Research map / Label-efficient and robust learning / Semi-supervised learning
AbdomenCT-1K: Is Abdominal Organ Segmentation a Solved Problem?
Jun Ma, Yao Zhang, Song Gu, Cheng Gang Zhu, Cheng Ge, Yichi Zhang, Xingle An, Congcong Wang, Qiyuan Wang, Xin Liu, Shucheng Cao, Qi Zhang and 5 more
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 |
|---|---|---|---|
| test on new-center data (KiTS 210) instead of same-source LiTS (27) | nnU-Net liver model · Liver segmentation | DSC (%): 97.4 -> 94.9 | Hurt |
| test on new-center LiTS (131) instead of same-source KiTS (42) | nnU-Net kidney model · Kidney segmentation | DSC (%): 97.1 -> 87.5 | Hurt |
| test on new-center KiTS (210) instead of same-source Spleen | nnU-Net spleen model · Spleen segmentation | DSC (%): 97.2 -> 86.6 | Hurt |
| test on KiTS (210) instead of MSD Pan. (56) | nnU-Net pancreas model · Pancreas segmentation | DSC (%): 86.1 -> 80.9 | Hurt |
| evaluate with NSD instead of DSC | nnU-Net multi-organ · Fully supervised benchmark liver | score (%): 97.0 -> 85.8 | Hurt |
| add LiTS/KiTS Plus cases with matching contrast phase to training (subtask 2 vs 1) | 3D nnU-Net · Fully supervised multi-organ benchmark, pancreas | DSC (%): 65.0 -> 78.1 | Improved |
| add 400 unlabelled cases via teacher-student pseudo labels (subtask 1) | Spleen-only nnU-Net · Semi-supervised benchmark average | DSC (%): 90.5 -> 91.3 | Improved |
| add 400 more unlabelled cases (subtask 2 vs 1) | teacher-student nnU-Net · Semi-supervised benchmark average | DSC (%): 91.3 -> 91.5 | Improved |
| add 400 more unlabelled cases (subtask 2 vs 1) | teacher-student nnU-Net · Semi-supervised benchmark average | NSD (%): 81.8 -> 81.8 | No clear effect |
| use labelled data for all cases (upper bound) instead of unlabelled | teacher-student nnU-Net · Semi-supervised benchmark average | DSC (%): 91.5 -> 93.6 | Improved |
| raise slice annotation rate from 5% to 15% | 2D nnU-Net · Weakly supervised benchmark | DSC (%): 78.0 -> 83.9 | Improved |
| raise slice annotation rate from 15% to 30% | 2D nnU-Net · Weakly supervised benchmark | DSC (%): 83.9 -> 84.7 | Improved |
Closest papers by meaning
- AbdomenAtlas-8K: Annotating 8,000 CT Volumes for Multi-Organ Segmentation in Three Weeks (NeurIPS 2023 · similarity 0.92)
- Touchstone Benchmark: Are We on the Right Way for Evaluating AI Algorithms for Medical Segmentation? (NeurIPS 2024 · similarity 0.86)
- PanTS: The Pancreatic Tumor Segmentation Dataset (NeurIPS 2025 · similarity 0.85)
- Versatile Medical Image Segmentation Learned from Multi-Source Datasets via Model Self-Disambiguation (CVPR 2024 · similarity 0.84)
- CAT: Coordinating Anatomical-Textual Prompts for Multi-Organ and Tumor Segmentation (NeurIPS 2024 · similarity 0.84)
- Towards a Comprehensive, Efficient and Promptable Anatomic Structure Segmentation Model Using 3D Whole-Body CT Scans (AAAI 2025 · similarity 0.83)
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.
