Mesoscale OCT Could Provide Cellular-resolution Images of the Whole Human Brain
Despite significant advances in imaging that have allowed researchers to peer ever deeper into the body, fundamental questions remain about human brain anatomy. Not least: How many cytoarchitectural areas are there (that is, parcellations of the brain based on the properties of individual cells)? What cell types can be found in these areas? To what extent do the areas vary between regions of the brain or across subjects? Answering these questions requires an imaging technology that can visualize the morphological and molecular properties of individual cells directly and without significant distortion. To date, though, no such technology exists.
This could soon change. In a recently funded grant, a team of researchers in the BU Neurophotonics Center described a means to bridge microscopic volumetric histological imaging, which offer cellular resolution but with inherent distortion due to sectioning of the tissue prior to imaging, and macroscopic MRI. The new tool will enable registration of the histological cell typing to an MRI-based atlas coordinate system – allowing reconstruction of undistorted 3D images of the human brain with high enough sensitivity and resolution to directly measure cells and their molecular properties in the brain.
“We’re thrilled to begin work on this grant,” says project lead David Boas. “The funding will enable development of cell type atlases of the human brain, showing expected variability across populations as well as changes induced by disease, all of which can be subsequently used to improve interpretation of anatomical MRIs of the living human brain.”
The key to the new approach is the use of mesoscopic optical coherence tomography (OCT). OCT is an optical technology enabling high-resolution cross-sectional imaging and 3D reconstruction in biological tissue up to several hundred micrometers in depth, non-invasively and contact-free. Integrated with a vibratome, an instrument used for sectioning slices of biological samples, OCT can be used to image the block-face prior to sectioning – thus preserving the spatial information across slices and removing the distortions and tears that are inevitable in histology sections of tissue. Histology is easily performed on the sectioned slices. At the same time, because OCT was performed prior to any cutting, avoiding introduction of deformations, OCT is easily registered to the MRI of the brain. Thus, the technology facilitates registration of subsequent histological cell typing to an MRI-based atlas coordinate system.
The proposed technology is complex and, not surprisingly, involves a number of challenges for the developers. One of the most significant of these is making the OCT data acquisition pipeline both robust and sustainable. This will be especially tricky because, in order to reduce the total acquisition time, the researchers are pushing the limits of acquisition speed with the OCT system they are using – namely, by using a larger field of view and less overlap between tiles. At the same time, they are pushing the sustainability of the acquisition pipeline – that is, the data management. Raw data from one human brain block measuring 4 cubic centimeters acquired with the technology could be more than 10 TB. Here, in order to reduce the data flow and disk load, they have implemented real-time pre-processing steps on the acquisition computer and then pushing the data to the Massachusetts Green High Performance Computing Center for further volumetric registration and reconstruction of the thousands of acquired image tiles.
The Neurophotonics Center at BU is well-suited to address these challenges: not least, says Jiarui Yang, a PhD student at the university who will be working with the technology, because it is an interdisciplinary center, well-suited to the multi-disciplinary scope of the project. “For example, the NPC has optical engineering specialists that can build an OCT system that meets our needs, wet lab technicians that can help with sample preparation, and engineering students and
faculty who can work on experimental design and image processing.” And in cases where particular resources aren’t available within the Center, the developers can turn to the myriad other facilities within the University. As just one example, the IT department at BU plays an essential role in setting up the data management solution while the researchers are using the campus shared computing cloud (SCC) service, which provides data storage space and computation power.
When completed, the proposed technology could help advance a broad range of questions pertaining to cell types and cytoarchitectural areas. “A complete human brain cell census could not only provide detailed brain anatomy information for ex vivo studies but also possible neuropathology insights for in vivo studies,” says Yang. “For example, our solution will lay the groundwork for automated whole-brain laminar modeling and architectonic segmentation, the investigation of structure-function relationships, and the assessment of histologic variability across healthy and disease populations, as well as conditions such as development and aging.”