Measure brain activity – Ryan Field on the Harry Glorikian Show
My guest this week is Ryan Field, who’s the chief technology officer at Kernel, and we’ll talk about how to measure brain activity and all about Kernel’s innovative technology in today’s episode.
They’re based in Southern California and their vision is to develop a consumer device that would work kind of like a pulse oximeter, but for your brain to measure brain activity.
The first version of this device is called the Kernel Flow.
It’s shaped like a bicycle helmet, and it contains more than 50 low-power lasers that beam light through your scalp into your skull, into the outermost layers of your brain.
Hundreds of detectors built in the helmet collect the light that’s scattered back.
Judging from the travel time and the intensity of the reflected light at different wavelengths, the Kernel Flow can measure hemoglobin concentration and oxygen levels in the brain’s blood supply.
And that’s an indirect measurement of neural activity.
A computer collects the data and uses it to reconstruct a map of how hard your neurons are working and which neurons are communicating with each other, all over your brain.
Field says company isn’t exactly sure why consumers would want that kind of data or what they would do with it.
But the company is already using the device in early studies designed to measure a user’s level of focus on a specific task, like whether you’re paying attention or getting bored during a driving simulation game.
Kernel is also using the device to study the brains of chronic pain sufferers before and after treatment with a virtual reality therapy.
And in the future, researchers and developers might come up with a totally new way to use Kernel’s neural activity maps.
Ryan says what Kernel has done is sort of like building the very first iPhone — but if the only app the device came with was Maps. Then it would be up to app developers around the world to figure out what else to do with the device.
I interviewed Ryan about the coming revolution in consumer brain monitoring back in early December of 2022, when the World Cup was still underway. So you may hear a soccer reference.
Here’s our full conversation.
Harry Glorikian: Ryan, welcome to the show.
Ryan Field: Yeah. Thanks for having me, Harry.
Harry Glorikian: I want to start off because we have a we have a pretty broad audience, right? If you can start off by explaining at a high level how the technology of functional brain imaging has advanced over the last few decades. I mean, because when you talk about how to measure brain activity in real time, right, I think most people probably imagine things like functional MRI, right. Which is a technology that’s been around for more than 30 years and usually relies on big, expensive hospital based machines to measure changes in blood flow in the brain, which is, you know, sort of an indirect measurement of neural activity. But that’s not the only way to make a movie, so to speak, of the brain in action. I think Kernel in particular uses a different technology called time domain functional near-infrared spectroscopy. You guys got to come up with a shorter name, but you got it.
Ryan Field: You got it, though.
Harry Glorikian: And you’ve managed to miniaturize it down to the size of a bicycle helmet. So, you know, I guess I’ll just say, like, how would you summarize the big advances? And, you know, I can throw some more questions when we go from there.
Ryan Field: Yeah. So that’s a good place to start. So you’re right. Brain imaging has been around for decades. There have been a lot of very talented researchers doing work in this space. And when people think about functional brain measurement or like if you think more broadly about just like BCIs or brain-computer interfaces, a lot of people like in their mind immediately go to like, Oh, I can think something and communicate to my computer, or I can control a mouse or a cursor or something like that. And yes, there are companies working on that and trying to build those like implanted devices that measure the direct electrical signals of the neurons. But you also mentioned in the intro, we don’t measure directly what the neurons are doing. We measure this indirect signal, which is how does the blood oxygenation around those neurons change when they do work? Whenever you’re thinking or concentrating or feeling sad or whatever it is, the representation of your being in your brain changes. And as it changes the oxygen levels in the blood around it also change. And that’s really what we’re measuring. And that’s very similar to what fMRI measures is just change in blood oxygenation. So the simplest way I like to present it is a lot of people are familiar with pulse oximeters these like finger clip devices that they put on at the hospital or doctor. We’ve built a device that is like that but generates a map of the blood oxygenation on the brain. So it’s like a pulse oxygenation, pulse oximeter map of the brain. And so you can think of it as just like tiling those things all over the head and measuring very small spatial volumes of changes in blood oxygenation in the brain. And so that’s like the most familiar thing that I can really connect to the technology behind what we’ve been building.
Harry Glorikian: So are we you know, we’re going to get into it here, but are we at a turning point where you might be able to get functional brain imaging technology out of the lab and into the homes or offices or other places.
Ryan Field: Yes, I think we’re right on the cusp of that. And that was really the reason that Kernel was started to begin with. So our founder and CEO, Bryan Johnson, he asked kind of why can’t we measure our brains easily? Like, have you ever had your brain measured outside of a hospital or like having a serious medical condition? The answer is probably no. And the hypothesis when Kernel was founded was that there’s a lot of information that we maybe want to have about what our brains are doing that we don’t currently have an easy and accessible way to get at. And so Kernel was started there and we set out to build technologies that had the characteristics of something that provides utility. Like you can get these same kind of maps and information that you would get from a research system in a lab. It had to be something that could be wearable. It needed to be something that you don’t have to sit inside a tube to get measured or be inside of a shielded room or something that requires a special environment. We wanted to be able to get this out of the labs and into more real world settings.
How to measure brain activity?
There are several methods to measure brain activity, including:
- Electroencephalography (EEG) which measures electrical activity in the brain through electrodes placed on the scalp.
- Magnetic Resonance Imaging (MRI) which uses a magnetic field and radio waves to create detailed images of the brain.
- Positron Emission Tomography (PET) which uses a small amount of radioactive material to measure brain activity by detecting the radiation emitted by the brain.
