Research On 3D Segmentation Method Based On Brain MRI

The three-dimensional treatment of medical brain images is an active research topic in computer vision.It provides meaningful brain tissue information for repair surgery,radiotherapy,stereotactic neurosurgery and other medical fields.The segmentation of the various parts of the brain is of vital importance for quantitative analysis,which is a key factor in assessing the progression of central nervous system disease and greatly enhances the ability to diagnose and monitor the evolution of the disease.Magnetic Resonance Imaging(MRI)technology is currently widely used in the diagnosis of human brain disease and other medical fields,because of its clear imaging,multi-parameter imaging,non-invasive and safe for human and other advantages,more and more people pay attention to and study.However,MRI images are susceptible to factors such as noise,bias field,partial volume effect,and so on,making the performance of computer-based automatic segmentation algorithms lacking medical prior knowledge seriously degrades.Therefore,the scientific method of brain tissue segmentation is still the focus of medical image processing research topics.This paper focuses on the three-dimensional image segmentation of human brain MRI based on kernel-based Fuzzy C-means(FCM)algorithm and Hidden Markov Models(HMMs).The HMMs model was trained by the feature vector sequence of multilayer brain slices,and the MRI images of the human brain were classified by HMMs model.Then,the improved kernel-based adaptive regularized FCM algorithm was used to segment the brain slices,while the information obtained by the classification reduced the dependence of the initial value,thus improved the accuracy of three-dimensional brain tissue segmentation.In view of the above ideas,this paper specific research work is as follows:1.The human brain slices were removed by skull tissue,denoised and other pretreatment,using the continuous HMMs model for each slice classification,and for the subsequent improvement of FCM algorithm to provide richer information.2.In this paper,an improved FCM algorithm that kernel based which adaptive regularized is proposed for the shortcomings of traditional FCM algorithm and related improved algorithm model which does not fully consider the influence of membership degree by spatial relation.The influence of the spatial information on the membership degree is adjusted by the relationship between the pixel and the membership degree of the pixel.The influence of the spatial information on the membership degree is reduced,and the influence of the algorithm on the noise is improved.Class effect.At the same time,the accuracy of segmentation of human brain slices by FCM algorithm based on kernel-based adaptive regularization is further improved by segmenting the slices of the previous frame,and the segmentation is done by slicing layers.3.The effectiveness of the improved kernel based adaptive regularized FCM method and the MRI adaptive segmentation algorithm based on HMMs is verified by several experiments.