Constructing  Brain Maps with Machine Learning - Dr. Stephan Doyen

Talking Biotech Podcast 397, Hosted by Dr. Kevin Folta
===

And today's guest is Stefan Doyen. He's a founder and chief day scientist at Omniscient Neurotechnology. So welcome to the podcast, Dr. Doyen. Hi Kevin. Thanks. Good to be here. Hi. Yeah, this is really, uh, a little bit outside of my normal focus. Like I, I'm, I, no d n a inside and out, but I've always stayed away from two things.

It's brains and immune systems. What your company brings to the table maybe is the best thing in the world to help me start to understand this. So the brain is complex, but it's organized in these specific neighborhoods that. Correlate maybe to some degree with function. So why is it necessary to think of brains in this par uh, form and that specific functions belong to specific neurological neighborhoods?

Well, you, you, you hint upon the exact term, which is complexity. And if we think about, um, the com, the company I co-founded with, um, a neurosurgeon, um, named Michael sru. The initial need actually came from the field of neurosurgery. So we didn't, um, start the company because we thought we're gonna create brain maps.

We started the company thinking, well, there, there's, there's an issue within the field that we can probably help fixing. And so the, the, the need that neurosurgeon had was to understand better. Um, the, their patient's brain, as they were to, um, prov, proceed with surgery. And so as a result, sometimes they would proceed to one approach and get a great outcome, and sometimes they would proceed to a similar approach to a patient and get a, an outcome was that was poor.

And, and the difference between the two, um, situations were an information gap, and that's where we thought we could fill that information gap with a map. So from there, um, we, we, we started using those maps and we found that there was a further opportunity, which was essentially organizing the brain in a, um, a set of functional areas.

And we can talk a little bit more about this, um, further, but essentially looking at the brain in discreet parts. All of which underpinning certain functions of, um, your, your, your behavior, your thoughts, your memory and everything else. Well, this makes a lot of sense, but if I'm going to develop a map of a town, I'll start with a satellite image that I can, you know, look at and start to draw on to, to figure out what goes where.

But how do you do this with the brain? Is it from imaging of the brain? Yeah, the, there, there are multiple, um, maps and Atlas out there. It's, it, it is actually, um, um, a similar analogy in the sense that if you were to go, um, somewhere, well you, you, you'd be looking at a road map, which would tell you the way to where you want to go.

Uh, if you were to climb a mountain, you'd be looking at a different type of map, which tells you. Uh, the elevations and, and, and, you know, the water streams and the pathways and such. Well, the, the same, the same applies to the brain. There's multiple ways of, um, describing the brain. The, the, the, the one we are using is a derivative from, um, a project that was an n I h uh, fondant initiative called the Human Connector Project.

And so what the Human Connectome Project, and there's a. A famous, um, paper by GLA Glasser Glasser in 2016, uh, in nature is basically providing, um, a functional map of the brain, a functional being that it segments the brain in discrete parts, all of which have a specific function in contributing to, uh, how the brain works overall.

Yeah. So the Glasser Atlas, this was, um, all MRI based, wasn't it? And, and or was it a functional connection with mri? It is a, uh, what, what they call a multi-model, um, Atlas and, and, and maybe a little bit of context about MRI and, and what you can derivate from this is, um, generally speaking, the way an MRI function is, is.

Essentially a, a big magnet that will align your, your hydrogen, uh, protons. And then as the magnet stops, it just releases them. And, and as a result, some energy is released in, in the space that is then captured by a sensor. And with this you can start to, um, build, uh, the, the, the movements of molecules of weather.

And, and another thing, depending on some of the, the frequencies you can use with your, your magnet. And so what Glasser and team did was using this MRI in different ways to produce different data. And so there's, there's three main types of data that we are using, um, and some of which that Glasser used as well.

One is called a, uh, a T1 or a t2, gives you the general anatomy of the brain, so you can see the, the structure as a picture. If you, if you were to dissect the organ, Another one is, uh, diffusion sensor imagery. So this one is specifically, Um, targeted at moving the molecules of water in the brain and reconstructing the white matter pathways.

