BRAIN POWER

Neurosurgeon Daniel Kramer explains the science behind neurological implants

Interviewed by Mara Kalinoski

April 2024

Daniel Kramer, MD, assistant professor of neurosurgery, studies neural signals and communication between brain areas that produce movement and somatosensation. He is developing a brain-computer interface (BCI) program at the University of Colorado School of Medicine focused on sensorimotor restoration.

Kramer offers an expert look into the science of BCIs and explores the possibilities of that work and discusses the January announcement by Neuralink, which was founded by billionaire Elon Musk, that it had completed its first brain chip implant in a human patient, raising questions about risks, opportunities, and ethics.

WHAT IS A BRAIN-COMPUTER INTERFACE?

Although generically a brain-computer interface means any computer interfacing directly with the brain, most commonly people mean a sensorimotor BCI. In this type of BCI, electrodes are placed in or around the brain that can record brain signal changes and interpret them to mean something, then use that meaning to control a robotic limb, a cursor on the screen, an exoskeleton, or really anything. These electrodes can also deliver electricity to the brain and create artificial sensations.

WHAT KIND OF DISEASES IS THIS TECHNOLOGY TARGETING?

Usually, sensorimotor BCIs are for spinal cord injury patients, those with ALS, or stroke.

There’s lots of interesting research going on in the wider field of neuromodulation as well, covering issues such as Parkinson’s disease, epilepsy, addiction, post-traumatic stress disorder, Tourette syndrome, and more.

CAN YOU TELL US ABOUT SOME SUCCESSES OF BRAIN-COMPUTER INTERFACES?

There is a small but robust number of groups around the country who are making opportunities to study different applications of BCIs for sensory-motor and visual restoration, and that leads to tons of interesting sub-projects.

For a patient with a spinal cord injury from trauma, they’re usually intact above the injury, so they can still talk, for example. But in diseases like ALS (Lou Gehrig’s disease), you lose all muscle activity. So, in the area of ALS and similar diseases, there’s been a lot of really cool research on decoding, handwriting, and generating speech. During decoding, brain signals indicate what the person wants to do, and we read those signal and convert them into real actions.

There’s also a group that’s working on reanimating the muscles of paralyzed limbs by stimulating into the somatosensory cortex. There are highly dexterous robotic arms, which aim to provide the same motion as a real hand, cursors, exoskeletons with which people can walk, and many more applications.

WHAT ARE SOME OF THE RISKS OF THESE SURGICAL IMPLANTS?

The biggest risks are always infection and wound healing complications. Particularly, the way most brain-computer interfaces are done, like with Blackrock’s Utah Arrays (high channel count microelectrode arrays that record and stimulate neurons), parts stick out of the scalp. That’s something that concerns implanting neurosurgeons, because we’re always worried about infection, however, so far, they have been well tolerated without wound issues. While implanting the devices, there’s always a risk of damage to the underlying cortex. Bleeding and strokes are always possible. But across all groups who have undergone implantations, there have been very few problems.

CAN YOU TELL US ABOUT NEURALINK'S TECHNOLOGY?

I’ll compare it to the traditional Blackrock Utah Arrays. The idea of all these brain-computer interfaces is the same: you record enough brain signals that you can interpret and decode what someone wants to do. In theory, someone just needs to think about or try to reach out and touch something, then that signal gets again decoded. And there’s enough variation in the neural activity that you can figure out if they want to move up and to the left or down and to the right, etc. All motor BCIs are the same in that way.

The main difference with Elon Musk’s Neuralink is that they made use of a system that is basically an extremely fancy sewing machine that can, in theory, do the surgery without a neurosurgeon, although they still have to open up the skull and manually monitor everything. It aims to implant extremely tiny threads, each of which has multiple electrodes on it, in higher channel counts than other technologies.

The other piece that’s unique is that they’ve established a communication system that aims to communicate through Bluetooth, so theoretically it could be fully implanted, meaning nothing sticking out of the skin.

WHAT ARE THE NEXT STEPS FOR BCIs OR NEURALINK TO BECOME MORE COMMON?

All the tools are in place. This technology has been theorized, and the barrier at this point for us moving forward and getting it out there into the world has been more logistical than scientific. The 5 biggest logistical piece is moving the technology to the next component for FDA approval.

In addition, there are many moving factors, including funding for BCI research. Neuralink had a huge initial investment from Musk, which no other company or group has. Most scientific groups at universities are moving from NIH funding to NIH funding. So, if a company (Blackrock or Neuralink) is persistent enough and well-funded enough to get a fully implantable system, they might just do it.

But it all very much remains to be seen. There are a lot of pieces to it also that aren’t trivial at all, such as durability of the implants, how reliable the Bluetooth communication is, recharging, and other real-world concerns.

WHERE IS THIS TECHNOLOGY IN TERMS OF FDA APOPROVAL, AND ARE THERE ETHICAL CONCERNS? 

Neuralink is not currently FDA approved. It has an IDE, an investigational device exemption, which means you’re allowed to implant it under very specific circumstances and careful watch by the FDA. Utah Arrays are FDA approved for 30-day implants, or longer through an IDE.

There’s a divide on ethical concerns in this field in general, not only with Neuralink. Most ethical concerns are theoretical at this point: if a certain problem happens in the future, how would we deal with it? These include things like hacking and even mindreading. None of these systems have gone live in the sense that patients have been sent home with it to use it unsupervised, so those concerns are worth debating, but not a major concern at this point. The science would have to advance quite a bit, and the usage mainstream before these things would be of real issue.