A deep-learning framework for brain tumor segmentation via three-dimensional mass-preserving geometric transformation
Tsung-Ming Huang, Kai-Qian Zheng, Wen-Wei Lin, Tiexiang Li, Shing-Tung Yau · 2026
Read A deep-learning framework for brain tumor segmentation via three-dimensional mass-preserving geometric transformation by Tsung-Ming Huang, Kai-Qian Zheng, Wen-Wei Lin, Tiexiang Li, Shing-Tung Yau (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-13158315.txt.
Description
Tsung-Ming Huang and co-authors reshape brain scans into cubes so a neural network can outline tumors more cleanly. Crooked anatomy makes tumor borders hard to trace in MRI volumes. Their mass-preserving geometric transformation uses a homotopy method to morph each irregular scan into a standard solid cube, keeping local mass shares and the global layout intact.
Fed into the popular nnU-Net, the tidy cubes sharpen the outlining. A two-phase segmentation keeps the runtime short, and a cleanup step lifts the scoring on single lesions. On the BraTS 2023 challenge data, the method's Dice scores matched or beat the top entries across whole tumor, tumor core, and the enhancing rim. The authors pitch geometry as a cheap ally for medical AI.
Catalog record
| CREATOR | Tsung-Ming Huang, Kai-Qian Zheng, Wen-Wei Lin, Tiexiang Li, Shing-Tung Yau |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 196 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |