Markov random field-based parcellation of the cerebral cortex: application to histology images

The cerebral cortex is composed of regions with a distinct laminar structure. Functional neuroimaging results are often reported with respect to these regions, usually by reference to a standard brain "atlas". Motivated by the need for more precise atlases, we have developed an algorithm that parcellates the cortex into regions of distinct laminar structures. The problem is cast into a segmentation framework and modelled using a Markov random field-based approach. An evaluation of the algorithm on histology images is presented. This technique is presented with a view to producing parcellations on 3D in-vivo magnetic resonance (MR) images of living human subjects.