative Simulation of Tumor Evolution for Treatment Planningcited by 31
  36. 2024The ULS23 Challenge: a Baseline Model and Benchmark Dataset for 3D Universal Lesion Segmentation in Computed Tomographycited by 26
  37. 2017DeepLesion: Automated Deep Mining, Categorization and Detection of Significant Radiology Image Findings using Large-Scale Clinical Lesion Annotationscited by 23
  38. 2022A Novel Unified Conditional Score-based Generative Framework for Multi-modal Medical Image Completioncited by 23

50 most cited of 69 All 69 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 4; 2026: 9.

13 papers.

Sort:

By area:

  1. 2025Learning Patient-Specific Disease Dynamics with Latent Flow Matching for Longitudinal Imaging Generationcited by 8
  2. 2025RadImageNet-VQA: A Large-Scale CT and MRI Dataset for Radiologic Visual Question Answeringcited by 6
  3. 2026RadThinking: A Dataset for Longitudinal Clinical Reasoning in Radiologycited by 4
  4. 2025See More, Change Less: Anatomy-Aware Diffusion for Contrast Enhancementcited by 3
  5. 2025USB: Unified Synthetic Brain Framework for Bidirectional Pathology-Healthy Generation and Editingcited by 1
  6. 2026Glance and Focus Reinforcement for Pan-cancer Screeningcited by 1
  7. 2026CORTEX: A Structured Reasoning Benchmark for Trustworthy 3D Chest CT MLLMscited by 1
  8. 2026A Real-Calibrated Synthetic-First Data Enginecited by 0
  9. 2026Beyond Masks: The Case for Medical Image Parsingcited by 0
  10. 2026Report Supervisioncited by 0
  11. 2026RT-Super: Learning Tumor Segmentation from Longitudinal Images and Reportscited by 0
  12. 2026Merlin Plus: A Large-Scale, Multi-Cancer, Image-Mask-Report Datasetcited by 0
1 more
  1. 2026PCaPaint: Prostate Cancer Inpainting by Mitigating Shortcut Learningcited by 0