Context‐informed personalized segmentation from radiology reports: A fully automated framework for long‐term liver tumor follow‐up
Jaehee Chun, McKell Woodland, Austin Castelo, Caleb O'Connor, Mais Altaie, Aashish Gupta, Shanli Ding, Allyson Nguyen, Eugene J Koay, Kristy K Brock · 2026
Read Context‐informed personalized segmentation from radiology reports: A fully automated framework for long‐term liver tumor follow‐up by Jaehee Chun, McKell Woodland, Austin Castelo, Caleb O'Connor, Mais Altaie, Aashish Gupta, Shanli Ding, Allyson Nguyen, Eugene J Koay, Kristy K Brock (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-13578125.txt.
Description
Jaehee Chun and co-authors build a fully automated way to outline liver tumors on follow-up scans. Tailored outlines beat general models, but their strength across long gaps in time is unclear. The new method adds facts from radiology reports to the scan data.
The team trained a U-Net style model on paired scans and reports from many patients. For follow-up, the model sums up each report and turns it into language codes. Those codes merge with image features to guide the outline. Personal tuning used the first follow-up scan, months after the first scan. The study tests if report context keeps outlines accurate over long-term follow-up.
Catalog record
| CREATOR | Jaehee Chun, McKell Woodland, Austin Castelo, Caleb O'Connor, Mais Altaie, Aashish Gupta, Shanli Ding, Allyson Nguyen, Eugene J Koay, Kristy K Brock |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |