 Live from Boston, Massachusetts, it's theCUBE. Covering Red Hat Summit 2019, brought to you by Red Hat. Well, good afternoon, welcome back here on theCUBE as we continue our coverage of the Red Hat Summit. And you know, every once in a while, you come across one of these fascinating topics. It's what Stu and I get so excited about when we do theCUBE interviews is that you never know where you're going to go, the directions you're going to take. And I think this next interview has been a fit into one of those wow interviews for you at home along with Stu Miniman, I am John Walls. And we're joined by Dr. Ellen Grant, who is the director of the FEDO Neonatal Neuroimaging and Developmental Science Center of Boston Children's Hospital, so far so good, right? And the professor of radiology and pediatrics at the Harvard Medical School. So Dr. Grant, thank you for joining us here on theCUBE. And Dr. Rudolph Pinar, who is the technical director at the FNN DSC and an instructor of radiology at the Harvard Medical School. So Dr. Rudolph Pinar, thank you for joining us as well. Thank you very much. All right, good. So we're talking about what's called the Chris Project, which was a technically based project, Boston Children's Hospital. I'm going to let you take it from there, Dr. Grant, if you would, and just talk about the genesis of this program, the project, what its goal was and how it's been carried out. And then we'll bring in Dr. Pinar after that. So if you would please. Sure, so the goal of the Chris Project was to bring innovative imaging analysis to the bedside, to the front end, where clinicians are not like I, are working all the time, but aren't sophisticated enough or don't have enough memory to remember how to do line code in Linux. So this is where I initially started when I was reading clinical studies and I wanted to run a complex analysis, but there was no way to do it easily. I'd have to call up someone to log into a different computer, bring the images over. Again, lots of complex steps to run that analysis. And even to do any of these analyses, you have to download the program, set up your environment. Again, many, many steps that are, if someone is a physician, I would rather deal with the interpretation and understanding the meaning of those images than all that infrastructure steps to bring it together. So that was the genesis of Chris, is trying to have a simple Windows point-and-click way for physicians such as myself to be able to rapidly do something interesting and then be able to show it to a clinician in a conference or at the bedside. And who's working on it then? I mean, who was supplying what kind of manpower, if you will, root off to the project? Kind of in the beginning, I would say maybe one way to characterize it is that we wanted to bring this research software which lives mostly on Linux onto a Windows world, right? So the people developing that software researchers or computational researchers who do a lot of amazing stuff with image processing. But those tools just never make it really from the research lab outside of that. And one of the reasons is because someone like Ellen might not ever want to fire up a terminal and type in these commands. So people working on it are all this huge population of researchers making these tools. And what we try to do, what I try to help with is how do we get those tools really easily usable and accessible? And to make a difference, obviously. So that was a genesis, that was kind of the need that we had in the beginning. So it started out really as a bunch of scripts, shell scripts, you start to type a couple of stuff, but not so many things. And gradually with time, we tried to move to the web and then it began to grow. And then kind of from the web, stretching to the cloud. And that's kind of the trajectory in a nutshell. And as each step moved along, more and more people kind of came in to play. So Dr. Graham, I think back, I work for a very large storage company and remember, object storage was going to transform because we have these giant files we need to be able to store them and manage them and pull them up. But let's talk about the patient side of things. What does this really mean? We have to talk about order of magnitude that cloud can make things faster and easier. But what does this mean to patient care and quality service? Well I think what it means or the goal for patient care is really to get into specialized medicine or individualized medicine and to be able to not just rely on my memory as to what a normal or abnormal image is or the patients I may have seen just in my institution but can we pull together all the knowledge across multiple institutions throughout the country and use more rigorous data analysis to support my memory? So I want to have these big bridal in front of lobes that are there at the cloud that help me remember things and to tie these connections and not have to rely on my visual gestalt memory which is obviously going to have some flaws in it. So and if I've never seen a specific disorder say for example at my institution if they've seen it at other institutions we can run these comparisons all of a sudden I may be aware of a new treatment that otherwise I may not have known about. All right so Rudolf my understanding is this is tied into the mass open cloud which I've had the pleasure of talking on the program at another show back here in Boston. Talk a little bit about how this is enabled. I mean massive amounts of data you need to make sure you get the right data and that it's valuable information and to the right people and it gets updated all the time so give us a little bit of the inner workings. Exactly so the inner workings that can be a pretty big story but kind of the short story. We have time don't worry. The short story is that if we can get data in one place and not just from one institution but from many places then we can start to do things that are not really possible otherwise. So that's kind of the grand vision so we're moving along those steps and the mass open cloud for us makes perfect sense because it's there's a academic link to Boston University and then there's the red hat being one of the academic sponsors as well in that effort. This kind of synergy that came together really almost perfectly at the right time as the cloud was developing as Rudolf was moving in as we were trying to move to the cloud it just began to link all together and that's very much how we got there at the moment and what we're trying to do which is get data so that we can cause medicine really it's amazing to me in some ways there's all these amazing devices but computationally medicine lags so far behind the rest of the industry. There's so little integration there's so little advanced processing going on there's so much you can do with so little effort you can do so much so that's part of the vision as well. So help me out here a little bit I mean maybe a before and after. Let's look at the situation maybe clinically speaking here where a finding or a revelation that you developed is now possible where it wasn't before and kind of what those consequences might have been and then maybe how the result has changed now so maybe that would help paint a practical picture of what we're talking about. I can use one example we're working on but we haven't got fully to the cloud because all of these things are in their infancy because we still have to deal with the encryption part which is a work in progress but for example we have mined our