Plug n’ Play in Brain Prothesis
Using machine learning techniques, scientists were able to read brain signals just by placing an array of electrodes on the surface of the brain.
In a significant advance, UC San Francisco Weill Institute for Neurosciences researchers working towards a brain-controlled prosthetic limb has shown that machine learning techniques helped a person with paralysis learn to manage a computer cursor using their brain activity without requiring extensive daily retraining, which has been a requirement of all past brain-computer interface (BCI) efforts.
“The BCI field has made great progress in recent years, but because existing systems have had to be reset and recalibrated daily, they haven’t been able to tap into the brain’s natural learning processes. It’s like asking someone to find out to ride a motorbike over and all over again from scratch,” said study senior author Karunesh Ganguly, MD, PhD, an associate professor in the UCSF Department of Neurology.
In a peer-reviewed paper in Nature Biotechnology, scientists from the UCS were ready to demonstrate so-called "plug and play" performance by using an electrocorticography (ECoG) array. ECoG arrays are pads of electrodes about the scale of a Post-it note that are surgically placed on the surface of the brain to record neural activity.
An array of electrodes are placed on the surface of the brain. Photo: UCSF/SkyNews
Throughout the study, the participant's brain was ready to amplify patterns of neural activity it could use to manoeuvre the cursor via the ECoG array while eliminating the less effective signals. This pruning process is believed to be how the brain learns any complex task, as per the researchers, who said the participant's brain activity looked as if it would develop an ingrained and consistent mental "model" for controlling the interface. Eventually, once expertise was established, the researchers showed they can put off the algorithm's urge to update itself altogether, and therefore the participant could simply begin using the interface every day with none needed need for retraining or recalibration. Performance didn't decline over 44 days on the absence of retraining, and also the participant could even go days without practising and see a little decline in performance. The establishment of stable expertise in one variety of BCI control (moving the cursor) also allowed researchers to start "stacking" additional learned skills - like "clicking" a virtual button - without loss of performance.
Such immediate “plug and play” BCI performance has long been a goal within the field but has been out of reach because the “pincushion-style” electrodes employed by most researchers tend to manoeuvre over time, changing the signals seen by each electrode. Also, because these electrodes penetrate brain tissue, the immune system tends to reject them, gradually impairing their signal. ECoG arrays are less sensitive than these traditional implants, but their long-term stability appears to make amends for this shortcoming. the steadiness of ECoG recordings is also even more important for long-term control of more complex robotic systems like artificial limbs, a key goal of a successive phase of Ganguly’s research.
"Once the user has established an enduring memory of the answer for controlling the interface, there isn't any need for resetting, The brain just rapidly convergences back to an identical solution," Ganguly said. We've always been mindful of the necessity to style technology that does not find yourself inside a drawer, so to speak, but which will be actually able to actually improve the day-to-day lives of paralyzed patients. These data show that ECoG-based BCIs may well be the inspiration for such a technology."
Thumbnail Credit: pinterest.ca
Reference: Daniel B. Silversmith, Reza Abiri, Nicholas F. Hardy, Nikhilesh Natraj, Adelyn Tu-Chan, Edward F. Chang, Karunesh Ganguly. Plug-and-play control of a brain-computer interface through neural map stabilization. Nature Biotechnology, 2020; DOI: 10.1038/s41587-020-0662-5