Integrating Real-time Video View with Pre-operative Models for Image-guided Renal Navigation: An in vitro Evaluation Study.

Abstract

To provide a complete picture of a scene sufficient to conduct a minimally invasive, image-guided renal intervention, real-time laparoscopic video needs to be integrated with underlying anatomy information typically available from pre- or intra-operative images. Here we present a simple and efficient hand-eye calibration method for an optically tracked camera, which only requires the acquisition of several poses of a Polaris stylus featuring 4 markers automatically localized by both the camera and the optical tracker. We evaluate the calibration using both the Polaris stylus, as well as a patient-specific 3D printed kidney phantom in terms of the number of poses acquired, as well as the depth of the imaged scene into the field of view of the camera, by projecting the several landmarks on the imaged object at known location in the 3D world onto the camera image. The RMS projection error decreases with increasing distance from the camera to the imaged object from 7 pixels at 15-18 mm, to under 2 pixels at 28-30 mm, which corresponds to a 2 mm and 1 mm error, respectively, in 3D space.