Surgical proficiency assessment using virtual reality (VR)-based hybrid simulation for minimally invasive procedures

Technological advances in extended reality (XR: technology incorporating virtual reality (VR), augmented reality (AR), and mixed reality (MR)) have spurred the development of surgical simulators with the goal of immersing the user in the environment. However, there has been only incremental progress in developing surgical proficiency assessments that are critical in utilizing this technology to train safer and more efficient surgeons. We hypothesize that baseline measurements of expert surgeons (ground truth) and comparisons between users of different proficiency levels can be analyzed to develop an accurate proficiency classification of trainees on a VR-surgical simulator. The foundation of this work is based on our functional endoscopic sinus simulator (FESS) which incorporates a hierarchical task-based analysis of sinus surgery with motion-tracking of the endoscope and surgical instruments. We utilized dynamic time warping (DTW) to combine motion-tracking and operative time of sinus surgery experts to estimate a ground truth, developed a classification network combining motion-tracking and operative time using Decision Tree C5.0, and validated the accuracy of the proficiency classification by conducting extensive experiments with novice (n = 28) and expert (n = 24) users. The proposed work is critical to provide personalized and directed feedback to efficiently train surgeons utilizing simulation.