Iman Kalantari - Department of Electrical Engineering ,Iran University of science and Technology, Tehran, Iran
Amir Hossein Foruzan - Department of Biomedical Engineering, Engineering Faculty, Shahed University, Tehran, Iran
Shahriar B. Shokouhi - Department of Electrical Engineering ,Iran University of science and Technology, Tehran, Iran

Accurate segmentation of brain tumors is ofimportance with respect to diagnosis, treatment planning andmonitoring. Several automatic and semi-automatic methodshave been proposed to tackle the problem. Variations of HiddenMarkov Models have been extensively used in this regard.Individual models are usually assigned to represent differenttissues. To consider intensity variations, compensation methodshave been employed as pre-processing step. In this paper, weemploy HMM models to automatically segment brain tissuesincluding tumors in MRI datasets. We assume a random texturefor brain tissues to cope with texture variations and use a newfeature set so as to train HMM models. The employed featuresets are both robust against noise and rotation-invariant.

Brain tumor, Image segmentation, Magnetic resonance Imaging (MRI), Ridgelet Transform, Hidden Markov Model (HMM)