Research Of Liver Image Segmentation Method Based On Multilayer Spiral CT

The combination of image processing in the field of computer technology andmedical image diagnosis technology has become a focused topic. The research in thispaper is based on Jilin province high technology industry development special project“The application of liver CT image based3D model in liver treatment”. For liver CTimages, researched the liver image segmentation method, and the region growingalgorithm and Graph Cut segmentation method were improved. The main researchcontent is as follows:(1)Research on the format conversion from CT image formats to the BMP file,and the filter algorithms, which is suitable for the transformed BMP liver image,including adaptive median filter algorithm and the Anisotropic diffusion filter.(2)Analyzed the liver intensity distribution feature extraction based on fuzzyc-means clustering. Firstly analyzes the intensity distribution characteristics of theliver images, and described the mathematical presentation, then clustering algorithmis studied in detail, and selected the fuzzy c-means clustering algorithm FCM for livergrayscale distribution parameters in this paper as statistical analysis and mathematicaltools.(3) Researched and improved the region growing based liver image segmentationalgorithm. On the basis of liver area distribution intensity statistical analysis, theimprovement of confidence connection threshold segmentation algorithm realized anadaptive segmentation of liver images without filtering, and achieved effective liverparenchyma extraction. In the simulation experiment, the proposed segmentationmethod is applied to pathological liver, without filtering, simplifying the process ofliver parenchyma segmentation, and verified that the algorithm is not sensitive to theinitial seed points, and achieved an effective segmentation results.(4)Researched and improved Graph Cut based liver image segmentation method. Researched the maximum flow minimum cut algorithm, the mapping principle ofimage to the Graph, as well as the boundary and the calculation method of energyfunction. Specific for liver image segmentation and liver image characteristics, basedon FCM clustering algorithm for the liver image intensity distribution, the statisticalresults intensity feature was used to improve the energy function of Graph Cut model.Simulation results show that, the improved Graph Cut image segmentation model canrealize effective liver image segmentation, even for the pathological livers.(5)Combined confidence region growing segmentation algorithm with GraphCut segmentation algorithm, liver sequence image segmentation method is researched.The improved Graph Cut segmentation algorithm is used for a single image, in whichliver account a large portion, the segmentation, and the corrosion result of thesegmentation is regarded as the seeds of adjacent slices before and after. Then theconfidence connection threshold segmentation algorithm is used to the segmentationof adjacent slices to realize the liver sequence image segmentation.