Multiscale tumor characterization in histopathology via self-distilled transformers and topology-aware visual encoding
Tanvir H. Sardar, P. Naresh, Sk Mahmudul Hassan, Praveen Kulkarni, Rajesh T. M, B. M. Ahamed Shafeeq · 2025
Read Multiscale tumor characterization in histopathology via self-distilled transformers and topology-aware visual encoding by Tanvir H. Sardar, P. Naresh, Sk Mahmudul Hassan, Praveen Kulkarni, Rajesh T. M, B. M. Ahamed Shafeeq (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/pmc-12708731.txt.
Description
Tanvir Sardar and co-authors characterize tumors in tissue slides across magnifications with a transformer framework. Their 2025 paper tackles whole slide images, whose size and detail strain models that must stay accurate at every zoom. Prior methods often ignore tissue structure or pixel context.
Their system blends a self-distilled vision transformer with topology-aware encoding. One part enforces steady predictions between two magnification levels, weighted by uncertainty. A graph-based regularizer bakes spatial priors from superpixel maps into attention. A curriculum scheduler orders training samples from simple to complex tissue patterns by entropy. A dual-modality branch joins stain image features with shape measures of cell clusters. The authors aim for diagnostic reliability that survives resolution and context shifts.
Catalog record
| CREATOR | Tanvir H. Sardar, P. Naresh, Sk Mahmudul Hassan, Praveen Kulkarni, Rajesh T. M, B. M. Ahamed Shafeeq |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |