Scientific Understanding of Consciousness
Human Brain Model Ultrahigh-Resolution
Science 21 June 2013: Vol. 340 no. 6139 pp. 1472-1475
BigBrain: An Ultrahigh-Resolution 3D Human Brain Model
Katrin Amunts, Claude Lepage, Louis Borgeat, Hartmut Mohlberg, Timo Dickscheid, Marc-Étienne Rousseau, Sebastian Bludau, Pierre-Louis Bazin, Lindsay B. Lewis, Ana-Maria Oros-Peusquens, Nadim J. Shah, Thomas Lippert, Karl Zilles, Alan C. Evans
1Institute of Neuroscience and Medicine (INM-1, INM-4), Research Centre Jülich, D-52425 Jülich, Germany.
2Jülich-Aachen Research Alliance (JARA), Translational Brain Medicine, Jülich, Germany.
3Section Structural-Functional Brain Mapping, Department of Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, Pauwelsstrasse 30, D-52074 Aachen, Germany.
4C. and O. Vogt Institute for Brain Research, Heinrich Heine University Düsseldorf, D-40001 Düsseldorf, Germany.
5Montreal Neurological Institute (MNI), McGill University, Montreal, Canada.
6National Research Council of Canada, Ottawa, Canada.
7Max Planck Institute for Human Cognitive and Brain Sciences, D-04103 Leipzig, Germany.
8Jülich Supercomputing Centre (JSC), Research Centre Jülich, D-52425 Jülich, Germany.
Reference brains are indispensable tools in human brain mapping, enabling integration of multimodal data into an anatomically realistic standard space. Available reference brains, however, are restricted to the macroscopic scale and do not provide information on the functionally important microscopic dimension. We created an ultrahigh-resolution three-dimensional (3D) model of a human brain at nearly cellular resolution of 20 micrometers, based on the reconstruction of 7404 histological sections. “BigBrain” is a free, publicly available tool that provides considerable neuroanatomical insight into the human brain, thereby allowing the extraction of microscopic data for modeling and simulation. BigBrain enables testing of hypotheses on optimal path lengths between interconnected cortical regions or on spatial organization of genetic patterning, redefining the traditional neuroanatomy maps such as those of Brodmann and von Economo.
We sought to create a human brain model at nearly cellular resolution by going considerably beyond the 1-mm resolution of presently available atlases, taking advantage of recent progress in computing capacities, image analysis, and relying on our experience in processing histological sections of the complete brain. Major challenges include, but are not limited to, the highly folded cerebral cortex, the large number of areas, considerable variability among brains, and the sheer size of the brain, with its nearly 86 billion neurons and the same number of glial cells. Compared with rodent or invertebrate brains, the human brain is extremely complex: For example, the volume of a human cerebral cortex is ~7500 times larger than a mouse cortex, and the amount of white matter is 53,000 times larger in humans than in mice. The recently published data set of the digitized mouse brain with 1-μm resolution has a total amount of uncompressed volume data of 8 Tbyte. The creation of a volume with similar spatial resolution for the human brain would result in ~21,000 Tbyte. The interactive exploration (as opposed to simple storage) of such a data set is beyond the capacities of current computing. Thus, among other methodological problems, data processing becomes a major challenge for any project aiming at the reconstruction of a human brain at cellular resolution.
To create the brain model, we used a large-scale microtome to cut a complete paraffin-embedded brain (65-year-old female) coronally, and we then acquired 7400 sections at 20-μm thickness and stained them for cell bodies. Histological sections were digitized, resulting in images of maximally 13,000 by 11,000 pixels (10-by-10–μm pixel size). The total volume of this data set was 1 Tbyte. The uninterrupted data acquisition time was ~1000 hours. To generate a data set with isotropic resolution, we down-scaled all images to 20 μm by 20 μm to match the section thickness of 20 μm.
Histological processing inevitably introduces artifacts, which pose problems at all stages of the three-dimensional (3D) reconstruction process. Defects include rips, tears, folds, missing and displaced pieces, distortion (shear), stain inhomogeneity, and crystallization. We performed both manual and automatic repairs to restore the integrity of all sections before the 3D reconstruction of the whole brain as a contiguous volume. The repaired sections were registered to the MRI, which served as an undistorted frame of reference, and further aligned section-to-section with the use of nonlinear registration. All calculations were carried out on high-performance computing (HPC) facilities within the Compute Canada network and were run on Jülich Research on Petaflop Architecture (JUROPA) at the Jülich Supercomputing Centre.
The present “BigBrain” model allows the recognition of not only the borders between primary cortical areas, but also between higher associative areas.
The directionality of hemispheric growth during embryonic and fetal development and the coupling of cortical areas via fiber tracts define the spatial organization of cortical areas and their connections, as well as sulci and gyri in the adult brain.
The present findings and data on the localization of cortical areas with respect to gyri and sulci support the notion that their topographical relationship is not merely a pure geometric phenomenon, but rather the result of an interference of developmental processes and the internal structure of areas, including their connectivity. A systematic analysis of cortical borders across the whole cortical ribbon is mandatory. The variability in this relationship across individuals requires the generation of additional BigBrain data sets in the future, labor-intensive work that is currently underway.
The BigBrain data set will be made publicly available to promote the development of new tools for defining 3D cytoarchitectonic borders. BigBrain allows the extraction of parameters of cortical organization by enabling measurements parallel to cell columns (e.g., cortical thickness, densities of cell bodies per column, surface measures) to provide a “gold standard” for calibrating in vivo measurements of cortical thickness and other measures.
The BigBrain data set represents a new reference brain, moving from a macroanatomical perspective to microstructural resolution. This model provides a basis for addressing stereotaxic and topological positions in the brain at micrometer range (e.g., with respect to cortical layers and sublayers). BigBrain will make it possible to localize findings obtained in cellular neuroscience and mapping studies targeting transmitter receptor distributions, fiber bundles, and genetic data. The BigBrain model can also be exploited as a source for generating realistic input parameters for modeling and simulation. It thus represents a reference frame with nearly cellular resolution, a capability that has not been previously available for the human brain, while considering the regional heterogeneity of human brain organization.
[end of paraphrase]
Return to — Modeling Brain Functionality of Consciousness
Return to — Scientific Understanding of Consciousness