Optimization Algorithm Of Medical Segmentation Based On Fuzzy C-mean Clustering

Image segmentation is the core of brain MRI image processing,is the basis of three-dimensional reconstruction of the image,and for clinical diagnosis and adjuvant therapy to provide a strong guarantee.Nuclear magnetic resonance imaging(MRI)has the uncertainties and ambiguities caused by partial volume effect,gray scale unevenness and noise.The fuzzy C-means clustering(FCM)algorithm has better effect in segmenting such images The Aiming at the shortcomings of FCM algorithm for initial clustering center and noise sensitive,this thesis mainly does the following work:(1)Aiming at the local optimization problem of FCM initial clustering center,this thesis proposes a hybrid algorithm based on genetic and particle swarm optimization.The algorithm can increase the diversity of populations in the process of optimization by random genetic variation or particle swarm optimization.,Which effectively increases the possibility of detachment from local optimization and improves the performance of FCM segmentation algorithm.(2)Research on image segmentation based on neighborhood pixel information.The Gaussian model introduces the influence of noise on the FCM segmentation results,and uses the genetic and particle swarm optimization hybrid algorithm to automatically select the feature field information of the neighborhood pixel.FCM segmentation of the objective function of the labeling field and the characteristics of the role of field strength parameters,so that the selection of parameters more reasonable.(3)Through the histogram data of the MRI image,the part of the calculation of the intensity parameters of the neighborhood pixel and the initial value for the initial clustering center optimization are selected to improve the computational efficiency of the optimization algorithm.