Sensory-Glove-Based Open Surgery Skill Evaluation

Abstract

Manual dexterity is one of the most important surgical skills, and yet there are limited instruments to evaluate this ability objectively. In this paper, we propose a system designed to track surgeons’ hand movements during simulated open surgery tasks and to evaluate their manual expertise. Eighteen participants, grouped according to their surgical experience, performed repetitions of two basic surgical tasks, namely single interrupted suture and simple running suture. Subjects’ hand movements were measured with a sensory glove equipped with flex and inertial sensors, tracking flexion/extension of hand joints, and wrist movement. The participants’ level of experience was evaluated discriminating manual performances using linear discriminant analysis, support vector machines, and artificial neural network classifiers. Artificial neural networks showed the best performance, with a median error rate of 0.61% on the classification of single interrupted sutures and of 0.57% on simple running sutures. Strategies to reduce sensory glove complexity and increase its comfort did not affect system performances substantially.