An Interpretable Local Editing Model for Counterfactual Medical Image Generation
Hyungi Min, Taeseung You, Hangyeul Lee, Yeongjae Cho, Sungzoon Cho · 2026
Read An Interpretable Local Editing Model for Counterfactual Medical Image Generation by Hyungi Min, Taeseung You, Hangyeul Lee, Yeongjae Cho, Sungzoon Cho (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/arxiv-2603-00423.txt.
Description
Min and four colleagues build InstructX2X, a medical image editor that answers what if questions by changing only the disease region, leaving demographic traits untouched. Past editors leak changes into demographics and hide their reasoning. Region specific editing plus a Guidance Map that shows where and why edits happen give users a readable trail. A new expert checked instruction set from medical question pairs supports training, and chest X-ray tests lead rival methods across major scores.
Catalog record
| CREATOR | Hyungi Min, Taeseung You, Hangyeul Lee, Yeongjae Cho, Sungzoon Cho |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29093 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.00423