Bridging minds and machines part 2
Raymond Yin
How can Brain Computer Interfaces help improve communication and mobility disorders? Dr. Dan Rubin, critical care neurologist at Massachusetts General Hospital and instructor at the Harvard University Medical School has joined The Tech Between Us today to discuss. I'm your host, Raymond Yin.
We've talked about ALS as being one of the afflictions that you're researching, that implanted BCIs could really help with. What other sorts of afflictions are you seeing that could really benefit from some of these therapies?
Dan Rubin
So, the mainstay of research so far has focused on people with more severe causes of paralysis. So, ALS… and ALS is one prototypical example of a larger class of motor neuron or neuromuscular diseases that have in common that people get progressively weaker and develop severe paralysis. But significantly, these conditions don't affect the cerebral cortex. They don't affect cognition. And that's important because when we're developing BCI, we're specifically tapping into cortical signals and we're trying to decode what someone's trying to do based on the signals in their motor cortex. So, their intentions, their goals are still there. We have got a signal, and that's an important contrast to some other conditions that may be amenable to work with BCIs in the future but haven't been explored yet. The other populations we've worked with are similar in that they cause paralysis without affecting cortical function. So, things like spinal cord injury, people who have stroke in the brainstem. So, the bottom part of the brain. So, a brainstem stroke is a kind of stroke that causes weakness similar to a spinal cord injury. It occurs sort of below the level of cerebral cortex and really is affecting the output of the brain to the spinal cord. They’ll often cause more severe paralysis. So, someone with a spinal cord injury may be paralyzed from the neck down, but someone with a brainstem stroke may be paralyzed, including their ability to speak.
Raymond Yin
Oh, wow. I thought a stroke was a stroke and it affects half of your body or something like that.
Dan Rubin
So that's an important distinction. So that's where most of the research has been so far. Again, spinal cord injury, ALS and motor neuron diseases, brain stem stroke, but brain stem strokes are pretty rare, whereas hemispheric strokes, embolic strokes, cortical strokes, the kind that one sort of thinks of when you think of a stroke where half of your body is weak and the other half isn't – much, much more common. One of the most common causes of disability in the world. And that's a space that we haven't yet really worked with BCIs. There's one or two groups that have been doing some groundbreaking work to try and enter that space, but it's more challenging because with that kind of stroke, the source of the signals that you would otherwise want to decode is the part of the brain that’s damaged. And so, it makes it more difficult.
Raymond Yin
Oh, I didn't know that.
Dan Rubin
So, if you've got a cortical stroke, so if you've got someone who's got a paralysis on half their body from a sort of more common hemispheric stroke, it's the motor cortex, the part of the brain that we typically think of recording from for BCI, I that's actually been injured by the stroke.
So, then that raises the question of, can BCIs help this much larger population? And although again, for a long time there's been not reluctance per se, but some reservations about going into this space because we don't necessarily want to bring people to the operating room and do surgery if we don't think that this stuff is going to work, and it's not going to be able to help them. But what we've learned, fascinatingly enough, from our participants who have these other conditions is that brain activity is highly localized. While true, is incomplete. And what I mean by that is what we've learned over the past 10 years with our clinical trial participants, we pretty much always placed the sensors in the left hemisphere.
And you may have heard that the left hemisphere is the dominant hemisphere, but more significantly for everyone, the left hemisphere controls the movement of the right side of the body. And since most people are right-handed, we record from the left hemisphere so that we can decode the signals associated with the intended movement of the right hand and arm. But what we've found is that even though the left hemisphere controls the right side of the body, when someone with sensors in their left hemisphere tries to move the left side of their body, we can actually decode that pretty well too. And we can figure out what they're trying to do with their left hand almost as well as we can with the right hand. And so even though people with a, for example, an injury to the left hemisphere become weak on the right side of their body, they don't get weak on the left side of the body, but those brain signals still had information about the left hand moving, whether that was communication that was received from the other hemisphere, or whether or not signals are being generated in both hemispheres to try and create coordinating control, we don't really know. But what this suggests is that for people who have a more, again, typical hemispheric stroke, there may be a way to use a BCI by simply placing the sensors in the hemisphere that wasn't affected and asking them to try and move the side of their body that's weak.
Raymond Yin
And still be able to pull the signals out.
Dan Rubin
Exactly.
Raymond Yin
So interesting. If you have a stroke, and let's say it damages the motor cortex on the right side, for example, is it the neurons no longer fire or is it the electrical signals are degraded to the point where you can't extract information? What does that look like? Once again, electrically?