And, um, we can talk a little bit about how it helps building this, the map, um, a little bit later. And then lastly, there's a, what we call a functional mri. So the way this this works is you, you, you put a subject in the MRI machine and then you have them doing a task, or sometimes you have them doing no task, which is called a resting state mri.

And what you want to do by this is to see the brain in action. And what happens is when neurons are firing, they essentially send a message to the adjacent vessels, um, which will call for more blood supply. And that blood supply is bringing some, um, iron and iron can be captured by this, this magnet, uh, which is the mri.

And so that's called a functional mri. So the, the Glasser. Uh, work uses, um, what they said to be multimodal, meaning that they're using, uh, obviously the structure, anatomical, but also the, the functional and in their specific case, they is using at a, as a task base. So they're asking people to do things and why these people are doing things well.

They see what part of the brain lights up. That's essentially how to date it. Yeah. Okay. Well that makes a lot of sense. I never knew that it was iron, that an mri, functional MRI was keying off of. I Is there, are there metabolic? Ways of testing for function, or is it really just a question of blood flow?

It it, it is, it is mainly a question of blood flow. I'm not too deep on the exact components that are being used, but the, the, the method is actually called bold for blood, oxygen level dependence, uh, response. So it's, it's, it's mainly a blood flow based technique. And I guess this is maybe a easy question, but how does knowing exactly where what is happening in the brain really help us have more effective therapies or even surgeries?

So if you think about it from a, um, neurosurgical standpoint, and I want to cave this because I'm not a neurosurgeon. My, my co-founder is, but I look at it from more of a technical AI standpoint, but from the eyes of a neurosurgeon, when, when you're, um, In surgery, you, you have in some cases what's called a, a guidance system.

So the guidance system essentially is a screen, uh, that is synchronized with the patient's head that is, um, in a frame. And so using a, a, a probe, you can actually navigate the, the MRI on the screen. Whil, you actually navigate the patient's brain. So it's actually a very nice piece of technology that's now quite widespread.

The problem though, with those systems is what you're seeing is an anatomical MRI now. Anatomy is useful in the sense that there are certain landmarks you can use to navigate the brain. So you know that if you're on the left hand side around your, your, you know, the, the, the temporal lobe and, and the mid side, you're gonna start hitting, um, things that are related to language or, or you know, if you're more the middle in tricity motor skills.

So, you know, these, these are known landmarks. Now there's things that are far more subtle. And the things that are more subtle are, um, for instance, networks. So one of the thing we do with the, the companies, we map networks and brain networks underpin certain function. There's a, there's a famous network we, um, we, we we studied, which is called a default mode network.

And the default mode network is essentially you're inner speech and serves all sorts of functions. That is invisible. You, you, you don't see it. There's nothing in the brain that tells you if you cut here, you're gonna cut through the default mode network. And if you cut there, you're not gonna cut through it.

But if you do, you're actually gonna profoundly alter the patient's behavior. Um, more often bring to, you know, consciousness and, and conscious awareness issues. Um, that's, that's invisible. So from guidance system. Um, which is a great plus. You also need to bring a functional view of the brain so that you're better informed when you're taking your, your surgical approach.

I see. So this makes sense, but are all brains pretty much the same anatomically, or, you know, people who are looking for work on a brain or, you know, for surgery or maybe something, typically have some sort of anomaly, which may alter structure or, uh, or make things a little bit different. So does that matter?

It. It does, absolutely does. So from a macro, cortical, and subcortical standpoint, Most humans are the same. And part of the reason is that we, we grew in an environment which had similar constraints. We generally have two legs, two arms, and we need to communicate and eat and so on and so forth. And so these things will constrain the development of the brain.

Uh, the brain wasn't creating a vacuum. It developed with the environments. Now will. The brain is also a organ that, um, reacts strongly to its environment and each and every single environment's. Our difference, our life experience is different. And so by this we introduce, um, slight differences. For instance, if you speak two languages or, or you know, if you had certain events in your history.

