ProtoFlow: interpretable and robust surgical workflow modeling with learned dynamic scene graph prototypes
Felix Holm, Ghazal Ghazaei, Nassir Navab · 2026
Read ProtoFlow: interpretable and robust surgical workflow modeling with learned dynamic scene graph prototypes by Felix Holm, Ghazal Ghazaei, Nassir Navab (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-13553708.txt.
Description
Felix Holm and colleagues introduce ProtoFlow, a method that models surgical workflows in a way surgeons can inspect and trust, publishing it in 2026. Detailed workflow recognition could power AI help in theatre, but labelled surgical video is scarce and most models hide their reasoning. ProtoFlow learns dynamic scene graphs, structured maps of tools, tissues, and their changing relations, then distils them into prototypes that stand for recurring patterns of interaction.
Tested on a fine-grained surgical scene-graph dataset, the method beat standard graph baselines and stayed strong when trained on very little data, even a single video. Its prototypes also picked out distinct sub-techniques and flagged workflow departures and rare complications in plain view. The work points toward AI support that explains itself in training, live guidance, and review.
Catalog record
| CREATOR | Felix Holm, Ghazal Ghazaei, Nassir Navab |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |