Segmentation of Abdominal Aortic Aneurysm (AAA) Based on Topology Prior Model

In this paper, we propose a statistical based method using a topology prior model, integrating both intensity and shape information, to segment abdominal aortic aneurysm (AAA) from computed tomography angiography (CTA) scans. The method was tested on a total of 48 slices taken from 6 different patients and has shown competitive performance compared with the best reported results in the literature. Our method has achieved a mean Dice coefficient of 0.9303(+/- 0.0499), and mean Hausdorff distance of 3.5703(+/- 3.1941) mm. This method overcomes the major problem faced by currently existing solutions of similar Hounsfield values of neighboring tissues to that of the AAA thrombus. This is a promising medical tool which can be used to analyze the AAA in order to generate an accurate rupture risk indicator.