A Novel Distance for Automated Surgical Skill Evaluation

Abstract

Objective evaluation of a surgeon's skill level is a crucial step toward automatic surgical training. If the surgical activity is captured using a set of sensors, then the problem becomes a task to define an evaluation framework for motion analysis and comparison. In this paper, we propose an evaluation framework based on a novel surgery skill distance, PDTW. that consists of two main components: Dynamic Time Warping (DTW) and Procrustes analysis (PA). The DTW method aligns two time series with different lengths by contracting/dilating both signals such that their lengths become equal. The Procrustes analysis, that include reflection, scaling, and translation, can then be used as a distance measure between two aligned sequences. We evaluate our framework on two surgical datasets, one simulated and another one produced by robot-assisted minimally invasive surgery (RMIS). Our results show significant assessment improvements of PDTW over the traditional distance measures in automatically classifying expert, intermediate, and novice surgeons on different tasks.