Skill assessment based on clutch use in cross-platform robot-assisted surgery

Area of Science:

• Surgical Skill Assessment
• Robotic Surgery
• Computer Vision

Background:

• Skill transfer in surgery is well-studied, but specific surgeon operational skills like clutch control in robot-assisted surgery (RAS) remain underexplored.
• Clutch control is a critical skill in RAS, necessitating objective evaluation methods.

Purpose of the Study:

• To develop a novel computer vision algorithm for automatic evaluation of surgeon clutch use in RAS.
• To assess the correlation between clutch metrics and surgical skill levels on different robotic platforms.

Main Methods:

• Trained 30 laparoscopic surgeons using robotic simulators and RAS animal experiments, collecting video and hand movement data.
• Developed a computer vision algorithm using a skin-color model and labeling to classify clutch use.
• Validated clutch metrics on multiple robotic platforms (dv-Trainer, EDGE MP1000, Toumai™ MT1000, DaVinci Xi) and across varying skill levels.

Main Results:

• On simulators, experts showed significantly higher clutch accuracy than the study group, with similar clutch counts.
• In RAS experiments, clutch counts decreased for both groups, but expert accuracy remained high while study group accuracy declined.
• The proposed metrics demonstrated significant correlation with surgical skill levels.

Conclusions:

• A novel, objective assessment method for surgical skills—clutch use metrics—was developed for cross-platform RAS.
• The validated metrics significantly correlate with surgeon proficiency in RAS, offering a tool for skill evaluation.