Enhancing weakly supervised 3D medical image segmentation through probabilistic-aware learning
Runmin Jiang, Zhaoxin Fan, Junhao Wu, Lenghan Zhu, Xin Huang, Tianyang Wang, Heng Huang, Min Xu · 2026
Read Enhancing weakly supervised 3D medical image segmentation through probabilistic-aware learning by Runmin Jiang, Zhaoxin Fan, Junhao Wu, Lenghan Zhu, Xin Huang, Tianyang Wang, Heng Huang, Min Xu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13604003.txt.
Description
Runmin Jiang and colleagues built a light-label pipe for 3D scan cuts that needs far less hand-marked truth. Full-label scan cuts eat hours of hand work, most of all in 3D. Their chance-aware pipe makes thick fake labels from thin marks, reads traits with a chance-aware transformer net, and trains with a loss that heeds mark trust.
On CT and MRI sets, the pipe matched full-label work and beat past light-label work, with Dice gains up to 18.1 percent for some organs. By heeding doubt all through training, the pipe cuts bias from noisy fake labels. Open code invites wide tests in clinic scan work.
Catalog record
| CREATOR | Runmin Jiang, Zhaoxin Fan, Junhao Wu, Lenghan Zhu, Xin Huang, Tianyang Wang, Heng Huang, Min Xu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |