Development of a neuro-fuzzy MR image segmentation approach using fuzzy c-means and recurrent neural network

Abstract

A neuro-fuzzy clustering framework has been presented for a meaningful segmentation of Magnetic Resonance medical images. MR imaging provides detail soft tissue descriptions of the target body object and it has immense importance in today's non-invasive therapeutic planning and diagnosis methods. The unlabeled image data has been classified using fuzzy c-means approach and then the data has been used for training of an Elman neural network. The trained neural net is then used as a ready-made tool for MRI segmentation. © 2009 Springer-Verlag Berlin Heidelberg.