- functional Near-Infrared Spectroscopy (fNIRS) which non-invasively measures changes in brain activity by detecting changes in blood flow. This is the method used by Kernel.
- Single-photon emission computed tomography (SPECT) uses a small amount of radioactive material to measure brain activity.
- Transcranial magnetic stimulation (TMS) which uses magnetic fields to stimulate small areas of the brain and measure the resulting electrical activity.
Each method has its own advantages and disadvantages, and the choice of which method to use to measure brain activity depends on the research question and the specific requirements of the study.
Ryan Field: And then third, it had to have the characteristics of being something that could be built at scale. So we didn’t want to build a device that required a lot of precise manual labor and like these expensive optical processes or expensive manufacturing processes, because those things don’t scale well. And instead, what we did is we really constrained ourselves to build something that could be created and manufactured around the existing consumer electronics supply chain. And so we’ve done all of those things with Kernel Flow, the device that we’ve built and are ready to launch our first production version of early next year. We haven’t hit the volumes, like we’re not selling iPhone numbers of headsets, so we don’t quite have the cost down to where we think they can get eventually. But at a point when the applications and uses of neuroimaging start to evolve more, I think we’re primed with a system that is all the things that I mentioned before. Useful, portable, easy to use and scalable, ready to kind of be built in mass. So.
Harry Glorikian: So before we get into the applications, I’d love to spend a little time like drilling down into the tech. Why have you focused on the, you know, TD-FNIRS, you know, as the technology you wanted to use to measure brain activity in this product that you’re calling Kernel Flow? I mean, what are the advantages of that approach? I know that you guys have experimented with other brain measurement technologies in the past. One of your earlier devices, I think was called Kernel Flux, and it used a method called OPMEG, which is what is optical pumped magnetometers for. Is it better than the other one that you were using? Right, for consumer measurement?
Ryan Field: Yeah. So it’s different. And if we go back to those three criteria of what what we wanted to build, it was a device that could well, I didn’t say this, but it was kind of implied. You need to be able to measure the whole brain, right? So if you look at things like Neuralink or some of these other invasive technologies, looking at just a very small part of the brain, and that’s really useful if you’re trying to restore motor function or restore a specific function or cure a very specific problem. But our hypothesis was that these kind of functional measurements are going to be, it’s going to be important to have a picture of the whole brain. So even though it’s not as high resolution as those implantable devices, you still get the whole image. And that lets you look holistically at how the whole brain is working. So we built these two devices, Kernel Flow, which is a TD-FNIRS device, and Kernel Flux, which is an OP-MEG device. And as we built them, we kind of started to see what the strengths and weaknesses of them are. Kernel Flow is really great for measuring precise spatial volumes of these changes in blood oxygenation. Kernel Flux was really great at measuring very high-speed transient signals as the brain was firing and doing things. So we’re looking at the magnetic fields that neurons generate when groups of them fire together. So all the neurons communicate. It’s like an electrochemical process, and you have electrons moving around, and moving electrons create a current, and that current creates a magnetic field. It’s very small, but there’s a magnetic field there to measure. And so we were using these OPMs, optically pumped magnetometers, in order to measure those very small magnetic fields.
Ryan Field: And we had some development paths where we were trying to shield the magnetometers from the background. So in the environment, the earth creates a magnetic field that’s on the order of microteslas. The brain creates a magnetic field on the order of femtoteslas. So there are nine orders of magnitude difference between the signal you want to measure and the Earth’s background. And then you’ve got all these other electronics in the world that are generating magnetic fields. And everywhere you go, there’s a new obstacle. So it’s a really big challenge to shield out the environmental noise so you can focus in on the neural signals. It can be done, and it’s typically done in a shielded room. So you just have a dedicated place where you shield it off and block those external magnetic fields from coming in. So Kernel Flux, when we got there, we tried to build these miniature shields and do all these different things. But at the end of the day, what we found is that there’s no way we’re going to get both whole-head coverage and a device that could be worn outside of a shielded room. Or not in the near term at least. So there are some very long term technology developments. So we decided to stop work on Kernel Flux and focus on Kernel Flow, where it met all three of our criteria, in being able to be high utility, scalable, and something that could be used outside of the laboratory environment.
Harry Glorikian: So this technology, though, has been around for about 30 years. I know it’s always been in lab technology, you know, way too big, way too heavy, way too expensive to build into a portable device. What sort of obstacles did you guys have to overcome to build this device that would be accessible to consumers? And, you know, I’m trying to think of like, what kind of engineering decisions did you have to make along the way going, you know, do we go left or do we go right? And does that get us to where we need to go to? Because you got to make some sacrifices or some changes along the way possibly to get to this level. Or has technology evolved to the point where, you know, we can actually do some of these things that used to be in the big machines, in something portable?
Ryan Field: Yeah, that’s a great question and it’s a combination of all those things. So when we started on Kernel Flow almost five years ago now, we didn’t know what we were really doing. We didn’t know what was required to build a TD-FNIRS device that could measure brain activity reliably. And so we started out on the development path where we bought a system like big boxes that were commercially available and use that to build a first prototype. Then we started to build our own custom hardware and built that in phases where each phase we could run some tests, learn a little bit more about what was required, where the key performance metrics were, and kind of step through understanding what was actually required in order to do the job and not just what had been used previously to do the job. Because you’re right, you have to make some tradeoffs and sacrifices. And that’s what engineering is. It’s taking a look at the application and taking a look at your technology stack and saying, how can I map this technology onto this application and what tradeoffs am I going to have to make between performance or some key metric here and what’s just physically possible from an engineering perspective? So we did that in a very, I think, systematic and steady way where we just kind of built through different prototypes and learned and got feedback along the way.
Ryan Field: The other key piece that came into this was we had this realization that there was, and I come from this industry, so I knew, and that’s part of the reason why I joined Kernel, is, there’s a lot of technology being developed for autonomous vehicles in a sensor called LiDAR, which is a laser based, pulsed time of flight system. And so there’s been a lot of innovation in that space where you’re doing these time of flight measurements to get range and distances. And we wanted in the brain to do a time of flight measurement on a much shorter time scale, but very similar in overall structure and architecture. So we took advantage of some technology development both on the laser side, to create these very short pulsed laser drivers, and also on the detector side, where a lot of commercial foundries, businesses that provide silicon wafers and CMOS process development for others, they were developing these technologies that could be used for LiDAR and doing the detection on LiDAR. So we took advantage of those technology developments and all the investment that went into LiDAR and repurposed those same kinds of technologies for measuring the brain. And so it was a combination of kind of systematically stepping through these different trade offs that we have to make and also taking advantage of some recent technology developments in an adjacent field.
Harry Glorikian: Yeah, I mean, I just you know, I’m always stepping back and fascinated by how you can, if you’re willing to step back a little bit, there’s other stuff going on in other fields and they’re moving so fast that you might be able to adapt what they’re doing to what you’re doing, as opposed to just trudging along your traditional path. And I think there’s so much happening in so many fields that you almost need to be able to do that these days to really get that stepwise function that you’re looking for and in sort of a breakthrough product. But…
Ryan Field: Yeah. I would add to that, I think one of the things that Kernel has done really well is we have a team of people who isn’t content with the answer “That’s the way things have been done.” So you ask a question and if someone says, “Well, that’s just the way we do it,” or “That’s the way that things have been done,” that doesn’t fly here. So you challenge it, you understand what were the reasons that that decision was made decades ago when the technology was founded, for instance. And then you kind of say, oh, well, the assumptions that were used, for making the decision the way it’s always been done, are totally different than what they are now. When you take a step back and like you were saying, like look at what’s going on in these other fields and the developments that are happening very quickly there.
Harry Glorikian: So now correct me if I’m wrong, but the current Kernel Flow has 52 laser light sources and 312 detectors, and I guess that’s six detectors per laser. That’s up from eight lasers and 48 detectors on the beta version of the device from 2020. And this is probably a dumb question, but, you know, A, r why are more detectors better? And B, can you walk us through why that was so important to add more source detectors? And what can you do now that you have that many?
Ryan Field: Yeah, that’s a great question, and I think really fundamental to why it matters to kind of do the level of integration that we’ve done. If you only need a few sources or a few detectors, you don’t have to be as extreme in integrating everything as we were. But if you want to get a very dense coverage of the head and have a very dense map with many sources and many detectors, then you really have to do a lot of engineering work. So what the number of sources and detectors, the impact that that has is, between every source and detector we form what’s called a channel. We consider the light path that goes from a source or a laser into the head and back out and is measured at a detector as a channel. And you can think of these as little, we call them banana paths. So you can think of like a bunch of mini bananas all over the brain between every source and detector pair. And where they cross over, you get different perspectives of the same volume of measure. And so you can use all of this information of overlapping source and detector pairs or channels to create a nice mesh that then tells you where exactly the change in blood oxygenation is happening. So if you only have a measurement along one dimension, you can only say, “Oh, the change happened somewhere along this one banana.” But if you have two bananas that cross and they both see the change, then you know, the change had to happen in the intersection of where the two bananas are. And so the more source and detector pairs you have, the more density, more spatial resolution you have about what’s changing.
Harry Glorikian: So, okay, let let’s switch gears here just for a bit and talk about the biology of what your technology actually measures. I mean, can you can you walk us through exactly what you’re capturing when you’re recording this data? I mean, how deep into the brain can it penetrate? What does hemoglobin concentration, oxygenation, deoxygenation tell you about what’s happening in those layers of the brain? And if I’m missing something, just tell me.
Ryan Field: Yeah. Yeah. No, it’s good. Those are the right foundational words for this discussion. So how does the technology work? I mentioned early on it’s very similar to a pulse oximeter, so it’s the same fundamental technique that’s used there. And it’s a technique called near-infrared spectroscopy. And so the near-infrared part just is describing what wavelength of light we use. So we use light that’s at two different wavelengths. One is at 690 [nanometers] and the other is at 850 or 905 in kind of next generation devices we’re building. But you use two different wavelengths. And the fact that you’re using these two different wavelengths is what the spectroscopy part is about. And so you’re interrogating the tissue with two different wavelengths, so two different parts of the spectrum. And why this is important is because there are absorption curves for the hemoglobin molecule when it’s either oxygenated or deoxygenated. And those two curves are different and they have different trends and one goes up as you increase wavelength and the other goes down as you increase wavelength and they cross right at 800 nanometers in the electromagnetic spectrum. So if you pick a wavelength on either side of that crossing point, that’s called the isobestic point, then you can get a ratio of how much oxygenated versus the oxygenated hemoglobin you have. So you can use this spectroscopic measure to determine changes in blood oxygenation because you can you can differentiate between the oxygenated versus deoxygenated hemoglobin. So that’s kind of the fundamental thing that we’re measuring. That’s the near-infrared spectroscopy part.
