Application of real-time artificial intelligence to cataract surgery

Abstract

Background/aims: Artificial intelligence (AI) in Ophthalmology has yet to be applied to real-time cataract surgery. This work explores a new AI tool, developed for phacoemulsification, and evaluates its potential uses.First, our study aimed to demonstrate the use of AI in phase recognition and analysis of phacoemulsification. Second, to evaluate the application of real-time AI to live cataract surgery.

Methods: Phase I: surgical video recordings of adult patients undergoing cataract surgery at Moorfields Eye Hospital were captured. The AI, via Touch Surgery Ecosystem, was developed and used to segment surgery into phases based on the International Council of Ophthalmology-Ophthalmology Surgical Competency Assessment Rubric tool.

Phase ii: having demonstrated the AI's functionality in phase I, a further group of phacoemulsification patients was recruited into a live surgery study arm. Three auxiliary screens were deployed in the operating theatres, displaying phase detection and phase relevant information in real time.

Results: Phase I: 192 videos were analysed between March 2020 and March 2021. Average case duration for consultants (n=68), advanced trainees (n=59) and junior trainees (n=65) was 11.18, 17.54 and 21.36 min, respectively. Efficiency benchmarks were determined using the median metric values for advanced trainee and consultant cases, respectively.

Phase ii: efficiency metrics for 74 cases with screen deployment and 26 without were compared. With real-time AI, consultant surgeons had a significant decrease in case duration.

Conclusions: We demonstrate the first use of a fully independent AI platform for analysing efficiency metrics in cataract surgery. Real-time AI has the potential to improve operative efficiency and surgical team training.

Keywords: Lens and zonules; Medical Education; Ophthalmologic Surgical Procedures; Public health; Surgeons.