And furthermore, um, and that's more so the, specifically the case for, uh, doctors and, and, and neurosurgeons. In particular, most often their, their patient's brain is not looking like, um, you know, the brain you'd find of an average, uh, 20, 28 years old university student, which is the most common M MRI scan you'll find out there outside of hospital.

More often, there's a, there's a tumor, there's a, there's a blood clot, there's, um, a previous surgery. There's something that completely altered the anatomy of the brain. And this is a problem, um, we've actually solved, which is how do you map a brain that has an abnormal shape or that isn't quite like you'd expect?

And we, we use machine learning for this and, uh, we, we, we published about this, um, it's a technique called, uh, structural connectivity Atlas. And grossly speaking, the way this works is that it, it, it uses, um, A D T I image, which basically is a mapping of the molecules of water in your brain, which is a proxy for mapping the white matter pathways and the white matter pathways as what is connecting each and every single.

Part of your brain together, and the way your brain is connecting itself through the various parts is actually informative about its function. And I, I explained the, the, the example of, um, language before, right? So you'll find that the language network connects with itself, or the default mode network connects with itself.

And sometimes parts that are unrelated won't connect as much or have thinner. Fibers that connect through them. Well, we use machine learning to actually make sense of this, and by doing that we can take that h e p atlas and we can tailor it to each and every single case, including, uh, gross abnormality in the brain.

Yeah. So if you're using this kind of approach, which is using this structure function, connection to, and machine learning to identify these networks and these pathways to make the maps, how is that different from, uh, what others have been doing so far? Right. So you, you, you mentioned a very important distinction, which is the, the structural and the functional and, um, side of the brain.

And it's something we can, we can explore as well. But the idea is that your brain is structured in a way that it's supports its function. Now there's a, there's, there's, there's overlaps there, but it's, it's essentially the idea that. What, what? Fires together. Wires together. That's a famous, um, principle from Candle was a.

Uh, no noble prize, um, for a study on the Aplasia. And so the brain organized in, in, in a way that what fires together, wires together. And so these create pathways that you can then detect using. Um, machine learning, right. So from there, how does that differ from, uh, what already exists out there? So what is available to doctors before we existed, what was available to doctors before we existed is, um, the anatomical image.

So you can have an mri, you can apply different contrast to that image, which is quite helpful. And you can also extract the white matter pathways. So there's, there's software that used to do that. Now what they didn't have is, uh, something that makes sense of those pathways. So you're basically seeing a picture, which is, um, you know, bushy fibers, colorful fibers, um, very, very visual.

Um, but knowing that this bundle of fiber does this versus that one, and this one is probably more important in the context of the patient than that one. That information wasn't available. They, they did not have the functional maps. And so this is the, the key difference we are bringing, uh, to the field of neurosurgery that is this functional view of the brain.

They can overlay onto their guidance system to better understand the case. They've got to, uh, deal with. And you're always have been speaking today about neurosurgery and guidance and helping to, you know, get these things correct in terms of location. But how much of a diagnostic value does this have? Is it the kind of thing where someone may present to an emergency room with very specific, um, very specific symptoms and someone who's familiar with your technology and the maps can say, well, this is where the problem is right here ever before they go for CT or M r I.

Um, see th this is not why we, uh, intended to build the device for we we're, we are a decision support device, so what we are doing is we bringing additional information that the doctors wouldn't have otherwise in the context of surgical planning. Now, not to say, you know, if you have better information or if you have more information than you had before.

You, you may consider certain approaches differently, and that's something that, uh, doctors can consider for themselves. That is the use of functional, uh, maps to, to, to tailor their diagnostic and better understand it. But that's not something we, we've, we've intended to build a device for. However, this is certainly something we can look at for, um, future improvements and future applications, and some of which we're, we're, we're actually working on.

