- 2 months ago
Category
🗞
NewsTranscript
00:00Brain imaging in children, if we have a better understanding of the brain in fetal development
00:03and for babies and toddlers, what will that allow us to understand? What does it prevent? What does
00:08it treat? Well, everything begins in utero, pretty much. So your life is an arc from infancy or when
00:14you're conceived through to adulthood. So the more we can understand the early development, the more
00:18we can start to understand how we make sure children are on the right trajectory. So the
00:22goal is to characterize brain development very early on. So we tell at the very earliest point
00:27when to start to deviate from a normal trajectory so we can get things back on track early as
00:32possible. And ideally, we want in future to be able to prevent diseases from happening,
00:37not just try to deal with them and try to correct them later on when the damage is partly done.
00:41So how early can we do it today and detect that there's something wrong? How early
00:45realistically do you think we can get it to? Yeah. We've started looking at fetuses at about
00:5111, 15 weeks, something around there at the earliest, closer to around 18 weeks. We start
00:56to characterize brain development, you know, 18, 19 weeks or so. So it begins quite early
01:01when we start to see and look at early brain development. Well, you and your team did a study
01:06a few years ago that gave you results to argue for earlier MRI during pregnancy. Yeah. Is that study
01:12enough to actually change the standard of care? Well, we do use it early here at Boston Children's. So
01:18when there's an indication, we do it as early as we can to better characterize the entire fetus
01:22because it's not just the brain, it's the body it's attached to, too. So we want to understand
01:26not just the brain development, but how that brain is developing in the context of the other organ
01:30systems. So we can do that. Because it doesn't necessarily run hand in hand. Like it can be
01:34very different, right? Like in terms of what's going on with brain development versus the rest of
01:39the system, they can disconnect? No. They're intimately connected through life. So if there's one,
01:45yeah. So that's what we want to understand. So let's say, for example, we deal with a lot of
01:48congenital heart disease here. That has effects on brain development. We deal with congenital
01:52diaphragmatic hernias. That has effects on brain development. So everything that's happening
01:56in the fetus, whether it's a brain or not, has the potential to have subtle effects on brain
02:01development. Why do kids, I mean, kids do need specialized tools for brain imaging. Yes.
02:08Talk to us about that one way. Yeah. That's the whole reason that I came to Boston Children's is
02:12industry's not interested in fetuses, infants, and young children. So it's really hard to get devices
02:18that are built specifically for these age range. So that's why I brought a team of technical people.
02:23So they're engineers, physicists, computer scientists, data scientists that help to either
02:29develop the devices or come up with better ways to analyze the data that we get with an eye on trying
02:36to understand pediatric disorders. So for example, we want to monitor and we're developing optical devices
02:41for the NICU to monitor cerebral blood flow. But the heart rate of a neonate is 150. So we have to sample
02:49at a much higher rate than you would in an adult to get the same information. So we have to build
02:53specifically devices to the physiology. And then if you think of a head of a premature baby, it's very,
02:59very small. So I can't take a probe that we use in adults and just put it on a preterm. So we have
03:04to develop the devices to fit the size of the infants. I want to just go, and I feel like we
03:10touched on this earlier. I mean, we are Bloomberg Business Week. We are Bloomberg and very entrenched
03:16in financial markets. And I feel like the more I've been doing this, money just follows everything.
03:21Money is why people do things or don't do things. Is that really it? Is it just the market? I hate to even
03:26make it that way. The market size. And so you don't have medical equipment companies building
03:31the things because they just don't think the market size is big enough? That is a big problem.
03:35Yeah. And I think that's where we're trying to get into more of a business perspective. Like if we
03:39do a small startup that starts to answer those questions, then a bigger company might buy it.
03:44But if we stay in the research realm, then it's sometimes really hard to go that last mile and get
03:49something into clinical practice. So how do you do that? How do you cross that? So what do you do?
03:54Yeah. This is what we're strategizing on right now is trying to figure out how we do those
03:58small startups, get industry interested. And a lot of the things we're doing right now,
04:05actually, one of the projects we're working on is AI strategies, right? And if we can get enough
04:10data on infants or fetuses and so on, we can start to build models that predict not just group
04:16outcomes, but we want to get to individual outcomes because that's what parents care about,
04:19right? So if we can figure out, get those models together. So that's what we're working on now is
04:23trying to create these AI models that are specialized for pediatrics and hoping to do
04:28startups around that particular concept. I have to ask one more question. Are venture
04:31capitalists interested? I don't know because we haven't really talked to any that I've heard about,
04:36but I think they always want something that's almost ready. So we're hoping to go a bit further
04:39along. Yeah. A little bit further along. Okay. Interesting.
