Research On Methods Of Medical Image Denoising And Segmentation

Medical images are affected by a variety of noise,and the noise makes the image blur.Medical images corrupted by noise can affect doctors diagnosing the illness.Medical image denoising is a necessary guarantee to obtain image information,and it is the foundation of the further operation such as image segmentation,contour extraction and visualization and so on.So first we must get rid of the noise.Image segmentation and contour extraction have very wide applications in many aspects,for example,tracking the moving objects,three-dimensional reconstruction of the model,the identification of image information,etc.In medical image processing,segmentation is very important.Medical image segmentation is the basis of some subsequent operations.And it plays an extremely important role in disease diagnosis,three-dimensional reconstruction of body tissues and the surgery planning and so on.This paper mainly studies the methods of medical image denoising and segmentation.About medical image denoising,we mainly study the basic knowledge of image filtering.It includes the mathematical model of the noise,several typical noise model,noise detection mechanism and several filtering methods commonly used in medical image processing,for example,the mean filtering,median filtering,gaussian filtering and improvement methods.Adaptive median filtering method is studied in detail.We improve its shortcomings and propose an improved adaptive median filtering method,which both can filter out noise and protect the image detail better.For image segmentation problem,we mainly studies some edge detection operators commonly used in image segmentation,Snake model,GVF Snake model and GNBGVF Snake model.In view of the shortcomings and deficiencies of GNBGVF model,we add an adaptive external force and propose FGNBGVF model.This model can solve the problem of deep sag better.Finally,the proposed denoising and segmentation method are applied to the left ventricle MR image segmentation process.We call the left ventricular internal and external membrane segmentation algorithm.And it can segment the left ventricular internal and external membrane efficiently and accurately.