Yeah, I had to ask, it was something, uh, of personal interest because in 2012 I had a seizure just out of the blue, probably from dehydration and lack of sleep, uh, day figure. And when I went to the, uh, neurologist, I explained to her how it propagated and the places it started, and she said, took out a model of a brain and split it in half and said, well, you probably had your stroke or your, or your tumor.

Right here in this neighborhood. And turns out that it was none of the above, but it was, uh, seemed very strange that just from the, uh, symptoms that she, that I described to her, that she was very, uh, compelled to, to, uh, to describe the part of the brain that was affected. So I had to ask, but Yeah. Um, and, and you're, you're, what you're highlighting is that there's still a lot to be done when it comes to understanding the brain.

Um, and, and so using machine learning on brain data is, is, is probably one of the most promising avenue in the next, uh, decade. And so you find that still nowadays there's a lot of questions that we can't answer even with our system, which provides a quantum leap in terms of evidence, um, and information that a doctor can deal with.

There's still a number of things that we just don't know and, and that includes also, you know, mental health related issues and other, other type of problems. And so what you're highlighting there is, is, is is fairly common when dealing with the brains is there a lot of things we just don't know. And so working through, you know, getting better information out there to practitioner is something that is so important to improve brain care overall.

Very good. So we're speaking with Steon Doen. He's the founder and chief's data scientist at Omni Neurotechnology, and this is the Talking Biotech podcast body Collabora. And we'll be back in just a moment. And now we're back on the Talking Biotech Podcast by Col Collabora, and we're speaking with Stefan Doyen.

He's the founder and chief data scientist at OmniGen Neurotechnology. And we're speaking about brain maps and how brain maps are being revolutionized using machine learning and ai, and how they're helping sci, how they're helping surgeons. Navigate the brain, uh, and provide guidance during the complex surgeries that we understand happened inside the brain and, uh, and other applications as well.

So let's start out with this, uh, idea of the machine learning based method. So how is that used to infer connectivity? Right. So one important property of the map is that it organizes brain data. And that's a key problem in the field of neuroscience and was a key issue in being able to apply any kind of machine learning on the brain.

So let me, let me zoom out a little bit and expand some of the work that's been done this far on brain data. And this is really good work. But you'll see very quickly what the limitations are often, um, when data scientists and others were getting a brain scan, Their approach would be to borrow from some of the computer vision methods, methods called ational, neural nets, and other things that are, uh, specialized in looking at pictures.

And it makes sense, right? You, you, you're getting a scan, it's visual. You're thinking, well, I'm gonna apply visual techniques. And also doctors visually, uh, they, they, they, they look at the scan and so they find abnormalities and, and other things. And then they use those techniques. Well, that's, that's good.

That's helpful. That gets you somewhere, um, it, it highlights some informations you wouldn't have otherwise. However, um, there's some issues, right? One issue is the sheer complexity of the information you're extracting through those models. They're inside of a scan. They're several tower, thousands and hundreds of thousands of voxels.

So what sense do you make if you see that two voxels to the left and three to the right and somewhere around there there's another, a cluster of voxels. What sense do you make of that information? It's really hard and it is, some sense can be made if you overlap it with preexisting knowledge, but otherwise you, you struggle navigating.

So there's an information and complexity issue. The other issue with those techniques is that, um, You know, you're, you're, you're landing into highlighting certain voxels, so you've made sense of it very well. Um, how do you connect this back to the function? And that's a critical issue in the sense that you might be highlighting voxels that are sitting over, um, different parts of the brain that do different things, and so say, You've got a few VS that have been highlighted by your machine learning algorithm that sits on the, on the motor strip and the sensit strip.

So you can see that it's both the motion and, and the way you're feel in your body that are showing certain issues. How do you make sense of that? Which one of the two? It is it. They're related, but they're very different and so. We took a different approach. We took, um, the approach that in order to make sense of brain data, you have to pre-process it in a way that you can make sense of it after you use machine learning onto it.

And the way we've done it is basically by using those brain maps. When you do that, you take unstructured voxel base data and you turn it into structured ation base or parcel base information, and it constrains your output to, to first a a handful of meaningful units. So if you find that it's this area that has been flagged by your algorithm, Well, it, there's this fine finite set of units.

