Structure-Aware Stylized Image Synthesis for Robust Medical Image Segmentation
2024
Read Structure-Aware Stylized Image Synthesis for Robust Medical Image Segmentation (2024) 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-2412-04296.txt.
Description
Jie Bao, Zhixin Zhou, Wen Jung Li, and Rui Luo restyle medical images to harden segmentation models in a 2024 paper. Scanners, settings, and patients shift image appearance between hospitals, and models trained in one place stumble in another. Worse, test sites are often absent from training data entirely.
Diffusion models can repaint images into one shared style, but they tend to smear the anatomy that matters. A structure-preserving network keeps lesion location, size, and shape locked during the transfer. On colonoscopy polyps and skin lesions, models trained with the restyled images segmented more accurately and robustly than baselines. One-shot stylization makes the trick practical.
Catalog record
| YEAR | 2024 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-04296