The Design And Implementation Of Liver3D Visualization System Based On GPU

3D visualization technology on medical images is an important study of images processing and analysis. It can show3D shapes of organs and help doctors to observe and analyze the lesion. It would improve the accuracy and science of medical diagnosis, and facilitate the development of optimal treatment and surgical planning. It plays a significant role in the medical research.The liver is the largest gland in the human body and largest substantive organ, its complex structure and a high incidence. Liver cancer is one of the most common diseases, known as the "cancer of the King." In order to simulate the distribution of Liver tissue and learn figures of liver tumors, the model of the patient’s3D liver needs to be generated. So the study of liver3D visualization can help doctors make right hyperthermia treatment planning, and provide reliable evidence for patients to undergo timely and exact surgery.3D visualization technology includes images segmentation and3D reconstruction. Image segmentation is the necessary preparation for3D reconstruction, image segmentation directly affects the application of3D reconstruction in the medical fields. This paper first analyzes the image segmentation algorithm, and compared the advantages and disadvantages of these algorithms. The paper presents an semi-automatic segmentation algorithm, this method first preprocess the image using watershed algrithm,then based on the classical graph cut algorithm and to improve it, you can quickly segment the liver tumor and normal liver tissue and other target objects. In addition, in order to solve the problem of low efficiency of the interactive segmentation,the paper also presents a fully automatic liver segmentation. According to the gray attenuation properties of liver region in the CT image, and the liver topological structure in abdominal CT, the method combined CT format conversion technologies and image processing algorithms such as regional growth and edge detection. The liver segmentation method also uses a three-dimensional restoration techniques, refine the two-dimensional segmented liver rough sequence.In the aspect of3D reconstruction, the classic volume rendering algorithms:ray-casting algorithm for the key in this paper, the algorithm is detailed and specific application of the principle, and based on parallel programming principles of programmable graphics hardware and the existing algrithms implement a GPU volume rendering algorithm based on CUDA mode.Based on the study of those theories about3D reconstruction of medical images, the paper designed a practical software for3D liver reconstruction based on the characters of abdomen slice serials. The proposed sofiware comprised of image pre-processing,3D liver reconstruction and image post—processing. It has been proved that the software cart achieve the full process of3D reconstruction based on slice image serials, and the object has obvious3D effect and can be operated interactively. So the software can meet the needs of hyperthermia treatment planning.