Ryan Field: The time domain part is that instead of just sending in a continuous amount of light so you can think of what happens when you put a flashlight over your hand, you get this kind of like just continuous glow. And that’s very similar to what we’re doing. We’re injecting light into the tissue and its scattering around through that tissue. And we’re sampling where it comes out using our detectors. Instead of doing that flashlight approach, we use a very short pulsed laser and then we measure how long each photon takes to travel through the brain or through the head, I’ll say, because it’s not just brain that we’re going through. And what that time information tells us is that if the photon goes in and comes out very quickly, it couldn’t have traveled very far into the head. So it’s most likely from a photon that only traveled through the scalp or the skull. So a very superficial layer of tissue. But photons that travel deeper have a certain probability that they’ve actually passed through the scalp, the skull and the brain. So we can use this very precise, and we’re talking on the order of nanoseconds or one billionth of a second, measurements that tell us how long photons have been in the head and use that information to more precisely localize where these changes in hemoglobin concentration are happening. Because one of the biggest challenges with near-infrared spectroscopy is your scalp. We know like there’s scalp, there’s skin, and there’s blood flow there and you don’t want to be measuring just changes in blood flow in the scalp. That’s not very interesting. What you want to be measuring are changes of blood flow in the brain. I should say, blood oxygenation, not flow. So we’re measuring changes in blood oxygenation in the brain, and we can use that time domain information to better localize where it’s coming from.
Harry Glorikian: So, you know, this, you know, begs the question of, you know, do bald people get better, you know, results than people with hair or does skin tone cause any difference in this? Like you mentioned, you keep mentioning pulse ox, and we know that that’s an issue there. So I just, begs the question.
Ryan Field: Yeah, it’s a great question and very happy you asked it. So, yes, bald people get the best signal and so you’d be an excellent user.
Harry Glorikian: Thank you.
Ryan Field: And, and the reason is that hair is opaque. And so we’re using light to measure the brain. And any time you do that, you have to first get that light to the scalp. And as long as you have hair in the way and even the follicles, you kind of have to work a little bit to get through the hair into the scalp. And so we’ve done a lot of work on the mechanical engineering side to build devices and mechanisms to help us get closer to the scalp and get through the hair. So there’s an engineering problem of how to easily get through the hair. Right now, what researchers will typically do is they’ll just one by one go in and move, move the thing around and get it in close to the scalp. And so we wanted to build a system that’s a little bit easier to use. So we’ve been designing mechanisms that help us with just like kind of we call it doing the Kernel Flow head massage. You just kind of like rock the headset back and forth on your head and it combs through your hair and you get a little scalp massage as you go in. And so we’ve been improving that over time and have gotten better with our new version that’s coming out early next year. And it just takes a lot of trial and error and experimentation to kind of work through the hair. So hair is a challenge for all optical systems.
Ryan Field: And I’m glad you brought up skin tone as well, because I’ll touch on that, if you don’t mind, before we move on. So skin tone is interesting. And you’re right, that pulse oximeters do have a problem with the darker pigmentation of skin kind of systematically overreporting oxygenation. And part of that is because those pulse oximeters are based on what’s called a CW-NIRS or continuous wave NIRS approach. So you can think of that as a flashlight, and that flashlight is just getting modulated, its intensity is getting modulated by the darkness of what it has to pass through. With time domain NIRS, we’re a little bit resilient to changes in skin tone because it’s a very superficial layer where the melanin is. So if you think about from the outside of the scalp down to the brain, we’re talking about r about 25 millimeters and the pigmentation is only in the outer layer of that. And we’re measuring these time of flight photons. So only the very short photons are affected by just that. But the longer the photons travel, a bigger portion of their time spent is traveling through the tissue we care about and not the skin pigmentation area. So yes, it’s a challenge with traditional devices, but we actually think our technology does a good job at solving the skin tone problem. Hair, as I said, it’s just it’s a challenge for everyone who’s trying to get light into the head.
Harry Glorikian: Let’s pause the conversation for a minute to talk about one small but important thing you can do, to help keep the podcast going. And that’s leave a rating and a review for the show on Apple Podcasts.
All you have to do is open the Apple Podcasts app on your smartphone, search for The Harry Glorikian Show, and scroll down to the Ratings & Reviews section. Tap the stars to rate the show, and then tap the link that says Write a Review to leave your comments.
It’ll only take a minute, but you’ll be doing a lot to help other listeners discover the show.
And one more thing. If you like the interviews we do here on the show I know you’ll like my new book, The Future You: How Artificial Intelligence Can Help You Get Healthier, Stress Less, and Live Longer.
It’s a friendly and accessible tour of all the ways today’s information technologies are helping us diagnose diseases faster, treat them more precisely, and create personalized diet and exercise programs to prevent them in the first place.
The book is now available in print and ebook formats. Just go to Amazon or Barnes & Noble and search for The Future You by Harry Glorikian.
And now, back to the show.
Harry Glorikian: So. I’m sure everybody’s asking this question at this point because I keep going through it in my head is, what are the possible at home applications for Kernel Flow? Like, why would anyone want to measure brain activity of their own? I’m not sure my wife would want to know what my brain activity is, or maybe she would at different points. But you see what I’m saying? I mean, what do you do with this when you have it at home?
