Google has mapped the human brain in the most detail

 

Around 4000 nerve fibers are connected to this single neuron

Google has helped make the most detailed map but the connection in the human brain. This reveals a number of surprising details, including connection patterns between neurons, and what might be a new type of neuron.

Brain maps, which are available online, including 50,000 cells, all of which are given in three dimensions. They joined by hundreds of millions of spirits of Spidery, forming 130 million connections called synapses. Data sets measuring 1.4 petabytes, approximately 700 times the average computer storage capacity on average.

The data set is very large so the researchers have not learned it in detail, said Viren Jain on Google Research at Mountain View, California. He compared it to the human genome, which was still explored 20 years after the first draft was published.

This is the first time we see the real structure of the human brain that is so great, said Catherine Dulac at Harvard University, who was not involved in the work. "There is something a little emotional about it."

Hose Mammoth began when the team led by Jeff Lichtman, also at Harvard University, acquired a small brain piece of a 45-year-old woman with a resistant epilepsy. He underwent surgery to remove the left hippocampus, his seizure source, from his brain. To do this, the surgeon must issue some healthy brain tissue that lines the hippocampus.

Lichtman and his team immediately soaked the sample in preservatives, then tarnished with heavy metals such as osmium, so the outer membrane of each cell was seen under an electron microscope. Then they instilled it at the resin to harden. Finally, they cut it into slices of around 30 thick nanometers, or about a thousandth of a width of human hair, and using an electron microscope to the image of each slice.

At this point, the Jain team on Google took over, assembling two-dimensional slices - called Jain "Deli slicer approach to the brain" - to form a three-dimensional volume. They use machine learning to reconstruct tendrils that connect one neuron to another and label different cell types.

All of these details are only a small part of the brain. Jain said the scale is best understood by thinking about the scanning of functional magnetic resonance imaging (FMRI), which is used to show activities in different brain regions. "The whole set of data we produce is cubic millimeter, which is usually one pixel in MRI's scan," he said. "It's interesting to uncover everything under the hap of one pixel MRI."

For Dulac, data sets are "Trove of Goodies for years to come". The team has made new discoveries about how our brain is wired: for example, there is a difference in Stark in the number of connections between neurons.

Usually, when a tendril from one graduate neuron is close to the others, it will form one synapse, or less two to four. But there are also several tendrils that make up up to 20 synapses to one target neuron, which means this centril by itself might be able to trigger the neuron to shoot.

It is not clear why, but Lichtman speculates that multi-synapse connections underlie learning behavior. "There are many things that your brain do with cognition, by thinking and confusing and making decisions, and there are many things you do automatically that are not possible to come genetically," he said, like braking when you see red light. , A very strong connection will allow messages to pass quickly through the network.

The team also found a mysterious neuron couple far in the cortex that had not been observed before. "The two cells point to the opposite direction to the same shaft," Lichtman said. No one knows the reason.

Brain mapping, or konekomic, has been far since its first breakthrough in the 1980s, when the researchers mapped 302 neurons in the worm nervous system called Elegans Caenorhabditis. Jain, Dulac and Lichtman are part of the group which, by 2020, argues to support mapping the entire mouse's brain at the same level of detail.

"The whole mouse brain is only 1000 times greater than this, Exabyte is not petabyte," Lichtman said. "There are scale where we might be able to do that in a decade, I'm suspicious." Dulas want to see how the cortex connects to other parts of the brain, and mapping the mouse brain will reveal it.

Mapping the whole human brain will require a greater data set of 1000 times greater, zettabytes, which Lichtman said is "proportional to the number of digital content produced in a year by the planet".

But do it may not be useful. "We might find that many of it is coding information that goes through experience, and therefore every brain will be something different from each other," he said. Without understanding how information is stored, the data will be gibberish, he said.

More direct benefits are exploring how cell maps differ in people with mental health conditions, said Dulac. "Similar studies can be made in patients who also have some mental illness," he said, to explain more light about how conditions such as schizophrenia manifest.