Medical tool tracking in fluoroscopic interventions: new insights in detection and tracking of tubular tools.

These days, fluoroscopic imaging is the modality used most widely to guide physicians during minimally invasive abdominal interventions. They involve transarterial chemoembolization for hepatocellular carcinoma, the placement of transjugular intrahepatic portosystemic shunts or needle aspiration of solitary lung nodules, to name a few. These kinds of interventions share the task of navigating a surgical tool to a specific anatomical site. Difficulties for the physicians arise from the fact that fluoroscopy, a two-dimensional modality, is used to support an actually three-dimensional navigation task. Thus, enhanced navigation procedures incorporating three-dimensional data are desired for fluoroscopic interventions. Not only should they improve the navigation but also offer the potential to reduce treatment time and radiation exposure for the patient as well as the physician. Due to the complexity of such a system and despite previous efforts, an integrated solution does not yet exist. Localization of medical tools and the estimation of their motion are core components of such a navigation system and this work focuses on related methods to improve the detection and tracking during fluoroscopic interventions. To this end, different image enhancement algorithms required for the tool localization are reviewed and analyzed in a unified framework. Based on the analysis a fast and robust localization scheme is developed. Furthermore, a novel method for tracking the tools in a non-rigid environment is introduced. The tracking algorithm is based on discrete optimization and thus it is derivative free and can deal with complex models. It is fast and robust and hence complying with application specific constraints. Moreover, the presented tracking method is generic and can be extended to closed curves, which makes it applicable to other problems such as segmentation. Based on the results, a novel respiratory motion estimation procedure is developed. It is the basis for the creation of a navigation system and is essential in order to establish the relation between the location of medical tools and three-dimensional data used for the actual navigation.