Surgical Performance Analysis and Classification Based on Video Annotation of Laparoscopic Tasks.

Area of Science:

• Surgical Education
• Medical Simulation
• Biomedical Engineering

Background:

• Current surgical skills assessment methods (VR simulators, motion sensors, checklists) can be complex to interpret.
• An alternative, simpler methodology for skills assessment is needed.

Purpose of the Study:

• To propose and evaluate a novel methodology for surgical skills assessment and classification based on video annotation of laparoscopic tasks.
• To compare the performance metrics of surgical trainees using this video annotation technique.

Main Methods:

• Two groups of surgical trainees (students and residents) performed peg transfer and knot tying tasks.
• Video recordings were analyzed using software to annotate surgical gestures (surgemes), their duration/counts, and transitions.
• A nearest neighbor approach was used for skill level classification, with accuracy, sensitivity, and specificity as assessment metrics.

Main Results:

• Significant performance differences were observed between students and residents in both tasks, with residents demonstrating greater efficiency.
• Video-derived metrics accurately classified trainee skill levels, achieving 0.71-0.86 accuracy, 0.80-1.00 sensitivity, and 0.60-0.80 specificity.
• Residents completed tasks with fewer, shorter surgemes and fewer penalty events compared to students.

Conclusions:

• Video annotation of surgical gestures provides an intuitive and effective tool for assessing and classifying surgical trainee skills.
• This method offers a valuable alternative to existing complex assessment techniques.
• The technique allows for detailed description of surgical performance, aiding in training and evaluation.