Liver Vessel Segmentation Algorithm Based On Improved Vessel Similarity Measure And Axis Tracking

Because the early symptoms of liver cancer are hidden, many patients are in the late stage of liver cancer when they come to receive medical examination due to the pain. At this point the best treatment is liver cutting and liver transplant which need to understand the liver internal structure. With the rapid development of computer image processing and digital imaging technology, the medical profession put forward an urgent need on the liver internal vessels segmentation and visualization technology. Because liver is prone to deformation and pathological, the liver vessel has no fixed form, and the blood vessel is very small, so the segmentation of liver blood vessels is very difficult.Based on studying and analyzing the related literature at home and abroad, with abdominal CT image sequence as the research object, this thesis puts forward a liver vessel segmentation algorithm based on improved Hessian matrix and axis tracking and make a 3D reconstruction of blood vessel. All this work can assist doctors to design a more reasonable preoperative plan.Liver vessel segmentation and three-dimensional reconstruction is composed of pre-segmentation, segmentation and smooth reconstruction three steps. In liver vascular pre-segmentation stage, the liver vessel is enhanced with the vascular similarity function which is made by the eigenvalue of Hessian matrix, and then do the coarse segmentation of liver vessels based on threshold. In liver vascular segmentation stage, we put forward the two-way region growing segmentation which is suitable for 3D vascular and corrugated type region growing segmentation algorithm which is suitable for the 2D blood vessels in order to keep segment the liver vascular connectivity. In this thesis, we solve the problem the vascular surface of algorithm of segmentation has fold problems base on the vascular axis of smooth reconstruction algorithm. What i mean is to extract vascular axis by refinement algorithm, then to finish branch and reconstruction according to the liver vascular topology. The final step is connecting the branch of blood vessels.In order to verify the validity of the algorithm, in this thesis, the actual medical CT image segmentation of blood vessels experiment was carried out. And the experimental results show that this algorithm can segment the liver blood vessels into good and smooth appearance, which is suitable for the quantitative follow-up of liver blood vessels.