Neuroscientists at the Seattle Allen Institute and other research institutions have completed a five-year, multi-million dollar project to release a high-resolution 3D map showing 200,000 brain clumps in mice The connection between the cells is about one grain of sand.
Data collection now Open onlineAs part of development Machine intelligence from the cortex network Program, or MICrONS for short. MICrONS was funded in 2016 US$100 million in federal grants Allen Institute and its partners Intelligence Advanced Research Project Activities, The US intelligence community is equivalent to the Pentagon’s DARPA think tank.
MICrONS aims to clear the way for reverse engineering of brain structure to help computer scientists develop machine learning systems that are more like humans, but the database may also benefit biomedical researchers.
“We basically treat the brain circuit as a computer. We asked three questions: What does it do? How is it wired? What program?” R . Clay Reid said at a press conference today. “Experiments are conducted to literally observe the activity of neurons and observe their calculations.”
The newly released data set contains 120,000 neurons and approximately 80,000 other types of brain cells, all of which are contained in 1 cubic millimeter of the visual neocortex of the mouse brain. In addition to mapping cells in physical space, the data set also tracks functional connections involving more than 523 million synapses.
Researchers from the Allen Institute participated in the project along with colleagues from Princeton University, Baylor College of Medicine, and other institutions.
Baylor’s team captured the neural activity patterns of rats when they watched images of natural scenes or movies. After these experiments, the Allen Institute team saved target samples of brain tissue, sliced them into more than 27,000 slices, and used electron microscopes to capture 150 million images of these slices.
The Princeton team then used machine learning techniques to convert these images into high-resolution maps of each cell and its internal components.
“The reconstruction we demonstrated today allows us to see the elements of neural circuits: brain cells and lines, which can draw connections between cells along the lines,” Reid said. “The last step is to explain the network. At this point we might be able to say that we can read the brain’s programs.”
The resulting insights can help computer scientists design better hardware for artificial intelligence applications. They can also help medical researchers find treatments for brain diseases that involve changes in cortical wiring.
H. Sebastian Seung, professor of neuroscience and computer science at Princeton, said: “Our five-year mission has an ambitious goal, which many people think is unattainable.” “Today, we have an amazing new mammalian cortex. The reward for the landscape. As we transition to a new stage of discovery, we are building a community of researchers to use data in new ways.”
The dataset is hosted online Brain Observatory storage service and databaseOr BossDB, through which Amazon Web Services is accessing it for free on the cloud Open Data Sponsorship Program. Google contributes storage and computing engine support through Google Cloud, database utilization Nerve scanner, An open source visualization tool developed by Google Research.
MICrONS’ emphasis on open access is consistent with the principles advocated by Microsoft co-founder Paul Allen when he put it forward. Founded the Allen Institute in 2003The Allen Institute for Brain Science is the oldest and largest department of the institute. Since Allen’s death in 2018, it has focused more on the study of neural circuits and brain cell types.