ProSona: Prompt-Guided Personalization for Multi-Expert Medical Image Segmentation
Aya Elgebaly 8 Nikolaos Delopoulos 2 Juliane Hörner - Rieber 3,4 Carolin Rippke 4 Sebastian Klüter 4 Luca Boldrini 5 Lorenzo Placidi 5 Riccardo Dal Bello 6 Nicolaus Andratschke 6 Michael Baumgartl 6 Claus Belka 2,7 Christopher Kurz 2 Guillaume Landry 2,7 Shadi Albarqouni 1,9 · 2025
Read ProSona: Prompt-Guided Personalization for Multi-Expert Medical Image Segmentation by Aya Elgebaly 8 Nikolaos Delopoulos 2 Juliane Hörner - Rieber 3,4 Carolin Rippke 4 Sebastian Klüter 4 Luca Boldrini 5 Lorenzo Placidi 5 Riccardo Dal Bello 6 Nicolaus Andratschke 6 Michael Baumgartl 6 Claus Belka 2,7 Christopher Kurz 2 Guillaume Landry 2,7 Shadi Albarqouni 1,9 (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/arxiv-2511-08046.txt.
Description
Aya Elgebaly and colleagues present ProSona, a segmentation tool that adapts to individual medical experts through language prompts. Doctors often outline the same lung nodule or prostate scan differently, and averaging their styles into one mask hides real disagreement. Keeping each expert style separate usually needs a branch per annotator.
ProSona instead learns a continuous space of annotation styles on a probabilistic U-Net backbone. Text prompts steer a projection through that space to produce a personalized outline. Contrastive training aligns words and images, and tests on lung and prostate data beat a leading rival on energy distance and Dice score.
Catalog record
| CREATOR | Aya Elgebaly 8 Nikolaos Delopoulos 2 Juliane Hörner - Rieber 3,4 Carolin Rippke 4 Sebastian Klüter 4 Luca Boldrini 5 Lorenzo Placidi 5 Riccardo Dal Bello 6 Nicolaus Andratschke 6 Michael Baumgartl 6 Claus Belka 2,7 Christopher Kurz 2 Guillaume Landry 2,7 Shadi Albarqouni 1,9 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-08046