A comparative performance analysis of fuzzy C-means based methods for segmentation of brain magnetic resonance images

Fuzzy C-means (FCM) is the classical clustering algorithm for segmentation of tissues such as cerebrospinal fluid (CSF), gray matter (GM) and white matter (WM) from magnetic resonance images (MRI). Medical images such as brain MRI are often corrupted by Rician noise thus, segmentation with FCM becomes problematic. Various variants of the classical FCM have been proposed which are robust to image noise. This paper presents a comparative performance evaluation of brain MRI segmentation methods based on Fuzzy C-means using Jaccard similarity (JS).