Gradient Vector Flow als Metrik zur Registrierung eines Herzmodells
Bachelorarbeit von Peter Stilow
As shown in the statistics of the World Health Organization almost the half of all death's in Germany are caused by cardiovascular diseases. Because of this it is most important to identify such diseases as early as possible. In order to get fast and correct diagnosis the data received from recording methods, like X-ray computed tomography, have to be suitable visualized, which is a difficult task.
A registration process was developed in the context of a projectgroup. The registration process gives some extra information to a program. With use of the information the visualization could be simplified. This process registrates a heartmodel with a dataset completely automated. To make it both more robust and more precise a new metric, the Gradient Vector Flow, is implemented in the course of this thesis. The Gradient Vector Flow is used in the machine learning part inside the already existing process of registration. An evaluation was done, to check if the Gradient Vector Flow could improve the results of the registration process.