Cortex Parcellation Associated Whole White Matter Parcellation in Individual Subjects

Schiffler P, Tenberge J, Wiendl H, Meuth SG

Abstract

The investigation of specific white matter areas is a growing field in neurological research and is typically achieved through the use of atlases. However, the definition of anatomically based regions remains challenging for the white matter and thus hinders region-specific analysis of the white matter in individual subjects. In this article, we focus on creating a whole white matter parcellation method for individual subjects where these areas can be associated to cortex regions. This is done by combining cortex parcellation and fiber tracking. By tracking fibers out of each cortex region and labeling the fibers according to their originating region we populate a candidate image. From the candidate image we derive the white matter parcellation by classifying each white matter voxel on the basis of the distribution of labels in the corresponding voxel in the candidate image. The parcellation of the white matter with the presented method shows a high reliability and is not as dependent on registration as white matter atlases. This method allows the parcellation of the whole white matter into individual cortex region associated areas and therefore the association of white matter alterations to cortex regions. We demonstrate a comparison between the results of the presented method and existing atlases. The areas the presented method generates are not as sharply defined as the areas in most existing atlases but are computed directly in the DWI space of the subject and therefore do not suffer from distortion caused by registration. The presented approach might be a promising tool for clinical and basic research to investigate modalities or system specific micro structural alterations of white matter areas in a quantitative manner.