Evaluation of a prototype wireless head-mounted display system in knee arthroscopy—a randomised cross-over study

Abstract

Introduction: Our wireless full-HD augmented reality head-mounted display (AR-HMD) aimed to eliminate surgeon head-turning and reduce theatre clutter. Learning and performance using AR-HMD versus TV monitors (TVM) was evaluated. Methods: 19 Surgeons and 19 novices were randomised into groups A and B and tasked to retrieve four loose-bodies from a bench-top knee arthroscopy simulator. They each performed 5 trials using (1) AR-HMD and (2) TVM following a specified sequence (Group A: 1-2-1-1-1 and Group B: 2-1-2-2-2). Trials 1-3 were cross-over sequences to evaluate “unfamiliar” performance with the user switching between devices. Trials 4-5 were “familiarised” performances using the same device. The outcomes are time-to-completion and the incidence of loose-body drops. Results: In the first 114 “unfamiliar” simulated surgeries, users had 67% longer mean time-to-completion using AR-HMD than TVM users (195s vs 117s, log-rank p<0.001) and same number of loose-body drops (57 vs 57). In the following 76 “familiarised” surgeries, mean time-to-completion was 47% longer for AR-HMD users (134s vs 91s, p=0.036) with equal number of drops (38 vs 38). With AR-HMD, surgeons were 32% quicker than novices in the “unfamiliar” phase (158s vs 233s, p=0.025) but they performed similarly in the “familiarised” phase (139s vs 129s, p=0.874). When familiarised, the 95% CI inferiority margin was 4.4% for the AR-HMD. Discussion and Conclusion: In terms of speed, AR-HMD users cannot beat conventional TVM users. The difference is smaller when familiarised. Novices improved faster than surgeons with the AR-HMD. We will further optimise and evaluate the system.