Evaluating Drilling and Suctioning Technique in a Mastoidectomy Simulator

In order to move towards the goal of enabling simulators to serve as intelligent, mostly autonomous virtual instructors of surgical skill, we have previously proposed [1, 2] a number of metrics intended to capture some of the most important aspects of good technique that a real instructor tries to teach his/her residents in the field of temporal bone surgery, using our simulator [3]. In this paper we present several new metrics related to bone removal and suctioning technique. Most existing surgical simulators, especially laparoscopic skill trainers, have attempted to incorporate a small number of simple metrics [4]. Most assume a simple global optimum value, such as minimize wall collisions, maximize path efficiency, or minimize completion time, and do not attempt to learn from runs of the simulators by experts or novices. Several have used learning algorithms such as Markov Models [5] or neural nets [6] to evaluate surgical performance.