CT Images Of Liver Vascular Segmentation And Reconstruction

Liver is an extremely important detoxification organ in the human body. In recent years, due to the improved quality of life, heavy drinking or improper diet cause fatty liver disease more and more. If did not receive timely treatment, it’s very easy to turn to cancer. So early condition judgment is very important for liver lesions. In general the tumor is removed by surgical resection. However, the relationship between tumor and blood vessels, the proportion of tumor volume to liver volume, both determine whether treat by surgical resection or by conservative treatment. For these problems, it’s difficult to give an accurate answer through information of body section given by traditional CT images. Therefore, we need to seek a more intuitive way to reconstruct the complex vascular system in liver.In this paper, the main step of three-dimensional reconstruction is divided into three steps:First, extract blood vessels, tumors in two-dimensional CT images. Second, reconstruct by a sequence of individual images. Finally, correct on the results of the three-dimensional reconstruction.Because the presence of large noise in CT images, we use a method of denoising pretreatment before the dividing of vessels. In the segmentation of vessels, the common image segmentation methods are based on threshold, the image area and some particular theory, such as cluster analysis, partial differential equations. However, the difference between bodies is large. It causes difficulties on the automatic processing a series of images. There is no one way can generally apply to all medical CT images. In this paper OTSU algorithm, region growing algorithm and Mean Shift algorithm are adopted to segment liver CT images respectively. There is little difference among the segmentation results on a single CT image. However, in practice, every patient has a large number of CT images. So, among the three methods, there are differences in the stability, adaptation and operability. Ultimately OTSU method with multi-threshold segmentation is adopted and achieves relatively good results.In the process of three-dimensional reconstruction, this paper uses the3D reconstruction visualization tools (VTK) for rendering vessels. The surface of the three-dimensional reconstruction is not smooth. To solve this problem, the denoising method for two-dimensional CT images is expanded to three-dimensional space. Eventually this paper won the ideal three-dimensional model. Finally, the results lay a solid foundation for the further surgical navigation.