Quantifying risky behavior in surgical simulation.

Abstract

Evaluating a trainee's performance on a simulated procedure involves determining whether a specified objective was met while avoiding certain "injurious" actions that damage vulnerable structures. However, it is also important to teach the stylistic behaviors that minimize overall risk to the patient, even though these criteria may be more difficult to explicitly specify and detect. In this paper, we address the development of metrics that evaluate the risk in a trainee's behavior while performing a simulated mastoidectomy. Specifically, we measure the trainee's ability to maintain an appropriate field of view so as to avoid drilling bone that is hidden from view, as well as to consistently apply appropriate forces and velocities. Models of the maximum safe force and velocity magnitudes as functions of distances from key vulnerable structures are learned from model procedures performed by an expert surgeon on the simulator. In addition to quantitatively scoring the trainee's performance, these metrics allow for interactive 3D visualization of the performance by distinctive coloring of regions in which excessive forces or velocities were applied or insufficient visibility was maintained, enabling the trainee to pinpoint his/her mistakes and how to correct them. Although these risky behaviors relate to a mastoidectomy simulator, the objectives of maintaining visibility and applying safe forces and velocities are common in surgery, so it may be possible to extend much of this methodology to other procedures.