An Improved Method Of Medicai Image Volume Rendering

Medical image visualization consists of a series of theories,methods and techniques,applying computer graphic theory,image processing technique in modern medicine,which change two-dimensional CT images produced by a set of procession and conversion from image acquisition devices to three-dimensional image intuitivily on screen,and it become the focus of the visualization of medical image study.It play an important role in medical diagnosis, surgical planning, remedial teaching etc,which has a very high value.This paper,mostly studies the technique of medical image three-dimensional reconstruction. Analyze their domestic and overseas development prospect, meanwhile introduce the concepts of volume rendering, which include medical image volume data,several ways of image preprocessing,and typical algorithms of it. For the feature of direct volume Rendering,we study speed technology from Software and Hardware.Directed towards the feature of medical image, we study the method of region-based and edge-based image segmentation, we found that the region-based segmentation could cause over-segmentation easily,but the edge-based segmentation inhibit noise weakly and could cause hollow inside the objects,so we find a new segmentation which could increase the impact of segmentation by combining those two methods.On the base of consulting many documents, the ray-casting algorithm were analyzed and researched. A ray casting algorithm based on BVH (hierarchical bounding volumes) is presented. Introduce BVH (hierarchical bounding volumes) to reduc the amount of the light. By traversing BVH tree to jumped empty voxel,and the alogorithm pretrate the date of the original medical image.Therefore,the image quality is improved,and the speed of rebuild is quickened.Finally, we implement our experiment on 3DMed open platform which provides open source code and comprehensive data interface solutions to enable researchers to test their algorithm by loading plugins dynamically. We use VC++ 6.0 to develop our plugins. The three-dimensional reconstruction result on 156 pieces real chest CT images shows that our method acquired better rendering effect and enhanced rendering speed through comparison of traditional ray-casting algorithm. The experiment verifies the effectiveness of our method.