Ryan Field: Yeah. So we we don’t really know yet. And we have a few ideas and we’re testing some things out. And part of the reason is the science. Like the scientists haven’t had a device like this to gather kind of the richness and volume of data that is really needed to build an at home application. Like you need things that work at home to work robustly for everyone, right? Some things that we kind of imagine, like you could, for instance, say this is hypothetical. I’m not saying we’re we’re working on this or this is a product we’re putting out there, but you could do some kind of cognitive testing at home very regularly. So just looking at how your prefrontal cortex, the part of your brain that’s responsible for cognition, is functioning over time. So if you could get a measure of that, you could say, Oh, I’m starting to observe changes in my cognitive function as measured by my brain. Maybe my behavior doesn’t show it yet because I can compensate for it. My brain can just work a little harder and I can be I can still do my Sudoku and my crosswords and like all the other things that I do to keep like, read my books and everything, that kind of keeps your brain active.
Ryan Field: But you wouldn’t know that your brain is working harder until you couldn’t do those things as well anymore. And so there’s this possibility that you could use it at home for kind of a continuous wellness check of the cognitive function of your brain. So it’s something that comes in as you think about aging or how our brains change over time. You could use it also maybe for learning, right? So it’s like when is my brain in a state where I am primed to absorb new information, right? Like if I want to sit down and read this very dense book on economics or something, is my brain ready to receive that information right now? Or am I better off just reading that fantasy book that I’ve been setting on the shelf and want to go to a different world and disconnect? So there are kind of all kinds of ways that I think eventually neuroimaging could be used in the home and could be used to help individuals understand more about themselves and how best to do things or spend their time or achieve the goals that they have or for their own personal reasons.
Harry Glorikian: So now you guys announced in May of this year that you’re collaborating with a digital therapeutics company called Applied VR to measure the effects of their VR therapy, which is for chronic pain such as lower back pain. I guess, you know, what’s the hypothesis that Applied VR is trying to use your technology for? Is it possible that a VR headset can lead to long term changes in brain activity that could reflect lower pain levels or greater tolerance for pain? And so how does the data from measuring your brain activity and your system help prove this case?
Ryan Field: Yeah, that’s a great question. So we love the guys or the team at Applied VR. They’re fantastic and they’ve built a product that’s actually gottenrr FDA clearance as a treatment for chronic pain, chronic lower back pain. So they are using this as a therapy, as you mentioned. And one of the questions that always comes up is why? Why this VR system? Why why do we care to use VR system versus other forms of treatment? Or like, why couldn’t I just sit down in front of a computer and do a similar exercise and get the same effect? So really what we’re looking at there, without getting into too many specifics before it’s done, is we’re looking at change in the brain over time. So is there a difference between before treatment and after treatment? And we’re also looking at what happens to the brain during the VR session. So we have a custom version of Kernel Flow that we built that allows us to use a VR headset and record data at the same time. And that’s that gives us a little bit of an understanding of what’s going on in the brain while someone’s using this VR therapy. And I think Applied VR, it’s a great application and like treating chronic pain is such a difficult thing to do and anything we can get to understand that better is fantastic. back pain, but you asked this question of like what’s actually going on in the brain during X.
Ryan Field: I also think it’s a good model for what Kernel could do to help other companies in similar spaces. It’s not necessarily just for chronic lower. So like, I have a new therapy, like what’s going on in the brain during the therapy? Or I’m doing a series of mindfulness exercises. What is going on during those mindfulness exercises? Am I doing good or bad? There are all kinds of questions that people could ask you. You could also think about it from an advertising perspective, right? It’s like, what’s going through your brain when you watch this advertisement, right? Is it a positive response or negative? Are you annoyed that you have to watch five ads right now? Right. Like, what’s going on? And so I think you can really start to ask some questions that are more at the core of what our perception of the world is, and get away from just relying on what we can verbalize. One of the challenges is like if we go back to chronic lower back pain, I say to you, like I have a chronic lower back pain and it’s like an eight today. But what is an eight mean? It’s my personal scale. And how do you quantify that? And is my eight today the same as my eight yesterday? And like, how does it change over time? So are there ways we can get away from kind of relying on ourselves to subjectively report what we’re feeling, experiencing or going through and use that more precise information to build better therapeutics or better products that help people more?
Harry Glorikian: Oh yeah, Pain is one of those horrible, you’re trying to come up with that objective scale that you can do a study on. But. So you guys, I mean, earlier this year, Bloomberg wrote a piece about you guys, about a small experiment you guys did inside the company where the founder, Bryan Johnson, wore the headset before, during and after having ketamine administered. Right. That’s a psychedelic drug that’s increasingly being used to treat depression. The results from the experiment were presented in the form of some diagram showing levels of what you’re calling functional connectivity between brain regions. First of all, what is functional connectivity like? How do you quantify it? Why is it important? You know, the experiment on him showed lower functional connectivity starting about 15 minutes after ketamine was administered. It persisted for several days after the ketamine dose. You know, what is what do you guys think that signifies? You know, what what message should people take away from this?
Ryan Field: Okay, so I’ll start with my high level explanation of what functional connectivity is, and then I’ll dive into a little bit more about the study that was done. So first, how I like to think about functional connectivity is, I mentioned early on we’re creating a map of oxygenation in the brain and how it changes over time. And so if you kind of replace the brain with the globe, the world, and you have a map of the world and you use certain key points of the world, we’ll say major cities, and you look at the airports in those major cities, then you look you take this map and you say on right now, this point in time, what are the routes of airplanes flying between those airports and major cities? And you can tell where the really strong connections in the world are. You’ll see like a lot of traffic between London and New York. You’ll see a lot of flights going into Dubai as they go on to other regions in the east or Southeast Asia. And you’ll see flights going over the Pacific everywhere. And that tells you really how strongly connected these different parts of the world are. And that’s really what functional connectivity is, where the brain is, wr re’re looking at correlations between different regions of the brain. And when they move in sync, that’s showing there’s a strong connection between those two regions of the brain. So it’s like a lot of air traffic between those two.
