Measuring surgical knot tying with 3D vision and VR gloves

Abstract

One handed surgical knot tying is one of the core skills of surgeons. The development of this skill takes time and one-on-one expert supervision. Our motivation is to provide good quality automatic evaluation in order to reduce the need for expert supervision and provide continuous evaluation that can support a deliberate practice training paradigm. Our work demonstrates the recognition of knot tying gestures a combination of 3D vision tracking and finger motion analysis from a pair of Manus VR gloves. We demonstrate segmentation and classification of the knot tying into its component knot types and estimate student skill level based on motion analysis.