By tuning to a subset of brainwaves, Michigan scientists have dramatically reduced the energy requirements of nerve interfaces while improving their accuracy ̵1; a discovery that could lead to long-lasting brain implants that can treat neurological diseases and enable mind-controlled prosthetics. and machines.
A team led by Cynthia Chestek, an associate professor of biomedical engineering and the nuclear faculty of the Robotics Institute, estimated their approach to reduce the energy consumption of neural interfaces by 90%.
“Interpreting brain signals according to someone’s intent now requires computers as tall as humans and a lot of electricity – it costs a few car batteries,” said Samuel Nason, the first author of the study and Ph.D. candidate in the Chestek Laboratory for Cortical Neural Prosthetics. “Reducing the amount of electricity by an order of magnitude will eventually allow brain-machine interfaces at home.”
Neurons, the cells in our brains that carry information and activity around the body, are noisy transmitters. Computers and electrodes used to collect data on neurons listen to a radio station jammed between stations. It has to decipher the real content between the buzzing of the brain. In performing this task, the brain is the focus of this data, which increases performance and processing beyond secure implantable devices.
Currently, scientists can use transcutaneous electrodes or direct conduction through the skin to the brain to predict complex behaviors, such as grabbing an item in the hand from neuronal activity. This is accomplished with 100 electrodes that capture 20,000 signals per second, and allows for activities such as reactivating an arm that has been paralyzed or allowing someone with a prosthetic hand to feel how hard or soft the object is. However, this approach is not only practical outside the laboratory environment, but also carries the risk of infection.
Some wireless implants created with high-efficiency application-specific integrated circuits can achieve almost the same performance as transcutaneous systems. These chips can collect and transmit about 16,000 signals per second. However, they still need to achieve consistent operation, and their tailor-made nature is an obstacle to being approved as safe implants compared to industrially produced chips.
“It’s a big leap forward,” Chestek said. “It would be completely impossible to get the high-bandwidth signals we currently need for wireless communication with the brain interface, given the power supply to existing devices in the pacemaker.”
To reduce energy and data needs, scientists are compressing brain signals. Focusing on fluctuations in nerve activity that exceed a certain power threshold, called the threshold crossing rate or TCR, means that less data needs to be processed while still firing neurons can be predicted. However, TCR requires listening for a complete focus of neuronal activity to determine when the threshold is exceeded, and the threshold itself can vary not only from one brain to another, but in the same brain on different days. This requires tuning the threshold and additional hardware, battery and time.
In another way, they compressed the data, calling the Chestek Lab a specific feature of neural data: bandwidth performance. SBP is an integrated set of frequencies from multiple neurons, between 300 and 1,000 Hz. By listening only to this range of frequencies and ignoring others, receiving data from straw as opposed to a hose, the team found a highly accurate prediction of behavior with significantly lower energy requirements.
Compared to transcutaneous systems, the team found that the SBP technique is equally accurate, taking as many signals as one tenth, 2,000 versus 20,000 signals per second. Compared to other methods, such as the use of a threshold overrun rate, a team approach not only requires much less raw data, but is also more accurate in predicting neuronal firing, even between noise, and does not require threshold tuning.
The team’s SBP method solves another problem limiting the life of the implant. The interface electrodes cannot read the signals between the noise over time. However, because this technique works just as well when the signal is half that required of other techniques, such as threshold crossings, the implants could be left in place and used longer.
While it is possible to develop new interfaces between the brain and the machine to take advantage of the team method, their work also unlocks the new capabilities of many existing devices by reducing the technical requirements for translating neurons to intentions.
“It turns out that a lot of equipment was sold short,” Nason said. “These existing circuits using the same bandwidth and power are now applicable to the entire brain-machine interface.”
The study “Low-energy band of neuronal spike activity, in which individual units dominate, improves the performance of the brain-machine interface”, is published in Nature Biomedical Engineering.
Is it possible one day to control the prosthetic limbs with human thinking?
The low-energy band of neuronal deflection activity, dominated by individual local units, improves the performance of the brain-machine interface, Nature Biomedical Engineering (2020). DOI: 10.1038 / s41551-020-0591-0, www.nature.com/articles/s41551-020-0591-0
Provided by the University of Michigan
Citations: Very low power brain implants find a meaningful signal in gray matter noise (2020, July 27), which was obtained on July 29, 2020 from https://medicalxpress.com/news/2020-07-ultra-low-power- brain-implants-meaningful .html
This document is copyrighted. Apart from fair treatment for the purposes of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.