clinical databases to get examples of normal images and using that I can run comparisons of a case that comes up to say whether this looks normal or abnormal so it flags the condition as to whether it's normal or abnormal and that helps when there's trainings or people not as experienced in reading those kinds of images. So again we're at the very beginnings of this it's one set of pictures there's many sets of pictures that we get so there's a long road to get to fully phenotype or characterize any one brain but we're starting at the beginning of those steps to very, to digitally characterize each brain so we can then start to run comparisons against large libraries of other normals or large libraries of genetic disorders and start to match them up. And you said you work in fetal neural imaging as well so you're saying you take an image of a baby in a mother's womb and many take hundreds, thousands, whatever it is and you develop this basically a catalog of what a healthy brain might look like and now you're offering an opportunity to take an image here on early May of 2019 and compare it to that cataloged look and determine whether it might be an abnormal normality that otherwise couldn't have been spotted before. Correct, and put a number to that in terms of a similarity value or a probability value so that it's not just me as a clinician saying well I think it's a little abnormal because then it's hard to interpret that in terms of how severe is the spectrum of normal how sure are you? So we put all these data together we can start to get more predictive value because we can then follow more kids and understand if it's that dissimilar what's most likely a disorder? What's the best treatment? So it gives you better phenotyping of the disorders that appear early in fetal life some of which are linked to we think can be treated say for example with upcoming gene therapies and other nutritional interventions so we can do this characterization early on we hopefully can identify early therapies that are targeted to the abnormalities we detect. So intervene well ahead of time. Absolutely. And also the other thing is I mean Ellen has oftentimes said how many images she looks at in a day and other radiologists and it's amazing she said the number 100,000 at one point so you can imagine the human fatigue, right? So imagine if you could do a quick pre-processing and just flag ones that really are abnormal but they can be grossly abnormal but at least let's get those on the top of the queue when you can look at it when you are much more able to think these things through. So that's one good reason of having these things sitting on an automated system say out on the cloud or where it might be. And where are we with the rollout of this and kind of expansion to maybe other partners? So a lot of stuff has been happening over the last year. I mean the entire platform is still I would say somewhat prototypical but we have all the pipelines kind of connected so data can flow from a place like the hospital it can flow to the cloud. Of course it's all protected and encrypted. On the cloud we can do any analysis we want to do provided the analysis already exists we can get the results back to a clinician. We have the interfaces, the web interfaces built they're growing so you can at this point almost run the entire system without ever touching a command line. A year ago it was partially there. A year ago you had to use a command line now you don't have to. Next year it'll be even more streamlined. So this is the way it's moving right now. And what's great for me personally about the cloud as well is that it's not just here in Boston where you can benefit from using these technologies. For the price of a cell phone and a cell signal you can use this kind of technology anywhere. You can be in the bush in Africa for argument's sake. And you can have access to these libraries of databases imaging that might exist. You can compare images that are collected wherever it might be just for the price of connecting to the internet. So you just need a broadband connection of some kind right? Yeah exactly. So when you think about it again about, even if we've talked about mobile technology, 5G coming on as it is, here in the U.S. rural health care. Yes, third world. Leveling that in third world. I was thinking more along the lines of here in the States even with some areas that just don't have access to the kind of obviously platinum care you get here in the Boston area. Exactly. But all those possibilities would exist or could exist based on the findings that you're getting right now out of the Chris Project. So where does the Chris Project go from here? Well what we'd like to do is get more hospitals on board. I think in pediatrics we have a lot of challenge because there are so many different rare disorders that it's hard to study any one of them from one hospital. So we have to work together. There's been some effort to bring together some genetic databases. But we really need to bring also the imaging databases together. So hopefully we can start to get a consortium of some of the pediatric hospitals working together. We need that also because normal, for normal you need to know the gender, the age, the ethnicity. So many demographics that are nice to characterize what normal is. So if we all work together, we can also get a better idea of what is normal and what is normal variance. And there's a lot of other projects that are funded by NIH building up some of those databases as well too. But if we can put them all into one place where we can actually now query on that, then we could start to really do precision medicine. And the other thing which we definitely are working on and want to do is build a community of developers around this platform. Because there's no way our team can write all of these tools, nor do we want to. But we want everyone else who wants to make these tools very easily hop onto this platform. And that's very important to us because it's so much easier to develop to Chris than it is to develop to Amazon. There's almost no comparison how much easier it is. Well, definitely a theme we hear echoing throughout Red Hat Summit here. Does that tie into the OpenShift community or what is the intersection with Red Hat? It definitely does because this is kind of the age of containerization which makes so many things so much easier. And this platform that we've developed is all about containerization. So we want to have medical or biomedical or any kind of scientific developers get onto that containerization idea because when they do that, and it's not that hard to do, but when you do that then suddenly you can have your analysis run almost anywhere. And that's an important part in medicine because if I run the same analysis on different computers I can get different results. So the containerization concept I think is something that we've been after which is the reproducibility that anybody can run it as long as they're using the same container we know we're going to get the same result. And that is critical. Yes. Especially with what you're doing, right? Yeah. You have to have that 100% certainty and standardization goes along the way towards achieving that. Fascinating stuff. Thank you both for joining us and good luck. You're an exciting phase that's for sure and we wish you all the best going forward here. Thank you so much. Thank you both. Thank you both. Back with more from Boston. You're watching Red Hat Summit Coverage Live here on theCUBE.