You're flagging RLA is 379 different constituents, and it's very manageable to say, well, it's parcel number 44, or it's parcel left eight av, or whatever their, their, their name is. And they're all, um, Available in the literature or on the website. Um, that's helpful in a way to better understand your data, but also it constrains your finding to these finite, finite areas.

So it removes a lot of the noise and it improves the quality of your finding. So that's an important thing to flag is that it mapping the brain is a real unlocker for using machine learning on brain data. Now, from there, the second question is, How do you use machine learning on brain data and then multiple ways you can use machine learning on brain data.

I've highlighted the voxel based method, which is extremely useful for triaging and, and, and helping human judgment. There's also another way which is, well, can machine learning tell me things I wouldn't know otherwise? Uh, given the complexity of the brain. And this is one of where we, we use it and we, we published about it quite extensively with our, our group and other research group, which essentially we start from a symptom.

Say you have a series of, um, patients with, uh, schizophrenia and, um, they're reporting, hearing, uh, voices, which is one of the symptom you can find in schizophrenia hallucinatory perceptions. Well, can we use a machine learning algorithm to map the outcome to an input data and then tell us what ended up mapping, got that algorithm to make that decision.

And so if you do something like this, and it's a, a technique, we, we, we, we've applied, we have, we call it hots. For whole of three. Super. So we, it's available online if you want to. It's been published in the, in the scientific literature. And so we, we are using this to basically understand what caused, not caused, sorry, what is associated with the symptom.

And then what you're revealing with this is brain parcels and brain networks that are topping up, um, some feature importance ranking. And in the case of auditory hallucination, for instance, in schizophrenia, we found that these were the auditory parts of your brain. So we didn't pre-wire this in the machine.

It's purely statistically driven. Uh, we also found that visual, as in hallucination in schizophrenia, are associated with anomalies in visual cortices. Now, again, we didn't prior prime the machine to do that. It actually found it, found it itself. And you think of this, this is actually very scalable. You can with the same set of idea, you can actually map a lot of symptoms, but you can also map all sorts of performance and if you have the right atlas for paring the brain as well as a sim sufficient sample size.

So we're talking about the several thousands of brain scans. Then you can start really finding markers of certain behaviors, certain symptoms, and certain traits. And that's where, um, the machine learning side becomes really interesting as to, um, how it helps to understand brain a lot better. And are there reproducible and, uh, Readily observable changes that occur in the map, which are oftentimes, or maybe predictive of diseases like disease states, like Alzheimer's or Parkinson's.

Yeah, so this is, this is a very new field, um, in, in neuroscience. And, you know, we're, the company I co-founded has been around for. Almost three years now, and we're a company that provides maps on on demand. So it is something that is quite news to be able to personate, um, the anatomically. Form brain. Um, as such.

So it's still a nascent, uh, field. However, um, there is evidence coming through and there's more and more, uh, papers that are published about this, that if you look at the structural connect home, so the way the brain is actually connected and, and to itself. And how it, um, produces the functional connectome.

So the way it all, it all works together, you can start seeing, uh, things that might be signs or precursors, signs of certain diseases. And again, this is all very new, but one example is we. We worked on, um, distinguishing cases of, uh, confirmed Alzheimer's diseases and, uh, mild cognitive impairments at a stage, um, where both of them were very much alike.

So the early stage. And so looking at these two things, we found that, um, uh, Alzheimer was displaying more of a profile of pathy. So what I mean by pathy is that in your brain there's certain areas that are highly connected. To, uh, other areas in the brain that we can call them hubs. They, they gather information from various networks or inside of a network.

And in the case of Alzheimer, we found that those hubs, um, were showing a, um, state of decay, which wasn't found. In, uh, the mild cognitive impairment. So this was a pretty, uh, striking finding. Uh, and, you know, we wish to have more evidence about this and we really encourage the community to, um, you know, look at, look at those things and, and, and try to understand the connection better and how it can be really used for these kind of, um, issues with the brain.

