Study On The Method Of Liver Segmentation Of CT Images

Medical image segmentation is one of the host spots in the field of image analysisand digital medical research, attracted many researcher’s attention in recent years. Theliver segmentation is a branch of medical image processing, is the importantprerequisite of3d reconstruction and image analysis. The result of the liversegmentation directly affects the precision of3d reconstruction, indirectly influences theresult of the disease’s the diagnosis and treatment. So the study of the medical imagesegmentation is not only having great academic significance but also have importantclinical application value.This paper is based on the analysis of the modern medical image segmentationmethod，Combined with its own characteristics of abdominal liver CT images andstudied some relevant problems of medical image segmentation, including image formatconversion between DICM format and BMP format, comparative study some denoisingmethod about the medical image, and selected the denoising method which not only hasbest effect but also will not break the edge of the original image. The paper’s mainworks include:Firstly, we analyzed the conversion problem between the DICM format of CTimage and BMP format of general image, do some image pretreatments such asconversion the abdominal CT image format and gray transform to enhance the image,which build a good foundation for the follow-up processing precision and efficiency.Secondly, analyzed the advantages and disadvantages of several denoising method,because of the medical image is sensitive to noise, and easy to be damaged, we selectedthe anisotropic diffusion filtering method to remove noise,which not only have a gooddenoising effect but also can protect the image edge, then, we give an improved methodbased on this filter.Thirdly, studied several image segmentation algorithm，because of the medicalimage contrast is small, the boundary of the liver with other organization is not clear,vascular system is huge and quantity of data is very large,we segment the liver bycombine Graph-Based method with four binary tree. This method not only hasrealized automatic segmentation, but also improved the speed of the division.Finally, experiments are carried out with the method put forward in this paper, andthe result said that these methods can reach a good effect.