Comparative validation of machine learning algorithms for surgical workflow and skill analysis with the HeiChole benchmark
Problem
Surgical workflow and skill analysis algorithms trained on single-center data lacked multicenter validation of generalizability.
Why it matters
Reliable workflow and skill analysis could increase operation safety and improve surgeon training.
What the AI did
Created the HeiChole multicenter benchmark of 33 laparoscopic cholecystectomy videos and comparatively validated machine learning algorithms on phase, action, and skill recognition.
Results
Dataset: 33 videos, 3 centers, 22 h, 7 phases, 250 phase transitions, 5514 action and 6980 instrument occurrences; prior single-center phase recognition up to 91% average precision; this study's comparative results truncated.
Impact
Multicenter benchmark tests generalizability of surgical AI toward safer cognitive assistance and training feedback.
Record
| Venue | Medical Image Analysis |
|---|---|
| Published | 2023 |
| Work type | article |
| Authors | Martin Wagner, Beat Peter Müller-Stich, Anna Kisilenko, Duc Tran, Patrick Heger, Lars Mündermann, David M Lubotsky, Benjamin Müller et al. |
| Citations | 100 (OpenAlex, at import time) |
| References | 87 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1016/j.media.2023.102770 |
| Open access | hybrid |
| Keywords | surgical workflow, skill analysis, benchmark, machine learning, laparoscopy, multicenter |
Importance rating
real but narrower advance
Abstract states cognitive surgical assistance could increase operation safety and improve surgeon training; multicenter benchmark tests generalizability, but no scale of surgeries or patients is stated.