That makes sense. Uh uh, and that would really allow this kind of mapping to potentially be able to use to, to be used as a predictive tool, right? Like if you started to see, uh, specific questions around given hubs, it could almost maybe inform. Uh, patients with a different strategy or potentially medication that could act before the onset of symptoms, that kind of thing.

Yes. So it, it, it is very early days, but it's not too far-fetched to think induced terms. Um, you know, there's one advantage of using connectomics over genomics in a sense that if you were to use, um, you know, gene to predict that you've, you might have Alzheimer later. There's, there's always this gap, right?

Um, which I'm. I'm sure you, you, you far more astute on than I am. Uh, that is the, the fact that Gene can express or not. Whereas if you look in connectomics, um, you know, the expression has happened in a way and so that that marker can be, can be closer to your outcome. Now, obviously, um, between finding the mark and the outcome as you, as you point to, there's, there's probably a space which can be filled with.

Um, strategies and therapies to address them. So it does bring hope, but like I said, um, early days and we need, we need, we need more work on this. So we'll talk about early days, you know, what is the next phase of work that's happening in your company? So we're, we're really at an exciting time in the sense that, um, never had we access to so much brain data, um, In such quality and quality meaning, um, mapped in a way that's digestible for machine learning.

And so one of the thing we're, we're, we're working on very actively is to essentially look into mental rela mental health related issues. That's one of the big mission we have, which is improving brain care overall. And that transcends, um, you know, neurosurgery goes beyond that. Uh, there's issues we talked about, neurology and, and your personal experience before, but also psychiatry and, and, and even per personal health.

And so until you have the information, until you can't start to quantify and describe your brain better. You can't really manage it. It's quite striking that, um, you know, if, if you turn up with symptoms of depressions, um, depression, you, you, you're probably gonna get a blood test and, and, and a lot of survey and, and monitoring, but you quite unlikely to get a brain scan.

However, that information that, um, you know, you're, you're depressed, that your, your biological brain isn't functioning quite well, it's making you consist consistently sad is all, is all is all in your brain. So, with the right tools, including acquisition technique and and the right mathematical color, AI approaches, we should be able to unlock this.

And that's kind of what we are working on. And we've got some very promising way of doing, ways of doing that. One of which I mentioned, which is the, the hots approach, which gets us to find markers in the brain related to symptoms. Um, but we've, we've gotta expand this and we've gotta make this more digestible for a practitioner in the field.

So now, um, in a way, one day connectomics would be routinely used for any problems of the brain. And right now, how frequently are these techniques being used in actual surgeries? And are these really just in addition to what you can learn from MRI and that kind of guidance, or where are we now? So in, in, in surgery, the, the, and you know, it depends, um, where you are in the world and, and obviously the type of surgery and doctors, medical doctors would know better, uh, than I do.

But in generally, um, for tumor resection, Um, that type that aren't on the surface of the brain. The, the guidance system is kind of getting into the, uh, the standards, uh, that you need to bring in. But functional maps, um, are on the rise. Um, we, we, we have a, a product that is now, um, used in the us. We worked with, um, you know, various centers over there, medical centers, and it's used more and more and more in medical practice.

And the ab adoption curve is there. So we'll see. Um, in the next year or so, these kind of techniques will, will become more and more routine in, uh, this type of surgery. And it's kind of down to us to make it as accessible as possible. But the bigger, um, the bigger issue past neurosurgery, which is kind of a natural place for brain map is, is mental health.

And, you know, this is something where brain maps aren't used at all as far as I know, or probably a few pioneers out there use brain maps, but it could actually be very useful there. And there's um, there's some proof points, uh, that we can use for this. So, um, there are things such as, um, brain stimulation.

So we, we hear. For instance, a technique called transcranial magnetic stimulation. And it's a technique by which you can pulse a small electrical current, very focused, um, in your brain. And it's a, it's a, it's a small electrical current, so it's, it's not, um, a full brain electro shock. And what this does is puts your, your, your synapses in a state of plasticity and then they kind of refuse a grossly speaking back in a different equilibrium.

