Mapping all the connections among the brain’s 100 billion neurons is one of the great scientific and engineering challenges of our time. Scientists, including those at the University of Chicago, began this work in earnest about a decade ago in a quest to build a complete wiring diagram of the brain. This diagram – known as the connectome – will help researchers reverse engineer the brain and unlock secrets of how it works at the most fundamental level.
So far, they’ve made tremendous progress on some of the initial technical challenges. The processes for preparing brain samples to take images under a microscope efficiently have vastly improved. Artificial intelligence, machine learning models, and raw computing horsepower are now up to the task of organizing and analyzing the vast amounts of data contained in these images. But one bottleneck remains: it still takes too long to capture images of each brain sample.
“We are kind of stuck at some hard limits on how quickly we can image the data, and there's not an obvious path how to get past those with the kinds of microscopes we’ve been using,” said Gregg Wildenberg, a staff scientist in the Department of Neurobiology at UChicago.
Wildenberg works with Narayanan “Bobby” Kasthuri, Associate Professor of Neurobiology at UChicago and neuroscience researcher at Argonne National Laboratory, who has been pioneering connectome research over the past decade. Now, they are joining a team of scientists from UChicago, Chicago State University (CSU), and the University of Illinois Chicago (UIC) to adapt a different imaging technology, called photoemission electron microscopy (PEEM), for connectomics and drastically increase the speed of capturing brain images. The group recently received a $4.8 million, three-year grant from the Brain Research Through Advancing Innovative Neurotechnologies (BRAIN) Initiative, part of the National Institutes of Health (NIH).
Brains on a deli slicer
Until now, connectome researchers have been using two kinds of microscopes: scanning electron microscopy (SEM) and transmission electron microscopy (TEM), each with their own advantages and disadvantages. SEM is fairly forgiving with samples, which can be easily deposited on standard silicon wafers, and it works with samples with a wide range of thicknesses. But SEM also scans across the sample pixel by pixel in a grid pattern, which can be very slow. TEM solves this problem by capturing large regions of an image simultaneously, but the samples must be extremely thin and mounted on fragile grids, which are difficult to work with, especially in large volumes.