AR navigation system for neurosurgery

Abstract

This paper presents a navigation system for an endoscope which can be used for neurosurgery. In this system, a wire frame model of a target tumor and other significant anatomical landmarks are superimposed in real-time onto live video images taken from the endoscope. The wire frame model is generated from a CT/MRI slice images. Overlaid images are simultaneously displayed in the same monitor using the picture-in-picture function so that the surgeon can concentrate on the single monitor during the surgery. The system measures the position and orientation of the patient using specially designed non-contact sensing devices mounted on the endoscpe. Based on this real-time measurement, the system displays other useful information about the navigation as well as the rendered wire frame. The accuracy of registration between the wire frame model and the actual live view is less than 2mm. We applied this AR navigation clinically in surgical resection of pituitary tumors in six cases, and verified its performance and effectiveness.