Perfect registration leads to imperfect performance: a randomised trial of multimodal intraoperative image guidance

Abstract

Objective – To compare surgical safety and efficiency of two image guidance modalities, perfect augmented reality (AR) and side-by-side unregistered image guidance (IG), against a no guidance control (NG), when performing a simulated laparoscopic cholecystectomy (LC). Background – Image guidance using AR offers the potential to improve understanding of subsurface anatomy, with positive ramifications for surgical safety and efficiency. No intra-abdominal study has demonstrated any advantage for the technology. Perfect AR cannot be provided in the operative setting in a patient, however it can be generated in the simulated setting. Methods – Thirty six experienced surgeons performed a baseline LC using the LapMentor™ simulator before randomisation to one of three study arms: AR, IG or NG. Each performed three further LC. Safety and efficiency-related simulator metrics, and task workload (SURG-TLX) were collected. Results –The IG group had a shorter total instrument path length and fewer movements than NG and AR groups. Both IG and NG took a significantly shorter time than AR to complete dissection of Calot’s triangle. Use of IG and AR resulted in significantly fewer perforations and serious complications than the NG group. IG had significantly fewer perforations and serious complications than AR group. Compared to IG, AR guidance was found to be significantly more distracting. Conclusion – Side-by-side unregistered image guidance (IG) improved safety and surgical efficiency in a simulated setting when compared to AR or NG. IG provides a more tangible opportunity for integrating image guidance into existing surgical workflow as well as delivering the safety and efficiency benefits desired.