Corornarographic Images' Centerline Extraction Based On Level Sets

In the clinical diagnosis and therapy of heart disease, the three-dimensional (3D) reconstruction of cardiovascular structures from 2D angiographies supplies doctors a tool to observe cardiovascular structures at any angle, and can be useful in the diagnosis and therapy. During the 3D reconstruction, how to extract the centerlines of the 2D Corornarographic images is one of the important steps,and also a very difficult step. In this paper, a skeleton extracting method based on level set in planar objects is presented, and also fundamental research of 3D objects'centerlines based on ahead 2D method.Three main parts of the study of this thesis are as following:1) First, we preprocess the 2D Corornarographic images: these images we gain are gray images. We should turn them into binary images. The illumination on Corornarographic images is uneven and the blood vessels are complex. We first segment the big image into small squares then give a different threshold to every part to be a binary image based on Otsu Method.2) In this system, input is the binary image, and output is the centerline of the image. During the process, we use level set algorithm. Level set method was used to solute geometric curve evolution problems. And here it is used to extract centerlines of images. There are two fast methods of level set: Narrow Band Method and Fast Matching Method.Based on FMM, we introduce an augmented fast marching method for computing the Corornarographic images'centerlines. This method is very simple and the result ispretty good.3) Then, based on ahead 2D algorithm, we try to extract centerlines of 3D objects.