Ryan Field: And that’s how I think about it, right? I’m an electrical engineer. I’m not a neuroscientist, but this is my my understanding of how we kind of look at activity patterns in the brain. These maps of activity in the brain are really looking at the connections between different regions and how they move together. So that’s that’s a visual picture of what I have going on in my mind. And then if you think about how changes tell us something about the brain. A few years ago, there was a volcanic eruption in Iceland. And during that eruption, the flight patterns between New York and London had to change. You couldn’t fly through the volcanic ash. And so you could see that represented by how the connection between London and New York changed. If you think about 9/11, you know, two decades ago, on that day, all the planes were grounded. So you could see that there was just like this major disruption to the system. You don’t know exactly what it is just by looking at the plane maps, but you can tell that something major changed. And that’s really what we’re looking for when we look at changes in functional connectivity. It’s,rr can we do something to change the way that different parts of the brain are connected? And can we start to build up data that explains what those changes mean? So is that a good foundation for what functional connectivity is? Do you have a good picture at least?
Harry Glorikian: No, no, I got it. You know, now it’s translating it into. What does that mean? What do I do with that? And I mean, it seems like you guys are trying to help, you know, psychedelic—I’m just assuming that that’s a play for the company to be able to assist in how these therapies might be impacting these patients.
Ryan Field: Again, it’s a play to try and quantify how these therapies are impacting patients because every patient will self report like “I feel better today” or “I had a an intense psychedelic experience.” And these are all kind of subjective reports. And the question or the hypothesis is could we measure the brain during the psychedelic experience and see how big of a change happened during that? So we can start to gauge how intense that psychedelic experience was. And then, two, can we use that information to predict the change that will come over time? So if we go to this study result—I’ll qualify all of this with, this was a pilot, single participant, that was done as a lead-in to a bigger study that we did with a partner, Cybin, who sponsored a 15-person study where we were looking at the ability to measure one’s brain during the use of ketamine. We’re actually going to be publishing that work probably early next year. We’re going to submit it before Christmas. But publication on that work will come next year sometime. And this pilot result that was done with just one person, we said, hey, we’ve got all this infrastructure set up. What happens if we do these measurements, these longitudinal measurements over many days and see what happened, what’s happening in the brain before the ketamine session and after it.
Ryan Field: And so what you saw was a representation of the magnitude of connection and what’s called the default mode network. And the reason we looked at the default mode network is because that’s one that’s, through scientific literature, and a lot of academic work has been associated with things like depression. So it’s like this this network that gets involved with depression. And that’s one of the hypotheses for psychedelic treatment is that, if you can use psychedelics to change the connectivity of this default mode network, you could help treat depression or the treatment of depression could be manifested in that. So again, I’m not a neuroscientist, so I’ll just qualify that. Go do your own fact checking. But that’s roughly the concept here. And so what we saw is that before the ketamine session, there was a pretty, pretty good stability in the default mode network. And then during the psychedelic session, it just oscillates wildly. So there’s a lot of stuff going on in the brain during the psychedelics. And then afterwards, there was this reduction in activity in the default mode network that persisted for a few days.
Ryan Field: And again, this is all kind of an N=1 thing. So we would need to do a bigger study that’s similar where we do these repeated measurements. That was not part of the 15-person study we just looked at during the 15-person study. So it was an interesting kind of, I want to say, just a single data points that suggest that maybe there’s something more there that we should be looking into and seeing if there’s a path to kind of do that. The challenge with doing these experiments with psychedelics is they’re expensive and complex. So we really need partners who are interested in solving the same problem and have a financial reason for wanting to solve that problem or understand it more deeply. And it’s hard for Kernel tor get out there and do this all ourselves and say, hey, we’re going to go out this alone and we’ll tell you all the answers you need to know about psychedelics. But we’re really interested in working with partners, especially in the psychedelics space. It’s a very hot space right now and so much potential around treating a lot of mental illness and other disorders. So it’s I think it’s really promising. And, you know, we’re always excited to talk with and work with potential psychedelic companies.
Harry Glorikian: You guys also work with gamers to see what’s going on in the brain of, say, expert players at, you know, first person shooter games like Call of Duty. What are you guys learning from a study like that? Are the brains of expert gamers different than those of the rest of us?
Ryan Field: Yeah, so it’s a great question. That line of work started really as just kind of a fun exploration. And it was, we had a connection to a Call of Duty world champion who goes by the handle Scump, and we said, Hey, can we come out and measure your brain while you do this specific training game? So it was a game called Gridshot that we had partnered with the company Statespace on kind of adapting for use with Kernel Flow. So we did this measurement and did this whole kind of like promo piece where Bryan and Scump looked at, we looked at their brain activities. And again, these are N=1. So it’s always like, Oh, there’s a difference, but what does it mean? You need a little more data to say like, this is what it means. So then what we did is we said, Hey, we’ll just recruit like I think almost 75 people from the L.A. area who had varying levels of gaming experience to come to our offices and do the same thing. And then we separated them by high, high performers and low performers. And we looked at differences in how their brain activity was between the group of high performers and the group of low performers.
