Scientists say that an artificial intelligence program that compares to ChatzPT has helped them create one of the details of the brain brain today, including 1,300 regions and suburines identified on the map.
Some of these subgraments have never been charted before – and researchers say there is more to come. “I think there are already hints that we can go out of what we see now,” Bosiljka TasikSeattle’s molecular Allen Institute for Brain ScienceThe
Mapping attempt led by researchers at the University of California at San Francisco and Allen Institute is detailed The study published today In the journal CommunicationThe
“Our model is built in the same powerful technology as AI equipment as ChatzPT,” Senior Writer Reza Abbas-ASLA neurologist at UCSF, Said in a news releaseThe “Both are built on one ‘Transformer’ network Which gains skills in understanding the context. “
That context may be important for the treatment of neurological illness, Tasik told Gikwire.
“The position is everything in the brain,” he said. “Defining the geography of the brain, and then defining all these regions and their activities, not only leads to better understanding, but also takes better power to treat.”
More detailed maps of the brain cellular structure can treat more targeting drugs that cause low side effects. “We always want to go to the treatment of the better, more precise brain, but to do this, where to interfere, what is wrong and what you need to know, you need to know,” said Tasik. “And if you don’t have the map you can find out where it is?”
Brain maping
The brain’s mapping attempt usually depends on the interpretation of the human physiology, but scientists are getting better to detect millions of separate brain cell locations and effectiveness. They need AI to help with explanation that they are getting better to collect a lot of data.
“We’re at a point where we have amazing experimental technology, so the next generation sequenceing has been completely revolutionized,” said Tasik. “Our way to define cell types – you can measure thousands of genes per cell and define similar cells as cell types – biology has been transformed.”
The availability of software that can deal with this national high-dimensional data is creating “an amazing time for a neurologist”, he said.
The key to the newly published study is known as an AI model Cell transformerThe To determine which cells include the same “neighbor” in the brain, the model creates a lot of data about data and functions of the brain cells.
Celttransforce has analyzed the data of about 9 million cells in more than 200 tissue cells taken from the brain of four separate rats. At first, researchers programmed the model to define the boundaries of 25 regions of the brain. Finally, they have increased the resolution to define the 670 regions and subgrades. At every level of resolution, the brain maps of the celttransformer were previously defined by human experts that were combined.
The dial was then made up to produce 1,300 regions and sub -region. At that level, successfully replicated maps of celttransformer brain cataloged regions. It has identified the previously unchanged, delicate subgrades in the region of the brain that is currently not badly understood.
Tasik said the process was like going from a map that only showed the continents or the countries on a map that showed the surrounding territories between states, cities and even cities.
“What we are saying is, let’s say we ask for a cell, ‘who is the neighbor?’ And then, on the basis of the generality of the neighbors it is called a region, “he said. “Originally, Celtransformer did this”
Some of the previously made subgrades are in the midbrane reticular nuclei, which is sensitive and a complicated role in the processing of motor data. Other newly identified subgraments are in Superior Caliculus, a part of the midbraine that processes sensitive information and starts the eye, head and body movement to concentrate on topics of interest.
Focus on new neuro-frontiers
Tasik said that it is possible to launch the dial on the algorithm of Celtransformer in order to create a brain map, which is even more detailed. “Now, the question is which one is meaningful, in what way and what do they present biologically?” He said.
Another question with the new feature subsi needs to be called. “Just imagine that you came to a new land, and you can see that it is there and there is but but now I need to be named now Now what else is around me,” said Tasik. “We want to give meaningful, systematic names and refer to how it is related to the old map” “
Probably the biggest questions will create lines with the map of the newly published map, which is based on the type of cell, which will connect to the cells between cells or brain-cell activities. “I am only expecting for more regulatory data collection, more regulatory data analysis and more multimodal models – models that will not only measure gene expressions and cell types, but will define the brain zones based on connection and productivity and all of these,” said Tasik.
Tasik said that the AI-based techniques that were made for the mouse brain mapping were “quite extended to the human brain”, but he does not expect to happen overnight.
“The limit is actually collecting data,” he said. “Human brain is huge, so it’s a problem. … I don’t want to give any guesses, but only to collect the only time it will take a decade [data about] The human brain is full at the level we have done for the mouse. “
UCSF researcher Alex Li Its chief writer Study to communicate natureTitle “Discover the mouse with transformers in the brain to discover data-driven fine granular zones” “ Other authors include Alma Dubub, Michael Cunst, Shenkin Lao, Nicholas Lask, Lydia NG, Hunkui Jeng, Bosiljka Tasik and Reza Abbasi-ASL.
