3d Segmentation Of Ct Images And Three Dimensional Visualization,

Since the discovery of X-rays and the invention of X-ray photography by the German physicist Wilhelm Konrad Roentgen, the medical diagnosis by medical images has been developed for over one hundred years. With the rapid development of Computed Tomography (CT) imaging techniques, which can produce several hundreds or even one thousand of images in a single study today, effectively searching and grabbing the useful information from those images is now a big problem presented to the clinicians.Image segmentation is one of the oldest problems and the most difficult problems in image processing, because no unified theory and standard method exist. However, successfully extracting interesting objects from medical images, including CT images, is the foundation of most medical image applications.In this thesis, a general framework for two different methods of 3D CT image segmentation is proposed. The one introduced first is a two-level algorithm based on region-growing algorithm and active surface model. In the first level, 3D region-growing approach is applied to obtain a coarse surface of our interesting object, and in the second level, GVF active surface model is used because of its properties of smoothness and continuities. Afterward, a smooth surface of our interesting object can be extracted from the real CT image dataset. The second method is a modified 3D level set algorithm that overcomes the difficulty in extracting splitting object. Because of the fuzzy boundaries of our interesting object and its complex topological structure, classical level set method is revised for the medical image application.This thesis focuses on the liver, and once the liver volume is successfully obtained, the visualization procedure is applied. Both segmentation results of the two algorithms are visualized by the method of maximum intensity projection.The developed algorithms were tested on real CT images. The results demonstrate that the proposed procedures are feasible for liver segmentation from CT images.