Related Experiment Video
Updated: Jul 9, 2025
05:12
Robotized Testing of Camera Positions to Determine Ideal Configuration for Stereo 3D Visualization of Open-Heart Surgery
Published on: August 12, 2021
2.0K
Analyzing Surgical Technique in Diverse Open Surgical Videos With Multitask Machine Learning
Emmett D Goodman1,2, Krishna K Patel1,2, Yilun Zhang3
1Department of Computer Science, Stanford University, Stanford, California.
JAMA Surgery
|December 6, 2023
Summary
This study developed an AI model using the largest annotated open surgery video dataset to analyze surgical behaviors and surgeon skill in real-time. The AI accurately distinguished experienced surgeons from trainees based on hand motion kinematics.
Area of Science:
- Surgical innovation
- Artificial intelligence in medicine
- Surgical education
Background:
- Open surgery limitations necessitate advancements in surgical training and assessment.
- Current artificial intelligence (AI) models lack generalizability due to limited annotated video data.
- Developing AI for real-time surgical behavior analysis requires large, diverse datasets.
Purpose of the Study:
- To curate the largest annotated video dataset of open surgery procedures.
- To develop a generalizable, multitask AI model for real-time analysis of surgical behaviors.
- To leverage AI for assessing surgeon skill and optimizing surgical training.
Main Methods:
- Programmatically queried YouTube for open surgery videos, manually annotating them to create the Annotated Videos of Open Surgery (AVOS) dataset.
- Trained a multitask AI model on the AVOS dataset, comprising 1997 videos across 23 procedures.
- Applied the trained AI model retrospectively to prospectively collected surgical videos from a tertiary care center.
Main Results:
- Developed a multitask AI model capable of real-time understanding of surgical behaviors and skill.
- Identified a compound kinematic hand motion feature that significantly discriminates between experienced surgeons and trainees.
- Experienced surgeons showed 3.6 times higher odds of being identified with increased compound feature values.
Conclusions:
- AI-deduced insights into surgical structure and skill can be derived from analyzing open surgical videos.
- The developed AI model demonstrates potential for optimizing surgical skill acquisition.
- This approach offers a valuable tool for improving surgical care through enhanced training and assessment.
