Based On Hybrid Genetic Algorithm For Mri Segmentation

Magnetic resonance imaging (MRI) is a medical imaging technique with high differentiation and noninvasive way. It can quantitatively provide rich information about human anatomy in two or three dimensions. Along with the development of the clinic medicine, the framework of the brain is applied extensively.In general, white matter, gray matter, and cerebral spinal fluid are three basic tissues in the brain. Brain tissue segmentation of magnetic resonance (MR) images means to specify the tissue type for each pixel or voxel in a 2D or 3D data set, respectively, on the basis of information available from both MR images and the prior knowledge of the brain (for notational simplicity, we use pixel for both 2D and 3D data).Because of both the hardware imperfections of the MRI devices and the connatural fuzzy framework of the human brain, it is difficult to segment the MR image according to the three brain tissues strictly. The genetic algorithm has been fleetly developed in recent years. Thus, the researches, which make use of both genetic algorithm and the methods to improve the segmentation in MRI have been greatly attempted and they have become more and more important.Therefore, two improved genetic algorithm is used to deal with the segmentation of the MR image.First, we analyze the characteristic of the fuzzy theory and the fuzzy clustering theory, and the virtue and the defect of the application in the image segmentation. Next, based on the fuzzy c-means cluster, a new and improved genetic algorithm is introduced, which form the genetic fuzzy c-means clustering algorithm and avoid the problem that FCMA is dependent on the choice of the initial distribution of the cluster center and consequently algorithm ends up in a local optimum. In this paper,by applying genetic algorithm,we can achieve global optimum and have applied the algorithm to MRI segmentation, which can divide white matter, gray matter, and cerebral spinal fluid effectively .The proposed algorithm provides the potential help to the study of the function and the framework of the brain.Firstly, we analyse the theory of the MRF theory and the SA .Secondly ,the method based on MRF takes the parameter of Gibbs as a representation of the conjunction of different pixels, so it is not affected by the noises. The paper provided a new method named Markov, Simulated Annealing and Genetic Algorithm (MSGA), which are different from some traditional methods, such as Simulated Annealing and Genetic Algorithm. The method designs a hybrid GA and SA based on MRF for MR image segmentation through integrating MRF not affected by the noises, the global search of the GA and the local search of SA. A series of experiments proves this algorithm can provide better results than SA and GA in the conditions of same parameters and computing time.