Ryan Field: And in that work, we did find that there was a difference in specifically the prefrontal cortex area in good versus bad gamers—it’s hard to talk about being a gamer, I guess if you’re having fun, you’re being a good gamer, in my opinion, I’m no professional gamer here—so it was interesting. We saw something there, but, you know, it was just like a fun question to ask because like, is there a difference in the brains of an expert gamer versus not? You could imagine how that that kind of sets a foundation for a lot of other questions you could ask. It’s like what sets a really good surgeon apart from an okay one, right? Like what? What characteristics of quarterback in their brain set it apart from an OK one, right? Like who’s a starter versus a third string quarterback. You could start to ask all kinds of questions like this around brain measurement. So you can really try and understand things about how people form strategies or make decisions and respond to stress. There are so many different things that you could kind of unpack from this gaming example and do a more detailed study.
Harry Glorikian: Yeah, like right now during the World Cup, you’d love to put that on Messi’s head and see how he’s looking at the field if what he’s going to do next. Right? But one of the underlying themes of this show is, you know, how is machine learning or other forms of AI giving us insights into all kinds of health data. And so can you guys. I mean, there’s a lot of data I believe, that’s going on in this product of yours. So, you know, what type of analysis can you do with your product? Do techniques like machine learning come into the picture at all? Or, you know, are you at the point where you have enough data to be able to look at the movie that is being produced, you know, that that’s being captured by the system and classify the activity or make a prediction about it, for example? You know, I don’t know. Could you look at the recording and say, “That person was probably taking ketamine” or “That person is playing a video game” or maybe “Is that person angry or anxious or depressed”?
Ryan Field: Yes. So we are just now getting to that point with our datarset sizes where we can start to look at some of these things. So we have looked at in one particular case, we built a classifier that classifies focus. So we looked at can we tell from the measurements of someone’s brain whether or not they’re focused? And we used a task, a test kind of, where we asked someone to do a driving simulator on a straight road through the desert for something like 15, 20 minutes. And we put up different things to kind of interrogate if they were focused or not, and then use that information, that behavioral information to build this classifier of what does a brain that’s focused or unfocused look like? And so we started to take that and apply it to real time algorithms and processing so we can, in real time give a presentation of are you focused or not. And it’s a nice demonstration of kind of what a brain interface can tell you about yourself, like a brain state. Again, kind of differentiating from kind of thought-to-speech or thought-to-cursor movement, but really measuring a brain state, something about how my brain is at this moment holistically. Is it focused or unfocused? And again, you can kind of think of a million different applications of something that can reliably tell you or tell if someone is focused or not. And this is just an example. This isn’t this isn’t a product we’re building. We’re not building a focus device. It’s just an example of a kind of highlight, what we can do with data, how we can translate it into a real time result, and how we can give that feedback to an individual as they’re wearing the device or after they have done a, you know, a session of something.
Harry Glorikian: Okay. Where are you guys—-I think you’ve said a couple of times about having a product early this year, but where are you in terms of commercializing the Kernel Flow? You gave a presentation at the IEEE Brain in November of 2021 where you said you plan to ship 75 Flow devices to developers this this year. I mean, have you guys gotten to that goal? And what’s next if you have.
Ryan Field: Yeah. So that 75 device system I’ll show you here for anyone who’s looking at the video. Right. So this is our system. You’ve seen this a lot online and this is what we built, the 75 units of, and srent it out to a number of partners and academics to get their hands on an early version of the system. So this is the first prototype we ever built. We call it Flow 1, and it’s the core technology. And it was our big learning ground to understand kind of what the strengths and weaknesses of the system were so that we could improve it before we made our first production release. So I won’t show you what our first production release looks like yet, but it’s going to be released early in 2023. So we’re targeting kind of end of February, early March, and that device is a huge improvement off of this one. We get better spatial resolution and coverage of the brain. We have higher sensitivity, we have lower power and all these other things. And we built a much nicer, sturdier, stronger headset. And the target for these devices is really the early scientists and developers. I like to say, rthe explorers of the brain, the ones who want to go out and ask questions and understand what the brain is doing under certain conditions so that they can then kind of build that into the applications. So again, I’ll use a bad analogy just to give a high level thing. What I think we’ve done is we’ve kind of built the iPhone, but we didn’t build any apps on it. The only app you have is a map. And so you get a map of the brain and you have to use an API to build an app on top of that map.
Ryan Field: And so it’s a platform that could be really versatile and used in so many different ways because our brains do so many different things. It’s just understanding what that map means in different contexts. So that’s, that’s how I like to think of kind of Kernel Flow and the stage it’s at now. Two years ago when we launched and kind of announced the first prototype version, we did set a goal and we said by 2033, so a little more than a decade from now, our objective is to make it so that Kernel Flow could be in every home. And that’s a that’s an objective from a hardware company building devices. But to get that out there into every home, you really need to answer the question of what you do with it. And this is what you opened with. Why would I want a neuroimaging device at home? What could I do with it? And my belief is that anything that you and I could think of today is probably wrong. So if you look at the iPhone, when it launched, it was like a phone device. SMS, email, like New York Times had an app, I think. There was like so many limited things. If you look at like the number one app in the App Store today, it’s TikTok. Who would have imagined when the iPhone launched that a decade later the most used, the thing that it’s most used for, is watching short clips or videos of people dancing and doing other things. So I think it’s hard to predict what types of things a transformative technology will ultimately be used for and where people will find the best uses for it.
