GRID Software Solution for the Segmentation of the Coronary Artery Tree in Biplane Angiograms

We are developing a computer system that will enable the precise diagnosis and the planning of therapy in the field of coronary artery disease. The implementation of this system is carried out within the framework of a computational GRID (Austrian GRID). Our aim is to obtain patient-specific simulations of coronary hemodynamics. The geometry of the flow domain is derived from biplane angiograms. In these images, the coronary artery tree must be segmented. Thereafter, a three-dimensional reconstruction method is applied. However, biplane angiograms are subject to noise, blur, and image deterioration caused by background structures. In order to eliminate these disturbances, several image pre-processing steps and a newly developed segmentation method must be carried out. In this paper we will concentrate on our new segmentation approach and show that the segmentation of the coronary artery tree can be solved advantageously within the framework of a computational GRID.