Graph-based tracking method for aortic thrombus segmentation

Abstract

In this paper, a method for segmenting the epithelial (outer) layer of aortic aneurysms in CT-angiography acquisitions is presented. The method starts by computing the lumen centerline between two user-defined seed points which serve as bounds (proximal and distal) to the extent of the lesion. This centerline determines the Multi-Planar Reconstruction (MPR) planes which display the vessel cross-sections. Then, the user manually segments the epithelial layer on the first proximal MPR plane to form a contour. From this point, a loop is executed: First, the contour is used to set up a restricted search region in the adjacent distal MPR plane along the centerline. The voxels of the adjacent MPR plane within this search region and the voxels on which the contour itself sits are used to create a directed graph. Then, the mincut of the graph is calculated, creating an optimal segmentation of the Aortic Aneurysm border on the adjacent MPR plane. This segmented contour serves as a starting point for the loop, which is then executed again. © 2009 Springer Berlin Heidelberg.