Research On Magnetic Resonance Imaging Brain Images Segmentation Based On Fuzzy Clustering Algorithm

With the development of technology, the use of MRI(Magnetic Resonance Imaging), X-CT(X-ray computed tomography), PET(Positron emission tomography) and SPECT(Single photon emission computed tomography) and other medical imaging technologies are becoming more and more wider. Because of the limited time, energy of doctors and the existence of subjective differences, computer-aided diagnosis systems play an important role in the clinic. Medical imaging techniques can provide a lot of useful information for the medical staff in the absence of trauma cases.Otherwise,it can be used frequently and plays a very important role in disease diagnosis, disease monitoring, results of operations and other aspects of feedback. MRI have advantages which are no radiation injury, high resolution, multi-parameter imaging at any angle, etc. As a basis of other medical image processing means such as three-dimensional reconstruction, accurate segmentation of brain MRI images have a great significance for the development of medical image processing and medical technology. However, in the progress of practical applications, there are some defects such as noise, bias field, partial volume effects and low contrast in the brain MRI images. Accurate segmentation of brain MRI images has become a difficult problem.This paper studies the issue for the segmentation of MRI images. Because of the importance of brain MRI image segmentation, domestic and foreign researchers on MRI image segmentation problems raised many effective methods such as active contour models, markov random field, atlas, fuzzy clustering, and so on. Fuzzy C-means algorithm (FCM) is a method which commonly used by scholars. But the traditional FCM algorithm is based on the gray scale information, without considering the spatial information of the image. High noise or high bias field will result in poor segmentation results. Bias field correction fuzzy C-means algorithm (BCFCM) based on FCM adds its estimate of bias field and spatial information, can be a good solution to affect the image of the split caused by the bias field. But BCFCM algorithm does not consider the impact from noise to the estimation of bias field.Aiming MRI brain tissue segmentation, we propose a fast segmentation method to remove the skull and its appendages in the image pre-processing. Furthermore, the algorithm for the presence of defects in BCFCM major improvements proposed two solutions. An improved algorithm is proposed based BCFCM. The improvements for the BCFCM algorithm can automatically change the size of window in the objective function by estimating the noise level in the iterative processing. The experimental results show that the improved scheme greatly enhances the ability of BCFCM algorithm to resist noise. Besides, considering the estimation error of bias field in BCFCM, we proposed an algorithm. The Gaussian kernel in the object function is utilized to smooth the intensity bias field and limits the estimation value of intensity bias field by using an experimental threshold which can effectively avoid the incorrect estimation of intensity bias field in the segmentation results. The experimental results show that the proposed algorithm can not only effectively and accurately segment the brain tissues, but also deal with high level noise and intensity bias field.