Research On Segmentation Method Of MRI Brain Image Based On Fuzzy Clustering

Brain diseases are one of the major diseases that threaten the health of human beings.Using brain imaging technique to analyze brain tissue qualitatively and quantitatively is of great help for the effective diagnosis of brain diseases.Magnetic resonance imaging(Magnetic Resonance Imaging MRI)technology has the characteristics of non invasive and noninvasive,affected by the small object,has become an important auxiliary means of medical clinical diagnosis of brain diseases.In the middle and high level medical image analysis,it is necessary to segment the image first,then obtain the quantitative analysis results about the lesion area,which provides the basis for the formulation and modification of the follow-up medical program.But in clinical application,because brain MRI images in the brain tissue of the partial volume effect,intensity inhomogeneity,noise and low contrast effects to brain MRI image accurately to bring great difficulties.Therefore,according to the brain MRI precise image segmentation,fuzzy C clustering algorithm based on,from the two aspects of the improvement of the objective function and the use of spatial information and local information,in-depth study of the brain MRI image segmentation algorithm based on fuzzy clustering algorithm and its extension.The main work and achievements of this paper are as follows:(1)Proposed a method of image segmentation based on the two objectives of fuzzy C means clustering(TLTS-FCM),the two level target constraint objective function is introduced based on FCM clustering algorithm,in the two level objective function under the constraint of the objective function reaches a minimum value,control of neighborhood pixels,reasonable use spatial information.(2)The paper proposes a segmentation method of fuzzy clustering algorithm based on anisotropic filtering based on kernel function of the image(KAF-FCM),anisotropic for each pixel of the filter to replace some constrained weighted operation neighborhood information data based on TLTS-FCM clustering algorithm,and the filtered image data is applied to the objective function of FCM in.The algorithm also replaces the distance formula in the standard FCM algorithm into the Gauss kernel function to make the algorithm more robust to noise.(3)A fuzzy clustering algorithm based on local information entropy optimization(LIEO-FCM Local Information Entropy Optimization FCM)is proposed.To realize the optimization of segmentation by using the information entropy theory,the algorithm can keep the locality at the same time,find the area affected by the minimum zone based on displacement field,improve the algorithm accuracy and improve the computational time.Two segmentation algorithms,with each cluster region as the center of dynamic search,search window and a plurality of corresponding segmentation,all the segmentation results with the original clustering region for the first time the segmentation results are compared,the two division of the pixel error results first segmentation segmentation,to further improve the accuracy of the algorithm.