AI Deciphers the Complex Wiring of the Human Brain

Although it may not seem like much, one cubic millimeter of brain tissue contains 230 millimeters of blood vessels, 150 million synapses, and 57,000 cells, holding a total of 1,400 terabytes of data. This is why the recent achievement by researchers from Google and Harvard is truly remarkable.

A portion of temporal cortex roughly half the size of a rice grain, the Harvard team led by Jeff Lichtman, the Jeremy R. Knowles Professor of Molecular and Cellular Biology and recently appointed dean of science, contributed to the creation of the largest 3D brain reconstruction to date, vividly depicting each cell and its network of connections.

The study, which was published in Science, is the culmination of nearly 10 years of work with researchers at Google Research. It uses AI algorithms in conjunction with Lichtman's electron microscopy imaging to color-code and reconstruct the incredibly intricate wiring of mammal brains. The first three co-authors of the paper are Daniel Berger, a Harvard postdoc, Michał Januszewski of Google Research, and Alexander Shapson-Coe, a former Harvard postdoc.

The ultimate objective, which is funded by the BRAIN Initiative of the National Institutes of Health, is to map the neural wiring of a mouse in great detail. This would require roughly 1,000 times the amount of data that the team recently generated from the 1-cubic-millimeter piece of the human cortex.

The word ‘fragment’ is ironic. A terabyte is, for most people, gigantic, yet a fragment of a human brain - just a minuscule, teeny-weeny little bit of human brain — is still thousands of terabytes.” 

Jeff Lichtman, the Jeremy R. Knowles Professor and Dean, Department of Molecular and Cellular Biology, Harvard University

The most recent map reveals previously unobserved aspects of the anatomy of the brain, such as a powerful but uncommon group of axons linked by as many as 50 synapses. The group also observed anomalies in the tissue, like a few axons forming large whorls. The researchers are unsure of the pathology or rarity of these formations because the sample came from an epileptic patient.

The goal of Lichtman's field, connectomics, is to compile comprehensive lists of the structure of the brain, right down to the individual cell. Completing these maps would provide new perspectives on brain function and disease, which scientists currently know very little about.

Thanks to Google's cutting-edge AI algorithms, three-dimensional brain tissue mapping and reconstruction are possible. Additionally, the group has created a set of freely accessible resources that scholars can utilize to study and annotate the connectome.

Given the enormous investment put into this project, it was important to present the results in a way that anybody else can now go and benefit from them.”

Viren Jain, Collaborator, Google

The mouse hippocampus formation, which is significant to neuroscience due to its role in memory and neurological disorders, will be the team’s next focus.