Intuitionistic Fuzzy Clustering Method with Spatial Information for MRI Image Segmentation

Fuzzy c-means is one of the popular clustering technique which has been utilized for medical image analysis. Intuitionistic fuzzy set theory based clustering is an extension of fuzzy c-means which is used for medical image segmentation due to its promising nature for handling the vagueness and uncertainty. The performance of image segmentation is not good in the presence of noise. Many fuzzy and intuitionistic fuzzy set theory based clustering methods have been reported in the literature to handle noise in the segmentation process. In the process of handling noise, most of these methods use smoothing which ignores the important structural information (such as edges and other fine details). In this research work to address this issue, the optimization problem of the proposed IFCM with spatial neighborhood information (IFCMSNI) method is formulated with a novel spatial regularization term which is based on the neighborhood membership value with the advantage of both the intuitionistic fuzzy set theory and a spatial regularization term to handle noise associated with medical images. In the proposed method, the image is represented in the form of Intuitionistic Fuzzy Sets (IFSs) using Sugeno's negation function. In order to validate the effectiveness of the proposed method, experiments have been carried out on a synthetic image dataset and two publicly available human brain MRI dataset. The segmentation performance of the proposed method is compared with FCM, IFCM, FCMS, FLICM and IIFCM methods in terms of dice score and average segmentation accuracy. The experimental finding endorses the proposed method for image segmentation.