Harry Glorikian: Yeah, that’s great. I mean, actually, it’s funny because you went right into my next question or series of questions, so you answered it, which is cool, but you know. Great to learn about the technology. You know, really interested in understanding more at some point of like, what does this stuff really tell us, how do we decipher that information, and then how do we put it to that practical use either for people at home or almost more important for, like you said, drug development and, you know, even digital therapeutics, because there’s a lot of digital therapeutics that are sort of coming onto the scene or want to come onto the scene. And I think people need to understand that if it’s neurological based therapy that these digital therapeutics can have a profound effect on that patient and have a clinical, positive clinical outcome. It will be interesting how this technology gets used there, But, you know, great having you on the show. Hope I covered the gamut of the technology and what you guys can do with it. And, you know, I wish you guys, you know, huge success.
Ryan Field: Yeah. Thanks so much for having me. And hopefully I can come back in a year or so and tell you all about the things people are doing with it once it gets out there and has a little bit more time to have these applications evolve in the market. So I think it’s exciting. I think we’re just at the beginning and there’s a lot more that will be coming out in the future. So.
Harry Glorikian: Cool.
Ryan Field: Thank you.
Harry Glorikian: Thank you.
Harry Glorikian: That’s it for this week’s episode.
You can find a full transcript of this episode as well as the full archive of episodes of The Harry Glorikian Show and MoneyBall Medicine at our website.
Just go to glorikian.com and click on the tab Podcasts.
I’d like to thank our listeners for boosting The Harry Glorikian Show into the top two and a half percent of global podcasts.
To make sure you’ll never miss an episode, just open Apple Podcasts or your favorite podcast player and hit follow or subscribe.
And don’t forget to leave us a rating and review on Apple Podcasts.
We always love to hear from listeners on Twitter, where you can find me at hglorikian.
Thanks for listening, stay healthy, and be sure to tune in two weeks from now for our next interview.
There are a variety of reasons why researchers might measure brain activity. Some common reasons include studying brain function and dysfunction in neurological and psychiatric disorders, understanding how the brain processes information and controls behavior, and developing new treatments and therapies for brain-related conditions.
Additionally, measuring brain activity can help researchers understand how the brain develops and changes over time, as well as how it responds to different stimuli. Overall, measuring brain activity can provide valuable insights into the workings of the human brain and help advance our understanding of the brain and its disorders.
fNIRS (functional Near-Infrared Spectroscopy) is a method of measuring brain activity by shining near-infrared light into the skull and measuring the amount of light that is absorbed by the brain. When light shines into the skull, it is absorbed by the blood and reflects off the skull, brain, and other structures. By measuring the amount of light that is absorbed by the brain, researchers can infer changes in brain activity.
fNIRS is a non-invasive technique, which makes it well suited for monitoring brain activity in infants, children, and adults who may not be able to tolerate more invasive techniques such as fMRI. It is also portable, and can be used in real-world settings like naturalistic observation, and mobile settings such as ambulatory and even wearable devices.
fNIRS has been used to study a variety of brain functions, including cognition, emotion, and motor control, as well as brain disorders such as stroke, traumatic brain injury, and Alzheimer’s disease. It has also been used to study brain development in infants and children and to monitor brain activity during surgery.
In general, the methods used to measure brain activity, such as fNIRS, are considered safe. Unlike other techniques such as fMRI, which uses a strong magnetic field, there are no known risks associated with exposure to the low-level magnetic fields used in fNIRS. Additionally, the near-infrared light used in fNIRS is not harmful to the eyes or skin.
However, there are some potential risks and considerations to keep in mind when measuring brain activity. For example, some people may experience discomfort during the procedure, such as feeling claustrophobic in the fMRI scanner.
Ryan Field is the Chief Technology Officer (CTO) of Kernel. He holds 3 US patents and does research & development involving integrated circuit design, sensors, system architecture, firmware, software, and imaging devices for automotive and other LiDAR applications. He is also the guest of today’s episode of The Harry Glorikian Show.
Brain activity refers to the electrical and chemical processes that occur in the brain to produce thoughts, movements, sensations, and other functions. These processes are generated by the firing of neurons, which are the basic building blocks of the nervous system.
Both EEG and fMRI are widely used techniques for measuring brain activity, but they have different strengths and weaknesses, and the choice of which to use will depend on the specific research question and the population being studied.
EEG (electroencephalography) is a non-invasive technique that measures the electrical activity of the brain using electrodes placed on the scalp. EEG has a high temporal resolution, meaning it can detect changes in brain activity very quickly (on the order of milliseconds). This makes it well suited for studying rapid changes in brain activity, such as those associated with sensory processing and cognition. Additionally, EEG can be used to study brain activity in people who may not be able to tolerate more invasive techniques, such as infants and children.
fMRI (functional magnetic resonance imaging) is a non-invasive technique that measures changes in blood flow to different areas of the brain, which can indicate areas of increased neural activity. fMRI has a high spatial resolution, meaning it can produce detailed images of the brain and identify specific areas of activation. This makes it well suited for studying brain structure and function, as well as for identifying brain regions associated with specific cognitive processes.
In summary, EEG and fMRI are two different techniques with different strengths and weaknesses. EEG is more sensitive to temporal resolution, and fMRI to spatial resolution. The choice of which to use will depend on the specific research question and the population being studied, and sometimes both techniques are used in conjunction to provide a more complete picture of the brain activity.