101 labeled brains and a new human cortical labeling protocol

Labeling the macroscopic anatomy of the human brain is instrumental in educating biologists and clinicians, visualizing biomedical data, and determining the locations and distribution of brain data. Examples of the latter include the analysis and reporting of brain imaging data with respect to landmarks or brain regions, which is routinely done for functional, diffusion, and structural magnetic resonance images (f/d/MRI) and positron emission tomography. Consistent labeling of the cerebral cortex is challenging due to the great anatomical variation in the cortical folds and difficulty in establishing robust, accurate reference landmarks across the brain. Accurate definitions for landmarks and label boundaries is important because they underlie our assumptions of correspondence across brain image data, and affect registration and region-based analyses. Hence an accurate, reproducible labeling protocol is crucial.
Here we outline a new surface-based cortical labeling protocol based on the Desikan-Killiany (DK) protocol [1], the “Desikan-Killiany-Tourville” (DKT) protocol that promises to be more consistent and accurate with respect to macro-anatomical landmarks than previous protocols. We evaluate this protocol by comparing FreeSurfer-automated DK labels with manually edited versions of these labels according to our DKT protocol. We created this protocol to set a new standard of labeling accuracy and consistency for use by the scientific community, as well as to create the largest and most complete set of labeled brains ever released to the public, a manually edited label set of 101 human cortices from the T1-weighted MR images of publicly available multimodal data acquired from healthy individuals (http://www.mindboggle.info/data/).

[1] RS Desikan, et al. 2006. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage, 31(3), 968-980.