Segmentation of Brain MRI Based on Modified Type-2 Fuzzy Compactness

Abstract

One of the important aspects in many algorithms for image analysis, object representation and visualization is the image thresholding that is usually applied as an initial step. Another issue in image analysis is the segmentation since is an important image-processing step by which regions of an image are classified according to the presence of relevant anatomic features. The success of image segmentation depends on object-background intensity difference, object size and noise measurement, however is unaffected by location of the object on that image. In this context, this paper proposes a new automatic thresholding method to accurately segment the brain in magnetic resonance images (MRI) on the axial plane. The proposed automatic method is based on a new algorithm that utilizes a modified type-2 fuzzy compactness algorithm, allowing it to solve the global thresholding for brain MRI in the axial plane. Associated with this method it was used has metrics the Uniformity measure (UM) and Shape measure (SM) with the purpose to analyse their effectiveness. Experiment using real-image data of MR medical images demonstrated the accuracy and robustness of the proposed method. With the purpose to analyse the capability of the proposed method, it was realized a comparison with other two approaches, namely with the Otsu and Kittler methods. The obtained results of the proposed method achieved state-of-the-art results in which UM reaches 0.998 and SM reaches 1.000.