Semi-automatic catheter model generation using biplane x-ray images

Abstract

Recently, techniques for the automatic detection or tracking of surgical instruments in X-ray guided computer-assisted interventions have emerged. The purposes of these methods are to facilitate inter-modality registration, motion compensation, enhanced visualization or automatic landmark generation in augmented-reality applications. Most techniques incorporate a model of the device as prior information to evaluate results obtained from a low-level detector. In this paper, we present novel approaches which are able to generate both 2-D and 3-D models of circular and linear catheters from biplane X-ray images with only minimal user input. We apply these methods in the context of Electrophysiology to generate models of ablation and mapping catheters. An evaluation on clinical data sets yielded promising results.