The Research And Implementation About Brain MRI Image Segmentation Based On Fuzzy C-means Algorithm

Magnetic Resonance Imaging(MRI), has being widely used in clinical medicine and the demand has been steadily on the increase. What’s more, with the development of computer technology the using of medical images has been ever wider. But, there are always plenty of noise,partial volume effect and biased field in the MR images which make it a difficult work to complete image segmentation perfectly. So in order to deal with the fault in the segmentation of MR images, we propose a method which bases on the basic fuzzy c-means method. In my method, the anisotropic diffusion filter has been introduced in and the neighborhood of the element has been take into consideration. My method can do well with the noise in the MR images and also make contributions to taking away the bias field in the images. The main works of my paper are as follows:(1)In this paper, an algorithm called(PMFCM) which based on fuzzy c-means algorithm and using anisotropic diffusion for image segmentation is presented. First, in this paper the traditional fuzzy c-means and its improve the algorithms are analyzed. Second, the anisotropic diffusion has been introduced into the fuzzy c-means to take place of the compute the average of the neighborhood of the pixels which improve the speed of the algorithm using neighborhood. And because of the usage of anisotropic diffusion the algorithm is robust to noise in the MR images, and during denoising the noise the edges of the image will be retained. At last, the fast fuzzy c-means method is been introduced into my algorithm what make my algorithm even faster.(2)In this paper, the method of taking away of the bias filed in MR image is discussed and then it is introduced in my algorithm. Then my algorithm can work better in MR image segmentation and estimate the bias filed. By using my algorithm, the clear MR image can be restored.(3)In this paper, the framework of MR image segmentation is discussed and then taking my algorithm into use. First the image is filtered by anisotropic filter. Then Second-difference method is used to decision the initial clustering center and The initial membership matrix by analysis the histogram of the image been filtered. The framework can speed up the segmentation of the MR image.(4)In this paper, A image segmentation system is designed and implemented according to the algorithm proposed as the subsystem of our Medical Imaging Processing System.