An automated recognition technique for aorta aneurysm using thoracic multi-slice CT images

摘要

An automated recognition technique was proposed for providing information on such as the skeleton of thoracic aorta, diameter of each cross-section along the skeleton, curvature of it and position of the aneurysm on the aorta using multi-slice CT images. The information was indispensable to design a stent with perfect fitting to the patient's aorta aneurysm region so that the stent kept the original position against the blood stream. The proposed technique was based on the fact that cross-sections of the aorta aneurysm region had deformation from circular shape while those of the normal region had circular shape due to blood pressure. The three-dimensional shape of the aorta was extracted by applying an algorithm Successive Region Growing. Cylinder model fitting well reduced noise and enabled us to detect the deformation with high accuracy. Skeleton and diameter of the aorta were obtained from the resultant fitting circle. The proposed technique was successfully applied to actual 90 slice of patient multi-slice CT data.