Research On Medical CT Image Segmentation Method

Medical image processing as a hot field of medical research nowadays,on the basis of in a noninvasive method to get images from the body's internal organizational structure,for the convenience of the doctor through the medical CT image diagnosis and analysis of the condition of patients,reduce the doctor's work.the CT image filter processing,can get rid of the noise the image which was not clear to achieve the objective of the restored image,and implement accurate and efficient CT images segmentation,target areas appear in front of the doctor is more clearly,help patients to have early diagnosis and surgical treatment.CT image are susceptible to noise interference,and image gray uneven between organizations,Interorganizational overlap,the diversity of lesions has been a problem in the segmentation of medical CT images.With the help of the classical algorithm,this paper realizes the denoising algorithm and segmentation algorithm of medical CT images improvement,improve the denoising performance of algorithm and accuracy of segmentation algorithm,the specific research is as follows:First in this paper is the study of image processing knowledge and technology,including noise model and denoising model,image filtering algorithm and so on.selected and study the nonlocal means filtering algorithm,the algorithm measures pixel similarity only by Euclidean distance,analysis of the algorithm without considering gradient information,combined with the gradient structure similarity and kernel function,and put forward improved nonlocal means denoising algorithm.About image segmentation,study edge detection algorithm and segmentation watershed segmentation algorithm,according to the characteristics of medical image fuzziness,focus on intuitionistic fuzzy mean clustering algorithm,according to it is not considering spatial information shortcoming,markov random field is added to improve pixel membership and objective function,and proposed the MRF-IFCM algorithm,it can more accurately separate the target area from the background.Proved through the matlab medical CT experiment,the algorithm has good denoising performance,at the same time can protect detail well.And the lung nodules on CT image segmentation experiments,verify the accuracy of the improved algorithm divided,false negative and false positive are less likely.The work of this paper on the one hand for medical CT image denoising and segmentation is provide some theory possibility,on the other hand also help medical image analysis,and provides a foundation to help for medical image 3d reconstruction.