Neuroscientists from Seattle’s Allen Institute and other research institutions have wrapped up a five-year, multimillion-dollar project with the release of a high-resolution 3-D map showing the connections between 200,000 cells in a clump of mouse brain about as big as a grain of sand.
The data collection, which is now publicly available online, was developed as part of the Machine Intelligence From Cortical Networks program, or MICrONS for short. MICrONS was funded in 2016 with $100 million in federal grants to the Allen Institute and its partners from the Intelligence Advanced Research Projects Activity, the U.S. intelligence community’s equivalent of the Pentagon’s DARPA think tank.
MICrONS is meant to clear the way for reverse-engineering the structure of the brain to help computer scientists develop more human-like machine learning systems, but the database is likely to benefit biomedical researchers as well.
“We’re basically treating the brain circuit as a computer, and we asked three questions: What does it do? How is it wired up? What is the program?” R. Clay Reid, senior investigator at the Allen Institute and one of MICrONS’ lead scientists, said today in a news release. “Experiments were done to literally see the neurons’ activity, to watch them compute.”
The newly released data set takes in 120,000 neurons plus roughly 80,000 other types of brain cells, all contained in a cubic millimeter of the mouse brain’s visual neocortex. In addition to mapping the cells in physical space, the data set traces the functional connections involving more than 523 million synapses.
Researchers from the Allen Institute were joined in the project by colleagues from Princeton University, Baylor College of Medicine and other institutions.
Baylor’s team captured the patterns of neural activity of a mouse as it viewed images or movies of natural scenes. After those experiments, the Allen Institute team preserved the target sample of brain tissue, cut it into more than 27,000 thin slices, and captured 150 million images of those slices using electron microscopes.
Princeton’s team then used machine learning techniques to turn those images into high-resolution maps of each cell and its internal components.
“The reconstructions that we’re presenting today let us see the elements of the neural circuit: the brain cells and the wiring, with the ability to follow the wires to map the connections between cells,” Reid said. “The final step is to interpret this network, at which point we may be able to say we can read the brain’s program.”
The resulting insights could help computer scientists design better hardware for AI applications, and they could also help medical researchers figure out treatments for brain disorders that involve alterations in cortical wiring.
“Our five-year mission had an ambitious goal that many regarded as unattainable,” said H. Sebastian Seung, a professor of neuroscience and computer science at Princeton. “Today, we have been rewarded by breathtaking new vistas of the mammalian cortex. As we transition to a new phase of discovery, we are building a community of researchers to use the data in new ways.”
The data set is hosted online by the Brain Observatory Storage Service & Database, or BossDB, and Amazon Web Services is making it freely accessible on the cloud through its Open Data Sponsorship Program. Google contributed storage and computing engine support through Google Cloud, and the database makes use of Neuroglancer, an open-source visualization tool developed by Google Research.
MICrONS’ emphasis on open access is in keeping with the principles that Microsoft co-founder Paul Allen championed when he founded the Allen Institute in 2003. The Allen Institute for Brain Science is the institute’s oldest and largest division, and since Allen’s death in 2018, it has sharpened its focus on studies of neural circuitry and brain cell types.
