3d Visualization Application In The Diagnosis Of Liver Disease Research

3-D Visualization of the medical images is one of the most important subjects in modern medical images processing because of the research value and the great prospect in clinical diagnosis. However, until now, most of the clinical doctors can only utilize their clinical experience to analyze the 2-D images obtained by MRI and CT etc, which might be very limited to determine the perfect spatial location and exact area of the disease. Therefore, there is a need to build a suitable PC software system, which can be applied to visualize the corresponding 3-D images of the 2-D images obtained by MRI and CT etc. Furthermore, through the necessary splitting on the 3-D images , and some useful analyses or simulations before the actual operation, the clinical doctors could have more confident and knowledge about the disease, which will eventually decrease the operation difficulties, and increase the successful ratios in some way.The present research status and some methodologies about the 3-D medical image processing are introduced in the first part of this paper, while two important contents, i.e., segmentation and rendering are more concerned. Based on the overview of the methodologies and technology about the segmentation and rendering, some application backgrounds and prospects of these methodologies are compared and evaluated.According to the detailed characteristic of liver, an automatic segmentation method and a 3-D rendering method for liver are presented in this paper. Since the liver posses many partition of the whole image, a supervised histogram estimation method can be applied to get a fuzzy segmentation of the image. Then an edge detector and a mathematic morphology method are utilized to optimize the boundary shape of the liver. A perfect boundary edge of the liver could be obtained while the organizations, which are weakly connected with the liver, are discarded. After that, a mixed rendering method is used to construct a 3-D liver image with those images obtained in above.Since the two adjacent slices have some similarity, the segmentation method is improved in our method. We dilate the liver edge of neighbor slice, and use it as the initial edge of the next slice. Therefore, the computational cost might be saved a lot. Two examples are implemented in the paper. One is based on some female sliced images taken from the Visual Human, and the other is based on some MR T2 images of a liver cancer patient. The results obtained by our automatic segmentation method are compared with those obtained by usual hand classification method.In the end of the paper, a visualization system dealing with the sliced images oflivers is implemented, which includes image pre-processing, segmentation & classification, 3-D rendering and other functions, such as 3-D rotation, color changing and incision in axial direction etc.