MRF models and multifractal analysis for MRI segmentation

Abstract

We demonstrate the interest of the multifractal analysis for removing the ambiguities due to the intensity overlap, and we propose a brain tissue segmentation method from MRI images, which is based on Markov random field (MRF) models. The brain segmentation consists of separating the encephalon into the three main brain tissues: gray matter, white matter and cerebrospinal fluid (CSF). The classical MRF model uses the intensity and the neighborhood information, which is not robust enough to solve problems, such as partial volume effects. Therefore, we propose to use the multifractal analysis, which can provide the intensity variations, to describe brain tissues. The value of the Holder exponent /spl alpha/ is calculated, and the corresponding multifractal spectrum f(/spl alpha/) is defined. The /spl alpha/ priori knowledge about (/spl alpha/,f(/spl alpha/)) is modeled and then incorporated into an MRF model. This technique has been successfully applied to real MRI images. The contribution of the multifractal analysis is shown.