Un-Supervised MRI Segmentation Using Self Organised Maps

Abstract

The primary aim in brain image segmentation is to perform partition a given brain image into different regions which are homogeneous with some criterion. Magnetic resonance image (MRI) segmentation plays crucial role in accurate representation of white matter (WM), grey matter (GM) and cerebrospinal fluid (CSF) provides a way to identify many brain disorders, such as Alzheimer's disease, schizophrenia or dementia. In this paper presents an unsupervised method for MR image segmentation based on Self Organizing Maps (SOMs). The proposed method is consist of five stages these are image acquisition, pre-processing, feature extraction using haralick features, feature selection using principle component analysis (PCA) and tissue classification using SOM. Our proposed method is performed over real MR data provided by Internet Brain Repository (IBSR 2.0) database. Performance evaluation using Tanimoto performance index indicate that the proposed method has good segmentation results. Tanimoto performance index gives mean and standard deviation of 0.68±0.03 for white matter and 0.59±0.05 for gray matter.