Level Set Based Automatic Segmentation of Human Aorta
S. Y. Yeo, IGOR A. SAZONOV, Xianghua Xie, Perumal Nithiarasu · 2009
Segmentation of aorta and other blood vessels from standard 3D CT or MRI scans needs a lot of hand work if to do it by a standard segmentation software like Mimimcs and Amira. In this paper, we present a new level set based deformable model for the segmentation of human aorta from 3D image dataset. Accurate 3D geometrical models are essential for realistic computational fluid analysis of the blood flow in human aortas, which can improve our understanding of flow-related aortic diseases. Segmentation of the human aorta is however difficult, due to its complex topology and intensity inhomogeneity in the image structures. The proposed method uses a hypothesized interaction force between the geometries of the deformable surface and image objects which can greatly improve the performance of the deformable model in extracting complex geometries, deep boundary concavities, and in handling weak image edges. The results show that the new deformable model can be used to efficiently segment complex structures such as the human aorta from medical images.