Surgical task and skill classification from eye tracking and tool motion in minimally invasive surgery

Area of Science:

• Minimally Invasive Surgery
• Surgical Skill Assessment
• Medical Technology

Background:

• Clinical risks in minimally invasive surgery are linked to surgeon's technical skill and anatomical knowledge.
• Quantitative assessment of surgical skills can mitigate procedural errors and enhance patient safety.

Purpose of the Study:

• To develop and validate methods for assessing surgical skill level and identifying task types in sinus surgery.
• To investigate the utility of surgical tool motion data and eye-gaze location for skill assessment.

Main Methods:

• Two methods were developed using Hidden Markov Models (HMMs) to analyze surgical tool motion and eye-gaze data.
• 14 HMMs were generated for seven surgical tasks across expert and novice skill levels.
• A dataset comprising 95 expert and 139 novice cadaveric sinus surgery trials was utilized with k-fold cross-validation.

Main Results:

• Eye-gaze data was found to contain significant skill-related information.
• Integrating eye-gaze data with surgical tool motion data improved skill assessment accuracy by 13.2% for experts and 5.3% for novices.
• The system achieved 82.5% accuracy for skill level quantification and 77.8% for surgical task type identification.

Conclusions:

• Eye-gaze tracking is a valuable addition to motion analysis for objective surgical skill evaluation.
• The proposed system offers a robust approach for quantifying surgeon proficiency and task recognition in sinus surgery.