Brain tissue classification in MR images based on a 3D MRF model

Intensity-based classification of MR images has proven problematic, even when advanced techniques are used. The partial volume effect and the inhomogeneity are usually sources of difficulties. Here, the authors propose a new classification method using 3D MRF models and the multifractal dimension measure for segmenting CSF, gray matter and white matter in MR T1-weighted images. Mixclasses (mixture of two pure tissue classes) result from the partial volume effect, are taken into account in the authors' tissue class model. Results are described with two acquisition sequences: IR-FGRE and SPGR. The accuracy of the classification is found by the way of a phantom validation study.