Segmentation Of Brain Image Based On Fuzzy Clustering

Brain disease is one of the current threats to human health. With the imaging technology, it does great help to diagnose brain diseases upon qualitative and quantitative analysis. The application of magnetic resonance imaging, with the characteristics of no intervention> not harmful、seldom effected by the motions of objection, has been widely used in taking pictures of medical images. It has become the primary means in biomedical research and clinical application such as study of anatomical structure, localization of pathology, etc. As a result, accurate segmentation method is crucial to the follow-up analysis for it can obtain quantification of tissue volumes which would help physicians to develop and modify the treatment plan.Because of fuzzy boundary between the internal organization as well as the inherent uncertainty of MR images, fuzzy clustering has been widely used in MR image segmentation, among which fuzzy c means algorithm is the most widely used. However, due to factors such as imaging mechanism, the MR images often contain both noise and bias which makes the traditional models can not obtain satisfied segmentation results. This paper wants to form a union framework for simultaneous estimation of the bias field and segmentation, embed in both the global and local image information, so it can reduce the impact of noise and bias at the same time. The primary work and remarks of this paper are as follows:(1) This paper proposes a new energy minimization framework for simultaneous estimation of the bias field and segmentation of tissues for magnetic resonance images. The bias field is modeled as a linear combination of a set of basis functions in order to keep it smoother, thus the proposed model can handle intensity inhomogeneity.(2) A novel model based on FCM which combines segmentation with bias correction is proposed, while the non-local regular item is used to reduce the impact of noise and keep the image structure at the same time. The experiments of the brain magnetic resonance images show that the proposed method can obtain better segmentation results as well as the bias estimation in an accurate way.(3) A novel model based on local entropy is proposed. We first propose a local energy function based on FCM model which combines segmentation with bias correction. As a result, the proposed model can handle intensity inhomogeneity. Then, the local entropy information is incorporated into the model which makes it more robust to noise and intensity inhomogeneity.