Fast 2D-3D Registration for Navigation System of Surgical Robot

This paper presents a new registration algorithm of 2D color images and 3D geometric models for navigation system of surgical robot. A 2D-3D registration procedure is used to superimpose a tumor model on an endoscopic image precisely, and is therefore indispensable for the surgical navigation system. And thus, the performance of the 2D-3D registration procedure influences directly the usability of the surgical robot operating system. One of the typical techniques developed so far is the use of external markers, but the accuracy of this method is getting worse due to the breath, the heart beat, or other unknown factors. For precise registration of 3D models and 2D images without external markers or special measurement devices, the new registration method is proposed which utilizes the 2D images and their distance maps created by the Fast Marching Method. We show some results of fundamental experiments using simulated models and actual images of the endoscopic operation.