Date:30/03/20

Researchers make brain-reading progress with silicon chips

The prospect of uploading minds to a computer just got one step closer to reality. A team of researchers from both sides of the Atlantic has found a way to connect the signals fired by neurons in the brain to a silicon chip just like the one working in your phone or camera.
 
The new method has shown significant enough results on mice to make it through peer-review and get published in the latest edition of scientific journal Science Advances.
 
"Here, we report a new strategy to take advantage of the scalability and electronic processing power of CMOS-based devices combined with a three-dimensional neural interface," the researchers said, with the core concept consisting of a bundle of insulated microwires connected to a large-scale CMOS amplifier array "such as a pixel array found in commercial camera or display chips".
 
To be able to dig into deeper parts of the brain, and simultaneously to scale recordings to larger areas, the scientists deployed bundles of thin microwires – up to 15 times thinner than human hair – into the brains of mice, and paired the wires to a semiconductor chip. The wires can be placed deep in the brain, and each reacts to the activity of one or two neurons.
 
The electric signals produced by active neurons travel up the microwires and are picked up by the silicon chip, which acts as an amplifier to those signals, and can in turn produce readable data for the scientific team.
 
Mihaly Kollo, from the neurophysiology of behavior laboratory at the Francis Crick Institute, who worked on the research, told ZDNet: "Each wire that conducts a signal out of the brain connects to an amplifier, and all the amplifiers sit on the same chip. We had to do some engineering at the tip of the wires to make sure that even the smaller neural signals would get picked up, but the amplifiers on the chip are very sensitive. It lets us pick up everything that the neurons are saying."
 
Using one chip to receive and transmit all the signals is a significant step-up, explained Kollo, from previous efforts to record and read brain activity. While probing the cortex with wires is nothing new, until now researchers have had to pair each electrode with a separate amplifier on the outer skull. In other words, the more wires you want, the more amplifiers you need, which makes the whole deal a lot less convenient and especially less scalable.
 
The new technology was tested on mice, which only required a few hundred wires; but the researchers said that the design could be scaled to a bundle of over 100,000 wires for larger mammals. And yes – by "larger mammals", the paper's authors do mean that the platform could be used on human brains in the future.
 
The Francis Crick Institute's Andreas Schaefer, who also participated in the research, told ZDNet that the new technology effectively bridges between advancements in neuroscience and in engineering. "The clever thing we've done," he told ZDNet, "has been to ask ourselves: 'Why has the technology for recording brain cells not moved as quickly as the technology behind cell phones and cameras?'"
 
And so, he continued, the team looked at how they could make the most of the huge strides that technology has taken in the electronics industry in recent years. And it didn't take the scientists much time to notice the potential of the ever-finer, ever-thinner structures built in ever-more powerful silicon chips.
 
"The problem is, a flat computer chip is useless to record a brain that works in 3D," explained Schaefer. "So we made the chip 3D by connecting it to a 3D bundle of wires."
 
An "ultimately really simple idea," in the words of Schaefer, that still required "a lot of complex engineering skills and collaboration", for example to make sure that the microwires connected reliably to the silicon wafer.
 
"We have demonstrated an effective method to combine the rapid progress in CMOS devices together with brain tissue-compatible probes," concluded the paper.
 
If the new method is successfully deployed in human brains, it could allow for better communication with people with paralysis or neurological conditions; but the system also works the other way around, meaning that the silicon chip can be used to inject electrical signals into precise areas of the brain.
 
"This technology could lead to tech that can pass a signal from the brain to a machine, for example helping those with amputations to control a prosthetic limb to shake a hand or stand up," said Schaefer. "It could also be used to create electrical signals in the brain when neurons are damaged and aren't firing themselves, such as in motor neurone disease."
 
The pitch is reminiscent of other scientific endeavors already underway – the most famous of which is probably Elon Musk's brain-to-computer technology company Neuralink, which has set out to come up with a perfect mix of humans and artificial intelligence.
 
The company has developed threads ten times thinner than human hair, which can – in theory – be implanted inside the brain to wirelessly monitor neuron activity. A surgical robot, also made by Neuralink, is put in charge of implanting the microscopic threads into brain matter.
 
Both Schaefer and Kollo said that they were unsure of how exactly Neuralink's team was planning to connect the wires to an amplifier. "From a research perspective, what Neuralink is doing seems very clever and exciting," said Schaefer, "but you still need to connect those flexible wires to a device that can read out the data. There are still a lot of unknowns to the technique."
 
The computer chip technology described in the study is the basis for a fully integrated brain-computer interface system that is being developed by a Texas-based company called Paradromics. The company says that it is working on an implantable chip that will record electrical activity in the brain to tackle critical conditions such as paralysis, blindness, deafness and mental health.
 
Paradromics has already placed a few patents on the silicon wafer technology, and Kollo believes it is a matter of years before the method is ready for the company to carry out clinical tests.

Views: 264

©ictnews.az. All rights reserved.