Improved fuzzy clustering with swarm intelligence for medical image analysis

Abstract

One of the challenges in the world today is the existence of a variety of diseases, some of which require the processing of medical images to diagnose and evaluate, such as images of brain tumors. One of the methods of analyzing and evaluating patients related to the brain is magnetic resonance imaging(MRI). Data mining methods such as clustering can be used to analyze magnetic resonance images. Clustering techniques can take the area of brain tumors from brain tissue and use it to diagnose disease. Various clustering methods have been proposed so far, one of which is the fuzzy clustering or FCM method, and it has a high accuracy for clustering and segmentation of brain tissues. Fuzzy clustering is less sensitive to the noise in these images and therefore its segmentation accuracy is somewhat desirable. To improve the performance of FCM clustering, in identifying the edges and borders of tumors, it is necessary to select the optimal clustering centers. The optimal selection of cluster centers increases its accuracy in learning and segmentation. Given that the optimal selection of cluster centers is an optimization method, metaheuristic algorithms can be used for this purpose. In this research, swarm intelligence algorithms have been used to optimally select cluster centers in FCM. The analysis of the proposed method on a set of images of brain magnetic resonance shows that the proposed algorithm has the specificity, sensitivity, and accuracy of 96.87%, 88.36%, and 91.32% in the diagnosis of brain tumors, respectively. The proposed method of hybrid methods, such as the fuzzy method, better detects brain tumors.