Adaptive Learning Strategies for Mitotic Figure Classification in MIDOG2025 Challenge
Biwen Meng, Xi Long, Jingxin Liu · 2025
Read Adaptive Learning Strategies for Mitotic Figure Classification in MIDOG2025 Challenge by Biwen Meng, Xi Long, Jingxin Liu (2025) 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-2509-02640.txt.
Description
Biwen Meng, Xi Long, and Jingxin Liu present a 2025 computational pathology paper on classifying atypical mitotic figures, abnormal cell divisions that signal disease, for the MIDOG 2025 challenge. Odd shapes and scanner differences make the task unreliable.
The team adapted a pathology foundation model called UNI2 in three ways. Variants used low-rank tuning alone, visual prompt tuning with stain normalisation, and prompt tuning plus a domain-adversarial component with test-time stain augmentation. Stain handling clearly helped generalisation. The strongest version combined prompt tuning with two stain normalisation methods applied at test time. It reached high balanced accuracy and ranking scores on the preliminary leaderboard, placing among the top ten teams. Prompt-based adaptation plus stain care looks promising for varied imaging conditions.
Catalog record
| CREATOR | Biwen Meng, Xi Long, Jingxin Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 9 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-02640