Application And Research Of Immune Clustering Algorithm In Mri Brain Image Segmentation

At present, the technology of the image segmentation has been widely used in various disciplines and fields, medical image segmentation is one of the important applications.For the study of MRI brain image segmentation, the most important clinical value is reflected in detecting internal tissue’s lesions, variability and soft tissue’s damage by MRI scanning to obtain patient’s information. The brain image segmentation can be used to accurately and quantitatively extract pathophysiological data to meet the different needs of medical research and clinical needs, providing an important reference for clinical diagnosis.Artificial Immune Algorithm(AIA), which learns biological immune system’s simulation, memory, identify,those intelligent behavior, is a bionics algorithm. Due to AIA’s good global search capability, robustness and adaptive ability, we proposed a way of MRI brain image segmentation based on AIA.AIA has good global search capability, however, it does not use the system’s feedback information, often leads to a lot of useless redundant iterations, has weak local search and prones to prematurity. Because K-means clustering algorithm is easy to implement, and has strong local search ability, based on the above considerations, we combined AIA and K-means clustering algorithm proposing a image segmentation method based on immune clustering algorithm(ICA), using AIA to find the clustering center of K-means clustering algorithm, the purpose is to not only to overcome the K-means clustering, sensitive to initial cluster centers, and improve its robustness,also enhance the ability of local search.Most existing ICA used in the literature,immune operator has no direction leding to slow down the speed of evolution, to solve this problem, this artical introduces an adaptive immune clustering algorithm applying to the search center, depending on the size of individual fitness of population and average fitness, adjust the crossover and mutation probability values. Adaptive immune clustering algorithm is proposed to improve the crossover probability and mutation probability of ICA.These three algorithms are used for MRI brain segmentation, the simulation results show, in the segmentation of cerebral white matter and cerebrospinal fluidcompare, on the PR, AICA added 8.25 percent compared with AIA, added 3.55 percent compared with ICA, on the FNVF, AICA reduced by 5.25 percent compared with AIA, reduced 3 percent compared with ICA, on the FPVF, AICA reduced by 0.017 percent compared with AIA, reduced 0.004 percent compared with ICA, on the TPVF, AICA added 5.25 percent compared with AIA, added 3 percent compared with ICA, in time, AICA is faster than ICA of 199.4s, but its running speed is slower than AIA’s.To sum up, in the MRI brain’s segmentation, the precision and accuracy of AICA’s segmentation is higher than that of ICA and AIA, and the speed is also raised, compared with ICA.