Human brain tissues segmentation based on DTI data

Brain tissue segmentation has an important impact in diagnosing and modeling of its structure and function. Some segmentation methods such as Otsu, K-Means and Iterated Conditional Modes (ICM) are employed to distinguish between the three components of the brain which are the Gray Matter (GM), the White Matter (WM) and Cerebrospinal Fluid (CSF) using Diffusion Tensor Imaging (DTI) modality. An overlap between brain components would be produced because of macroscale scanning resolution as well as the produced noise due to artifacts. It is possible to separate brain components with less segmentation error and less computational cost when employing the Diffusion Volume (DV) quantity in segmenting CSF and non-CSF components using ICM method. The removal of the CSF before further processing will guarantee minimum overlap between the WM and CSF as well as between GM and CSF. Segmentation of WM and non-WM is accomplished using ICM method by employing Fractional Anisotropy (FA) images after removing the CSF. The obtained ratios of CSF, WM and GM in the final clustering maps are almost 18%, 34% and 48% respectively.