Real-time artificial intelligence evaluation of cataract surgery: A preliminary study on demonstration experiment

Area of Science:

• Ophthalmology
• Surgical Technology
• Artificial Intelligence

Background:

• Cataract surgery requires precise execution.
• Objective evaluation of surgical skill is crucial for training and patient safety.
• Existing methods for assessing surgical performance can be subjective or retrospective.

Purpose of the Study:

• To develop and evaluate a real-time artificial intelligence (AI) system for assessing cataract surgery.
• To compare surgical risk indicators and duration between resident surgeons and experienced supervising doctors.
• To investigate the utility of AI in intraoperative surgical risk identification.

Main Methods:

• Implemented a real-time AI-based evaluation technology for cataract surgery.
• Compared risk indicators for continuous curvilinear capsulorhexis (CCC) and phacoemulsification (Phaco) procedures.
• Analyzed data from 18 cataract surgeries (9 by residents, 9 by supervising doctors).

Main Results:

• Supervising doctors exhibited significantly lower mean risk indicator values than residents in both CCC (0.433 vs 0.556) and Phaco (0.377 vs 0.511) procedures.
• Statistical analysis confirmed significant differences in risk indicators between the two groups (P = 0.0003 for CCC, P < 0.0001 for Phaco).
• The AI system successfully collected real-time data on surgical performance.

Conclusions:

• A real-time AI surgical technique evaluation system for cataract surgery was successfully implemented.
• Experienced surgeons demonstrated superior performance based on AI-derived risk indicators.
• AI holds potential as an objective tool for intraoperative identification of surgical risks in cataract surgery.