Classification And Segmentation Of MRI In Brain Tissue

Modern society,the pace of life is faster and faster,the material and cultural enrichment brings a lot of convenience,but a lot of diseases,especially tumor diseases are more and more frequent.Of course,the progress of society also brings about the progress of science and technology.The research and treatment of tumor diagnosis also promotes the development of computer vision field.Medical image do not need to encroach on the body tissue,can clearly show the related lesions,compared with anatomy,it has an irreplaceable advantage.Magnetic Resonance Imaging(MRI)is one of the most widely used techniques in medical image,so there are a lot of researches on it.The segmentation of images is one of the most important steps in the analysis and diagnosis of MRI images.FCM is a widely used segmentation algorithm in image segmentation,But it is sensitive to noise,which has affected its application,so a FCM algorithm based on kernel has been proposed,ARKFCM(Adaptively Regularized Kernel-Based FCM)is one of the best methord.In recent years,ARKFCM is an excellent brain segmentation algorithm.Compared with other brain segmentation algorithm in segmentation accuracy rate,it has obvious advantages.But in the face of the MRI image non uniform intensity,its accuracy is somewhat unsatisfactory.In view of this,this paper proposes an improved algorithm based on ARKFCM,the improved algorithm can effectively improve the segmentation accuracy of MRI images.In the subsequent experiments,we first verify the negative impact of image on the segmentation accuracy of ARKFCM algorithm.After that,it can effectively improve the segmentation accuracy of the MRI image by preprocessing the uneven intensity.Finally,it has been proved that the improvement is successful by comparing with the brain segmentation algorithm in recent years.Subsequently,the classification of brain tumors was studied.Because of the different biological characteristics,benign tumors and malignant tumors are significantly different in appearance,so it is feasible to distinguish the types of brain tumors by MRI images.This paper uses the improved algorithm to extract part of clinical tumor in MRI images,then remove the tumor and irrelevant parts by morphological algorithm,then the processed image is converted to grayscale co-occurrence matrix,and matrix eigenvalue extraction.Finally use the SVM(Support Vector Machine)classifier,classification is performed by the eigenvalues of the tumor MRI image.And the classification accuracy and superiority of the algorithm is verified by experiment.