Zenonos, Georgios; Monaco, Edward A. III; Friedlander, Robert M.
After a labor-intensive 10 year period, a unique full-scale ultrahigh-resolution three-dimensional (3D) model of the human brain has been created. Researchers from Germany and Canada have recently published their experience developing this model, providing near cellular level detail at 20-μm resolution (Amunts K, Lepage C, Borgeat L, et al BigBrain: An altrahigh-resolution 3D human brain model. Science 340(6139): 1472-5). The project has been coined the “BigBrain” and is available to everyone for free (http://bigbrain.cbrain.mcgill.ca). This one-of-a-kind model combines the previously available macroanatomic details with microstructural correlates, opening new doors into understanding how structure and function down to the cellular level combine to create the human organism.
To date, interactive structural 3D brain models have been limited to high-resolution MRI images of the brain with a level of detail on the order of 1mm. These models, although delivering excellent macroanatomical detail, yield no information about corresponding cellular organization. On the other hand, cytoarchitectonic maps, such as those originally proposed by Broadman, or Von Economo and Koskinas, were based on cellular features and were intimately associated with function, but are limited to two-dimensions (2D). Because the brain is a convoluted three-dimensional structure, the definition of borders between complex areas in space has been fraught with subjectivity and uncertainty. The “BigBrain,” as a 3D stereotactic cytoarchitectural reference atlas of the entire brain, not only allows an objective definition of anatomical boundaries in space, but also provides unprecedented possibilities for integrating multimodal information, such as functional maps, genetic expression, chemoarchitectural, or myeloarchitectural maps at both the macro-anatomical, as well as the microstructural level of cortical layers, columns, and microcircuits.
The creation of the “BigBrain” was a technically challenging process, requiring more than 1000 hours uninterrupted data acquisition time alone. A total of 7400 20-μm coronal histological sections were obtained from the paraffin embedded brain of a 65 year-old woman using a large scale microtome. After being registered to a high-definition MRI of the specimen obtained prior to sectioning, each section was stained for cell bodies and its microscopic image was digitized. Despite meticulous handling during the process, given the highly folded cerebral cortex, the large number of areas, and the sheer size of the brain, artifacts were inevitably encountered. To correct these artifacts, seemingly endless hours of manual or automated repairs were required to restore the integrity of the specimens prior to 3D reconstruction. Finally, high performance computing facilities were used to integrate all the data into a smooth and isotropic 3D digital continuum of one terabyte. Notably, Amunts et al comment that providing further detail is not currently feasible, mainly due to the limitations in computing power. For example, increasing the resolution to the level of 1 μm would require approximately 21,000 terabytes of storage space!
BigBrain is a breakthrough that bridges the gap between the perspective of macroanatomy and microscopic resolution. It redefines traditional neuroanatomical maps, opening new horizons to the neurosciences with regards to integration of multimodal data, testing hypotheses, modeling, and simulation. Furthermore, it provides groundwork for validation of concepts regarding ontogenesis. Currently cytoarchitectonic and functional probability maps from the Montreal Neurological Institute Space, embedding information about inter-subject variability, can be mapped to the BigBrain database at the website. BigBrain may also prove to be a valuable addition to the neurosurgical armamentarium in a multitude of ways. First, even in its simplest form, the high-resolution 3D perspective of the brain's structures can be a powerful learning tool for neurosurgeons in training. The accurate 3D definition of microstructurally and functionally distinct areas and their relationships may enable neurosurgeons to refine their surgical approaches through simulation, especially when combined with tools like ultrahigh-definition fiber tractography. Additionally, BigBrain may serve as a new reference basis for stereotactic procedures with micrometer detail, and through integration with genetic expression and functional maps in various disease states, it may help identify new deep-brain stimulation targets. BigBrain opens the door to an exciting new era of multimodal brain atlases. Future directions include the expansion of the palette of multimodal data that is assimilated into the atlas, as well as creation of additional subject brain models to account for variability in sex, age, race and different disease states. Furthermore, with the exponential growth of computing capabilities, cellular and subcellular detail maps may be possible, providing yet more powerful tools for unraveling the mysteries of the brain.