And so when you identify markers, um, with, uh, a brain map and machine learning overlay to identify anomalies, you can flag certain areas to be problematic, which can serve as targets for, for transgender magnetic stimulations. And there are other ways of stimulating the brain, obviously. And that actually, um, there's a growing body of literature that shows that it has some really, really good effects on, uh, for instance, uh, treatment resistant depression.

So you've failed three medications and you're still not going better while these kind of techniques. Could help, uh, with the right targeting system and such. And that's kind of what the, the, the, the bigger deal about brain maps is, is, um, you know, tackling into this enormous society crippling problem and improving brain care overall.

Yeah. It also seems like, and correct me if I'm wrong, that when you're looking at some, uh, let's say mental health, Type of issues that they tend to be complex and, and very, uh, heterogeneous. So a spectrum. Okay, so the autism spectrum, there are many different behaviors that are associated with someone who is on the autism spectrum.

And so is this kind of brain mapping useful in helping to differentiate maybe different parts of that spectrum to better focus the appropriate therapies for different patients? Yeah, you, you, you said the world, right? Which is heterogeneity. And that's, um, that's a big issue when you look at, um, you know, psychiatry today, let's say, you know, a person who turns up with symptoms of depression, um, and then the, the symptoms of depression.

You have something related to your sleep and that can express in, you know, struggling to fall asleep or waking up in the middle of the night. Or actually waking up too early. Um, and, and these are three different ways sleep problems can, can, can turn up. And that's only the sleep part. You know, you've got the eating part in depression, which is eating too much or not enough.

And, and these are very, very different behaviors. So where brain maps can help is that they can. Break away from those classes to look at the symptom only, and those classes are helpful. If you read the DSM five and you find those categories, they, they are very, um, useful ways to look at your, at your patients.

But the, the thing that the brazen produces. Is the behavior and sometimes behavior correlates. Right? In the case of depression, your, your eating disorder is correlated with sleep disorder and the whole thing working together is kind of creating depression, but precursor to that is the actual symptom.

And the symptom is generally underlied by specific functional areas. And so if you can start measuring them and you can start seeing how they're working together, you get a better understanding of how they can produce the symptom. And if you break that down into its constituents and use brain maps again to measure those things.

So you say, well, this specific eating behavior is based on, um, you know, five different components that interplay into a, some sort of reward system. And we can see that the, the, the piece of your brain that is working with the. Anticipation of your reward isn't working quite well, it's, it's actually tuned down.

So nothing brings you joy in life. Um, well it becomes very actionable. You're getting towards, um, a specific part of the brain in a network that is not functioning quite well and it helps you to develop better therapies. And that's kind of where, where we headed with the kind of brain maps we're putting together.

And if people wanna learn more about your company and, uh, what brain mapping is all about, where can they look? So, the, the best place to, to star is our website. So you find it at, um, oh eight t.com, O as Oscar, eight as the number and T as the letter. t.com stands for tions, which has, um, eight letters in the name.

There you'll find, um, a lot of information about brain maps. You also find the, uh, breakdown of functional, um, parcels of the brain. So we've got all that published as well as, uh, all the papers we've been putting out there, plus, um, how you can get access to our software. And if you are interested to go further, please get in touch, uh, either through the, uh, general email address or, um, I'm happy to provide my details through here.

So Stefan Doan, thank you very much for your time today. I think it's fascinating that we're able to start to unravel the complexity of the brain in such a useful, functional way that really is making it easier for physicians and neurosurgeons to be able to, uh, deliver their therapies or therapeutic methods a little bit more precisely.

So thank you very much for your time today. My pleasure. Thank you. And as always, thank you for listening to another week of the Talking Biotech podcast. Be sure to tell a friend about what you're learning on this particular podcast. Our numbers continue to grow over the course of seven and plus years, and we'd like to even keep that influence going stronger into year eight.

This is a Talking Biotech podcast, and we'll talk to you again next week.