Addressing data annotation scarcity in Brain Tumor Segmentation on 3D MRI scan Using a Semi-Supervised Teacher-Student Framework
Jiaming Liu, Cheng Ding, and Daoqiang Zhang · 2026
Read Addressing data annotation scarcity in Brain Tumor Segmentation on 3D MRI scan Using a Semi-Supervised Teacher-Student Framework by Jiaming Liu, Cheng Ding, and Daoqiang Zhang (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/arxiv-2602-08797.txt.
Description
Segmenting brain tumors in MRI scans is hard because labels are costly and scanners vary across sites. Jiaming Liu and two co-authors propose a teacher-student framework for limited labels: a teacher that labels with uncertainty guides a student that learns in stages by confidence. Their arXiv paper ranks unlabeled scans by image-level confidence and introduces them in stages.
A dual-loss objective trains the student on high-confidence regions while unlearning low-confidence ones, and agreement-based refinement improves label quality. On BraTS 2021, validation Dice rises from 0.393 at ten percent data to 0.872 at full data, with the largest early gains. The student even surpasses the teacher on tumor subregions. Readers get a data-efficient segmentation recipe robust to noisy labels.
Catalog record
| CREATOR | Jiaming Liu, Cheng Ding, and Daoqiang Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08797