04:42Can what we learn and what you understand through imaging about the brain's development
04:46be applied to how adult brains are treated? Everything in adult life has its genesis in
04:54infants, right? So we learn a lot. We were all there once. Yeah. Yeah, exactly. And some of the
04:59ways the adult brain response is more prominent in a pediatric brain. So in some disorders,
05:05you go to pediatric models to see a physiology that's more prominent in neonates or infants,
05:10but also occurs in adults. You know, there was a, in, in doing the research, uh, in the prep for our
05:15interview with you, there is a picture of, um, a physician, uh, or a therapist doing, uh, some,
05:24what's I think is called therapeutic hypothermia to a brand new baby's head. Uh, and my understanding
05:32is that oxygen deprivation around birth is one of the leading reasons that you actually see babies
05:38come into the, the NICU. Yes. That was one of the, one of the main reasons. Yes. And, and the,
05:43the therapy for this is as simple as. Yeah. You cool them down or down their normal. Yeah. Okay.
05:49At least it's a normal thermic is when they have injuries, the response to sort of the whole
05:53physiological response to an injury is to have a fever and that is detrimental. So we want to keep
05:58them cool so that they don't set off these cascades of brain injury. And that's partly why we built
06:03this one device because we want to be able to monitor, um, through the NICU stay and optimize,
06:08um, management. But it's interesting. We don't even know what the right blood pressure for a newborn
06:13is. So this is why we wanted to have a probe that could measure cerebral blood flow to the brain
06:17because there is no way to monitor whether there's enough brain or oxygen getting to the brain,
06:22um, with the tools that we have right now. I feel like we never even talk about blood pressure
06:26when it comes to like infants, right? Yeah. We just don't, but you need to know. Yeah.
06:30You did your residency and fellowship in the 1990s. Uh, curious how imaging has changed since
06:35then. And where do you think, how will it improve in the next, I don't know, 10 years? Yeah. I don't
06:40know. What's a smart benchmark. Yeah. Yeah. No, when I was in training, MR was just starting
06:43and it was very slow. Um, so where we come now is the acceleration acquisition is just incredible.
06:49What used to take us an hour to do, we can do in 10 minutes now. Um, so the, the speed,
06:55the speed of acquisition is huge. Um, we also developing, um, a lot of, uh, post analysis that
07:01we can do after the images are acquired to give us more quantitative metrics because the whole thing
07:06in medicine to get past the qualitative read of a radiologist, which is helpful, but we want to put
07:11more numbers on it so we can have a more dynamic range on how we describe each child. And this,
07:17this, therefore we can get into better precision medicine and open prediction. So we're getting more
07:22to that quantitative aspect of imaging now and not just brain, but all body parts, of course,
07:28and, you know, down to fetal age. Is it for kids too? Every case is very personal and individual,
07:35or are there trends and things that you can help in so that one case can help another? Like,
07:40is there a body of knowledge that gets built off of this? Yes, there's body of knowledge gets
07:44built off of this, but this is where we come back to AI. I only can remember so much, you know,
07:49even though I've been in practice for a long time, things fall out.
07:51So this is where I'm really excited about AI because I can, you know, mine our databases to
07:58find where's an individual child just like that one I'm treated now. What did they respond to?
08:04What worked for them? And how are these two similar? So I can mine the databases to start
08:08to come up with individual outcome prediction, which is what we're doing right now with databases
08:12we've got from some of the major trials for hypothermia. And so we can use this large database
08:18to start to take individual outcomes. You can say, well, I have a newborn with this pH that had
08:24these, you know, and I'm a mother of this age and put in features and they could give you from that
08:28database an outcome prediction. So working on that and also working on making data more available to
08:34parents, because I think we've, a lot of parents are very frustrated with trying to read the literature,
08:40even if you're using chat GPT or open arts, it's really hard. And then you get group statistics.
08:45And then where does my kid fit in between the 25 to 75%, you know, good outcome or something like
08:50that to get chatbots that can work with some of our databases. So people, anybody can talk to,
08:57you know, a physician, so to speak, to give the answers that they want.
09:00That's, that's pretty remarkable because, you know, I just think about the, um, the tone of
09:09these chatbots and if there's a way that they can be, um, you know, we talk about, we talked earlier
09:16this week about, um, what a challenge it can be for people to actually interact with them in a,
09:22in a quote unquote normal way. But is there a way for, for them to actually be empathetic and,
09:27and, and work with patients, work with parents, work with families. And forgive us, we've got
09:31about 40 seconds. Yeah. Yeah. No, we're working on that, but I can't tell you all the secrets
09:33because we're going to start. You can't wait. No, you can go longer than that. Can I ask you something?
09:39When you guys do use AI in chatbots, do you have hallucinations? Like do the AI hallucinations,
09:45do you, or how do you, especially when you're dealing with medical prevention, there's a lot of
09:49safeguards we put around that. So it's, it's, we have, again, this is sort of more the secret sauce
09:55that I can't talk about yet, but, um, there are ways to constrain chatbots to give you reasonable
10:01answers that are statistically sound.
Be the first to comment