Dan Rubin
It's a great question. To some extent, we don't know. We've never placed sensors directly in damaged cortex. My suspicion would be that there just wouldn't be any electrical activity.
Raymond
So, it'd just be a lack of activity rather than reduced voltage or reduced activity.
Dan Rubin
That's right. But again, but not all strokes are created equal. And so, there are some people who may have a relatively small stroke on say MRI or CAT scan who are very severely weak, because it’s just in the wrong place. And so, for those people, it may be not a matter of placing sensors on the other side of the brain, but placing the sensors near, but not in the area that was damaged by the stroke. And there's probably enough information there about the signal that's trying to get through the damaged part of the brain that you could, again, extract useful signal for restoring function.
Raymond Yin
I think it's incredible that some of this is still unknown. I mean, it just kind shows how complex the brain truly is.
Dan Rubin
I think it's one of the most exciting things about being in this space. I would say the second most exciting thing. The most exciting thing is working with people with paralysis and restoring function and seeing them use it.
Raymond Yin
I bet
Dan Rubin
That's the really rewarding part. But as a neuroscientist as well, it's really exciting that when we're working with our participants, we're engaging them not just in research on developing tools to help other people with paralysis, but we're learning about the brain. And ultimately by just doing what they're trying to do, by doing the movements that we ask them to do, by doing the different sort of tasks that we put together to help develop these tools, we are necessarily learning stuff about the brain that just was unknown. Although there's hundreds of years of neuroscience and basic neuroscience in mice and rats and monkeys, they’re not people. They can't do what people can do. And significantly, they can't learn the way that people learn. They can't integrate information the way people integrate information. When you want to teach a mouse how to do a particular task so that you can study how that task is performed, you've got some sensors in the mouse's brain, and then you have to very tediously train the mouse over weeks to months so that it's doing the task the way you want it done. And then you can record the brain activity.
But with a person, you can ask someone who's never done something, this is the game you're going to play. I'm going to tell you the instructions once and you're going to do it. And you can see how the brain activity responds the first time they try, the second time, so you can see how they're learning. You can see how the information is being integrated. You can see subtle things because you can change instructions on the fly and the person will do it perfectly the first time because they understand the language that you're providing them. You can see how things change when they're integrating different sensory signals. We're going to do this now, but instead of telling you what to do by putting words on the screen, we're going to have a monitor display, we're going to have a speaker play sounds, that will be your instructions. And all of a sudden, the brain looks completely different depending on whether or not the instructions for the same tasks are written or spoken.
Raymond Yin
Even though the action's the same?
Dan Rubin
Even though the action may be the same. So, one study we did in our group a couple of years back that was really interesting. We don't really know what happens in motor cortex or really in any part of the brain, but particularly the motor during sleep. So, we know that if you practice some motor skill and then you get sleep, then you'll be better at it the next day if you learn how to play the piano. Both of my kids are just started piano lessons. So now they're learning how to the piano, it's adorable, and you’ve got to practice, and it doesn't sound very good. And then the next day it sounds a little bit better and sounds a little bit better. And we all have this experience from being in school. You'd study a lot, but people would say, oh, make sure you get a good night's sleep. So that way you consolidate everything you've learned, and you have that memory available, but we don't really know how or why sleep seems to help that. So, a couple of years ago with one of our participants who said, well, let's do a little investigation. So, we had one of our participants play a little video game basically that programmed using the BCI. So, this was a young man, who had paralysis from a spinal cord injury, and we programmed a little game that was very similar. I don't know if you've played a little handheld electronic game called Simon.
Raymond Yin
Four quadrants. You have to replicate the pattern.
Dan Rubin
Exactly, precisely. And so, what we did is we had a little computer version of that, and he would play it by moving the cursor on the screen to each of those four, a little pattern. So, the decoder that we used was very straightforward, very simple, the kind I described earlier. It's simply a two-dimensional decoder that decodes where he's intending to move his hand in a two-dimensional plane. And we had him play that game for a couple of hours, and he was a good sport about it. It was kind of boring. But what we did was within that couple hours of playing without telling him, we embedded a particular sequence that appeared much more often than any other random, four pattern sequence. It was up, down, left, right. He was much more likely to be asked to make an up, down, left, right than any other sequence over the course of the afternoon. You wouldn't notice it.
