Maryam Talebi Rostami - Electrical and Computer Department Babol University of Technology
Reza Ghaderi - Shahid beheshti University
Mehdi Ezoji - Electrical and Computer Department Babol University of Technology
Jamal Ghasemi - Electrical Department Mazandaran University

one of the most commonly used methods forMagnetic Resonance Imaging (MRI) segmentation is Fuzzy CMeans(FCM). This method in comparison with other methodspreserves more information of the images .Because of using theintensity of pixels as a key feature for clustering, Standard FCM issensitive to noise. In this study in addition to intensity, mean ofneighbourhood of pixels and largest singular value ofneighbourhood of pixels are used as features. Also a method forsegmenting MRI images is presented which uses both FCM andRadial Basis Function (RBF) neural network and partly decreasesthe limitation of standard FCM.

OTK ugiogpvcvkqp= HEO= TDH pgwtcn pgvyqtm0