Mitigating Biases in Surgical Operating Rooms with Geometry
Tony Danjun Wang, Tobias Czempiel, Christian Heiliger, Nassir Navab, Lennart Bastian · 2025
Read Mitigating Biases in Surgical Operating Rooms with Geometry by Tony Danjun Wang, Tobias Czempiel, Christian Heiliger, Nassir Navab, Lennart Bastian (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-2508-08028.txt.
Description
Tony Danjun Wang and co-authors present a medical AI paper, published in 2025, on bias in models that watch surgical operating rooms. Deep networks often cheat by learning surface cues instead of the real signal. In operating rooms, standard gowns hide body landmarks, and the authors show models fixating on shoes, eyewear, and other incidental clues to tell staff apart.
Their fix drops color images and encodes each person as a moving three dimensional point cloud, keeping body shape and motion while stripping appearance. On realistic clinical data the image models lost much accuracy while the geometry method held up. The authors argue such representations better suit future assistants that track skill and teamwork, not clothing.
Catalog record
| CREATOR | Tony Danjun Wang, Tobias Czempiel, Christian Heiliger, Nassir Navab, Lennart Bastian |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 10 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-08028