Raymond Yin
Ok, sure.
Dan Rubin
What we did on that day was we did that and then we said, alright, we're going to let you go to sleep now in your bed. But unlike most other days, we were going to leave, we we're going to leave the sensors recording.
Raymond Yin
Right.
Dan Rubin
Which we had never done before. And then we'll just take a look at the data. And we did that, and we did that a couple times. And what we saw, which was really cool because we had built decoders that would take the brain's activity and turn them into the movement of the cursor on the screen when he was playing the game, we had a monitor in front of him and he could see where the cursor was going. Now, when he was sleeping, there was no monitor on, but what we found was if we took his brain activity from the nine hours that he was sleeping and just fed it into that decoder and just had the monitor show where the cursor would have gone had it been on.
Raymond Yin
Oh my gosh.
Dan Rubin
What we saw was that all night he was playing the game, and he kept making that pattern that up, down, left, right that he had been practicing doing over and over in his sleep. He was replaying the new task he learned all night while he slept.
Raymond Yin
Wow. It's like mental muscle memory.
Dan Rubin
Something along those lines. And of course, the next question we had to ask him the next morning was like, so did you have any interesting dreams last night? No, nothing that I can recall. Although it was interesting about this because a question that we got was, were we playing his dreams back on a computer screen? That's the question that people asked. And the short answer is no for any number of reasons. But in particular, because what was surprising was that, so sleep has different stages, different types. So slow wave sleep, sort of light sleep, and then REM sleep. And so, dreaming is typically associated with REM sleep, this paradoxical state where your brain looks active, but your muscles don't move.
Raymond Yin
Right.
Dan Rubin
And slow wave sleep is just, your brain kind of looks not off, but just like on standby almost.
Raymond Yin
Right.
Dan Rubin
But what was interesting is that all of this replay of motor activity occurred during slow wave sleep, not during REM sleep.
Raymond Yin
Oh, really?
Dan Rubin
Yeah. Which was again, to some extent somewhat counterintuitive. So, this standby mode where again, we think when you see very slow brain waves on an EEG, you tend to think of someone who's sort of just either in deep sleep or otherwise out of it, but there seemed to be this important memory consolidation event occurring specifically in that sleep. So even what we think of as sort of the slower sleep was clearly, or not clearly, but at least by this hypothesis, part of an active learning process.
Raymond Yin
Right. Yeah. I mean, I have read that - just articles and things like that, that while you sleep, your brain kind of cleanses your memory. It kind of goes through your memories and just kind of reviews them. And then I'm not sure what it does, but it prepares the brain for more the next day.
Dan Rubin
That's the sort of prevailing thought, is that - exactly that you can... So, there's this notion of consolidation of memory where you move things from short-term memory into a more distributed network that can remember things for a longer period of time, and the short-term memory... So short-term memory is someone gives you an address of a sound studio to show up to, and you can remember that long enough to get there. But if you were to ask me a couple of weeks from now what the address of the studio was, I probably wouldn't remember it. But I can still remember my third-grade teacher's name. It's been a while.
Raymond Yin
As a matter of fact, yeah!
Dan Rubin
So that's something that's made it into long-term memory. And the process by which things move from short-term to long-term is this notion of consolidation. We don't really know how it works, but we also know that we don't have capacity to learn an infinite number of things. So, some things you don't need to go. And so, the notion is that during sleep, there's both consolidation, you're recording things that are important, but then also exactly said cleaning things out and freshening out the closets, and something to do with the levels of calcium in your synapses, or something along those lines. Although interestingly, it seems that at least the last time I reviewed the literature, that it's actually during REM sleep, during dreaming that you're actually doing that refreshing, and it's, I guess during this slower week sleep that you're doing the consolidating part.
Raymond Yin
Oh, interesting. Because logically you would think that while your brain is active, this is when all this is happening.
Dan Rubin
Right? Again, it was an interesting thing to explore. This of course, has all just been a very long tangent to describe the way in which it's exciting to do BCI work because you get to work on technology to help people, but at the same time, you get to learn some really fascinating neuroscience along the way. And again, the same thing that we talked about earlier. This notion that neurologic function in particular motor function is so much more distributed than we previously appreciated.
Raymond Yin
If you're interested in hearing more of our conversation with Dr. Ruben, subscribe to Mouser’s Empowering Innovation Together Content series. You'll also gain exclusive access to technical articles, videos, and more. Sign up today at mouser.com/empowering-innovation.