Analysis And Research Of Tubular Structure Based On Digital CT Image

In recent years, the digital image processing technology has been widely applied in the field of medical imaging, with the rapid development of the computer in the direction of the intelligential, medical images provide the image information has become an important basis for the diagnosis and treatment of disease. However, the diagnosis of the disease is correct or not, it depends on the two aspects of the image acquisition and interpretation. Medical imaging equipment using computed tomography(CT) scanning technology has reached the image digitization, the density distribution of the different organizations by recording the human body, until to produce a series of planar images of layer by layer, the data set three-dimensional reaction inside human body anatomy. In terms of image interpretation, medical diagnosis and treatment are now mainly rely on the rich experience of imaging specialists; however diverse pathological features and overlapping anatomical structure enable the data information of the image becomes complicated, huge amounts of image data and repeated workload of diagnosis volume could easily lead to omissions and misdiagnosis of the disease. Thus, it is very important to use a fast, accurate and intelligent image analysis method.High resolution images of the blood vessels in the surgical planning and medical diagnosis play an important role, the blood vessels can provide some critical information, such as vascular landmark location function, because the vascular system is not consistent with the complexity and size of the structure, it is difficult to repair once vascular injury. The distribution of blood vessels in three-dimensional structure is like a tree, some of which are linear form. This feature determines the specificity of the extraction algorithm. Accurate and rapid extraction of blood vessels is the key to computer aided diagnosis, and it is also the basis of the subsequent processing operations. So the theory of algorithms for blood vessels segmentation and edge detection has been a hot topic in the field of medical graphics.On the basis of vascular tree segmentation, extract medial axis line, the original structure of the 3D model is represented by a connected, highly accurate single pixel set. The skeleton information of medical image has important clinical significance in image registration, and the spatial position of the diseased body with respect to the skeleton can be express the severity of the case clearly. At present, the researchers have put forward a variety of thinning algorithm about skeleton, mainly based on distance, topology and energy model in three category, each one of them has different characteristics. The situation of disease assessment directly related to the degree of skeleton feature matching.Three dimensional quantitative assessments are to detect abnormal tissue area by computer, and to evaluate this area, and compared the clinical data, and outputted in mathematical descriptive expression. The goal is to avoid human-man error, reduce the difference between different doctors in the observation, to improve the level of treatment for clinical diagnosis.In this paper, I proposed my own opinion and solution to the problems mentioned above. Firstly, the medical image format is explained in detail in this paper, and gave a demonstration of DICOM display principle and image data access rule, and also analyzed the format conversion method. To explain the algorithm principle according to the classic Live-wire edge detection algorithm, summarized its characteristics and gave improvement suggestions, programmed algorithm. Then study the Hessian matrix model, analyzed level set algorithm's advantages and disadvantages, then a new blood vessel fast extraction algorithm based on level set threshold is proposed, and the efficiency and accuracy of the algorithm is implemented in the software platform. Finally, focuses on the analysis of topological thinning algorithm, proposed a thinning method based on adjacency link of the graph model, compared different thinning methods, show the workable of the algorithm; Based on the existing space coordinate transformation formula and the open source visual library code, an integrated quantitative method is proposed, and it is verified in the medical software system which is written by ourselves.