Aortic Aneurysm Segmentation

pfadAbdominal aortic aneurysms can be treated either by open surgery or minimally invasive by implantation of a stent graft. The latter technique is relatively gentle, however complications can occur. Therefore regular postoperative examinations are necessary to ensure that there are no leakages at the implant. Those examinations are done on the basis of CTA images. The indicators for leakages are beside the apparent appearance of contrast agent in the aneurysm sack unfavorable changes in size and shape of the aneurysm. One has to take into account that the aneurysm shrinks over time when successfully separated from the blood flow. This shrinking has to be accounted for in comparing measurements.

Several methods for automatic preoperative aneurysm segmentation have already been published. However, the postoperative examinations are today (early 2008) still carried out manually. Beside the search for contrast agent inflow, a measurement of the maximum aneurysm diameter is necessary. For an accurate measurement of the diameter, it is necessary to measure exactly orthogonal to the aneurysm centerline. We therefore developed an algorithm which calculates the centerline automatically given a user defined start and end point in the aortic lumen.

The algorithm uses the 3D-CSC to segment the aortic lumen and calculates the centerline through the lumen by a shortest-path algorithm. The centerline is the basis for a curved reformation. In this image representation, the user can easily measure size and diameter of the aneurysm. Further on a straightened aorta longitudinal cut is generated in which beside the maximum diameter the general shape of the aneurysm is visible.

On the reformated slices the aneurysm is automatically segmented. This is done by first detecting possible aneurysm tissue and then approximating the aneurysm by a circle which is detected by circle detection.