Related Experiment Video
Updated: Sep 18, 2025
Clinical Application of Single-Surgeon, Three-Port, Laparoscopic Resection for Colorectal Cancer with Natural Orifice Specimen Extraction
Published on: March 24, 2023
Developing a deep learning-based surgical-skill assessment model focused on instrument handling in laparoscopic
Kei Nakajima1, Shin Takenaka2, Daichi Kitaguchi2
1Department for the Promotion of Medical Device Innovation, National Cancer Center Hospital East, 6-5-1 Kashiwanoha, Kashiwa, Chiba, 277-8577, Japan; Department of Gastrointestinal Surgery, Graduate School of Medicine, Institute of Science Tokyo, 1-5-45, Yushima, Bunkyo-ku, Tokyo, 113-8510, Japan.
Automated surgical skill assessment using computer vision is feasible for evaluating tissue grasps in laparoscopic surgery. The number of grasps accurately reflects skill level, though distinguishing successful from failed grasps requires further model refinement.
Area of Science:
- Minimally Invasive Surgery
- Surgical Skill Assessment
- Computer Vision in Medicine
Background:
- Poor instrument handling is linked to reduced surgical proficiency.
- Automated assessment models can objectively evaluate surgical skills.
Purpose of the Study:
- To develop and validate an automated recognition model for tissue grasping in laparoscopic surgery.
- To determine the feasibility of automated surgical skill assessment based on tissue grasp metrics.
Main Methods:
- Intraoperative videos from high-, intermediate-, and low-skill surgical groups were analyzed.
- Tissue grasps (total and success/failure) were manually and automatically quantified.
- Computer vision models were used to automatically distinguish skill levels.
Main Results:
- Automated grasp counts strongly correlated with manual counts and were higher in the low-skill group.
- Manual analysis showed more grasps in low-skill and fewer failed grasps in high-skill groups.
- Automated models accurately differentiated skill levels based on grasp count, but not success/failure.
Conclusions:
- Automated recognition of tissue grasping is feasible for laparoscopic surgical skill assessment.
- The number of tissue grasps is a reliable indicator of surgical skill level.
- Further improvements in automated recognition accuracy are needed for distinguishing successful/failed grasps.
