Analytical centerline extraction and surface fitting using CT scans for aortic aneurysm repair.

Master's thesis, Cornell University, May 2005.

Endovascular aortic aneurysm repair promises to provide great advantages over traditional open surgery. Though this nascent form of treatment has already been demonstrated to be safer and easier for the patient, there is a strong need for technological advances which can improve precision and simultaneously reduce the amount of human effort involved in planning surgery. Computational analysis is also essential in predicting the future behavior of endovascular stent grafts, because at this stage there is very little empirical knowledge of long-term stent dynamics.

We present a geometric analysis procedure which aims to address this need. Our goal is to begin with CT scan data collected from aortic aneurysm patients and to produce a fully analytical model of the patient's arterial geometry with minimal user interaction. The product of this highly automated process can be a powerful tool for surgery planning and stent design. Furthermore, it provides a basis upon which computational fluid dynamics and mechanical behavior simulations can be built. The analytical nature of this representation allows for direct mathematical analysis or for arbitrarily precise discretization.

Our geometric analysis begins with a voxel segmentation. The segmented data set is then used as input into a unique robust centerline extraction procedure. This centerline then serves as the basis for an analytical fitting procedure which produces a lofted B-spline surface model. After describing this analysis process in detail, we present results from three input datasets.

This paper is available as a PDF file Goe05.pdf ( 22M).