MPGR-Net: multi-scale patch graph reasoning for weakly supervised pathological image segmentation
Xu Zhang, Huaiju Ge, Yanling Wang, Chongqi Wei, Yang Liu, Xinbin Zhang, Cheng Li · 2026
Read MPGR-Net: multi-scale patch graph reasoning for weakly supervised pathological image segmentation by Xu Zhang, Huaiju Ge, Yanling Wang, Chongqi Wei, Yang Liu, Xinbin Zhang, Cheng Li (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/PMC13505138.txt.
Description
Xu Zhang and co-authors present a computer method for outlining diseased regions in pathology images when only rough labels are available. Drawing exact boundaries by hand is slow and needs expert time, so their system learns from image-level labels instead. The design builds on a vision transformer, a model that first cuts each image into small patches.
Their network, called MPGR-Net, treats the patches as nodes in a graph and links neighbours at several scales, so local detail and wider context can inform each other. A light attention step then ties global features back to local ones through channel adjustment. Tests on pathology benchmarks showed better accuracy and steadiness than earlier weakly supervised methods. Removing the graph module weakened the results, which supports its central role in the design.
Catalog record
| CREATOR | Xu Zhang, Huaiju Ge, Yanling Wang, Chongqi Wei, Yang Liu, Xinbin Zhang, Cheng Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 80412 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |