Research On Segmentation And Recognition Based On Fuzzy Clustering For Brain Image

With the mutual penetration and development of clinical medicine and computer science,medical image processing can avoid the influence of subjective factors,accurately diagnose tumor type and malignancy as soon as possible,and help to develop effective treatment for patients.Therefore,segmentation and recognition based on image processing for brain image is particularly important.This paper mainly studies segmentation based on the fuzzy clustering and recognition based on SVM and related algorithms.Aiming at the problem that the fuzzy clustering algorithm is extremely sensitive to noise and artifacts in brain images,this paper proposes an improved FLICM algorithm based on neighborhood information.Based on the FLICM algorithm derived from the gradient descent method,the objective function is improved by fusing the gray relation of the pixel and the spatial information of the pixel.The experiment results show that compared with other fuzzy clustering algorithm,the improved algorithm has better segmentation result and stronger anti-noise performance.In order to improve the segmentation efficiency of FLICM algorithm,this paper proposes an improved FLICM algorithm combining SCoW.Firstly,it segments the image into the blocks with SCo W and modifies segmentation result by the threshold method for compressing data preprocessing.Then,it extracts the mean feature of each superpixel block.Finally,it clusters the superpixel blocks by the improved FLICM algorithm for achieving image segmentation.The experiment results show that the improved algorithm has better segmentation result and its segmentation efficiency is increased by 30%.Since the performance of SVM depends largely on the choice of model parameters,and the particle swarm optimization algorithm has problem of premature convergence so as to fall into the local optimum,the classification result of SVM based on particle swarm optimization is not good enough.Therefore,according to the shape and edge features of brain tumors,this paper uses the second optimal SVM training model optimized by improved particle swarm optimization with mutation mechanism and dynamic parameter settings,to achieve classification of brain tumors.The experiment results show that the proposed algorithm has faster convergence speed and higher classification accuracy.