There are more possible connections between neurons in the brain than there are stars in the galaxy, making old noggin the most complex supercomputer we know of.
Mapping the connections in just one millimeter of cerebral cortex tissue required more than a million gigabytes of data and 225 million images.
Now assembled by researchers from the Lichtman Laboratory at Harvard and Google, the 3D image is the most detailed ever taken of our neurology. The hope is to replicate this method long enough to create a puzzle of the human brain called a “connectome.”
To say that the connectome could take a lifetime to complete is not an exaggeration. To complete the connectome of a C. elegans worm that owns only 302 neurons it required 12 terabytes of data. Google has also made a change, along with the Howard Hughes Medical Institute, in a fruit fly connectome, which has about 100,000 neurons. It took a year and 100 terabytes to handle about half of the brain of the fruit fly.
Here is a 2D image of the C. elegans Brain matter is extrapolated by machine learning and other scientific gubbins onto a 3D map.
Much more detailed
With between 100 and 86 billion neurons and 100 trillion synapses, electrical exchanges and signals between cells, the human brain is far more detailed than that of its invertebrate neighbors.
However, taking a cubed millimeter section of the cerebral cortex, the researchers coated it with resin and cut it into approximately 5,300 individual slices approximately 30 nanometers thick to reduce detail low enough to obtain an image with a electronic microscope. After all that, they still needed more than 200 million images before they could reassemble it into a full 3D map.
The Data set H01, as it is called, it measures 1.4 petabytes, more than the Pentagon gathers in a month, and contains 50,000 cells and 130 million synapses, making it the most complete picture of a part of the brain ever created.
Interesting observations were made, for example, that glial cells, a specialized immune cell in the brain, outnumber neurons two to one, and are available for study in a paper accompanied by a previous review on BioRxiv.
In a video produced by Google, one can see a process of zooming in and out that highlights different aspects at different levels of the sample.
SHARE this trailer with your friends on social media …