Automated Objective Surgical Skill Assessment and Visualization in the Operating Room Using Unstructured Tool Motion for Improved Surgical Training

Abstract

Past attempts at surgical skill assessment using tool motion in the operating room have focused on highly-structured surgical tasks such as suturing. These methods considered only generic descriptive metrics such as the operating time and the number of movements made, which are of limited instructional value. In this thesis, we develop and evaluate an automated method of surgical skill assessment of flap elevation in nasal septoplasty in the operating room. The obstructed field of view and highly unstructured nature of septoplasty hinders trainees from efficiently learning how to effectively perform the procedure. Thus, we also present the development of a real-time visualization system that allows trainees and instructors to better observe tool motion with respect to patient anatomy during the operation. In this work, we propose a descriptive structure of septoplasty that consists of the following two activity types: (1) the brushing activity directed away from the septum plane that characterizes the consistency of the surgeon’s wrist motion and (2) the activity along the septal plane that characterizes the surgeon’s coverage pattern. We computed features related to these activity types that allow classification of a surgeon’s level of training with an average accuracy of about 72%. Further, as opposed to previously-measured generic motion metrics, the presented features provide surgeons with personalized, actionable feedback regarding their tool motion.