recentlyResearch publishedIn the project, the Google team will map all the connection structures of the hippocampus in the mouse brain, which will become the largest biological data database in human history and is expected to cost US$3300 million to achieve.
In this study, the Google team will collaborate with teams from Harvard University, the Allen Institute, MIT, Cambridge University, Princeton University, and Johns Hopkins University to address challenges in neuroscience under the National Institutes of Health's Innovative Neurotechnologies for Brain Research program.
Among them, the hippocampus structure in the mouse brain, which is responsible for memory, attention and spatial navigation, will be analyzed and mapped, which will serve as an important reference for future research on the human brain.
Prior to this, Google and Harvard University had collaborated on a map of the human brain that occupied 1 cubic millimeter of space and released it to the public as the H01 database. However, this is only a very small part of the human brain. To establish the connection structure of the entire human brain tissue, it would require a cumulative capacity of almost 10 billion TB and the problem of depicting the intricate connection structure must be solved, which is almost as difficult as exploring the universe's galaxy network.
Therefore, the current research method is to first focus on mice, whose genetic composition is 90% similar to that of humans. By using the relatively small mouse brain, which has many similarities with the human brain, as a reference, we can conduct in-depth research on the connection structure of the human brain and find ways to improve Parkinson's disease, amnesia and other human brain disorders.
In contrast, recording the connection structure of human brain tissue requires more than 10 billion TB of data, while the amount of data required to record the connection structure of the mouse brain is only 25000 TB, which is relatively easier to achieve earlier.
Google also used a single Pixel phone with 512GB of storage capacity as a unit, explaining that it would take about 2 Pixel phones to store the image data of the roundworm brain, 100 Pixel phones to store the fruit fly brain, and 48800 Pixel phones to store the data of the mouse brain's hippocampus. The stacked height is about the same as the Empire State Building. The number of Pixel phones required to store the entire mouse brain structure data is about the same as the height of Mount Everest, the highest peak on Earth.
To successfully record this massive amount of data, the Google team used deep learning technology to track and segment the pathways generated by each neuron, and recorded key features through the self-supervised learning technology SegCLR to identify the connections between different cell types and neurons. The team also plans to strengthen and expand the analysis of core connection structures to record a more complete picture of the brain's connectivity.
Currently, brain research has become an important topic in many scientific developments. Perhaps by using technologies such as artificial intelligence to fully analyze the structure of the human brain, more diseases that were difficult to solve in the past will be improved, and it may even unlock unknown functions of the human brain.
