 Hey everyone, this is Raghu. I'm a solution architect at Red Hat. It's very nice to be between you and lunch, but I promise not to take much time. Let's get started. So what will we discuss? We're going to look at key challenges and what's Chris in a box, Chibox, and then some of the architecture and see if we can do some demo. I know we don't have a lot of time, but I'll see how much we can get through this. So when it comes to key challenges, analytical techniques, they can have a huge impact on radiology, but there are always challenges, right? When it comes to innovation, all the innovative techniques, they're not useful if you don't integrate them into clinical workflows. Now the second part is when you think of hospitals like in rural or developing countries, they don't have a lot of on-site computational power to do all the analytics. And the researchers, they also don't have access to the data so that they will be able to develop some new techniques on the, or they don't even have the computational resource. So let's see how Chris in a box can address these challenges, right? Now, what's Chris in a box? It's a self-provisioning system and it can schedule containerized compute on the edge. From a domain perspective, it can autonomously analyze the patient data and then it can push rules back into clinical systems that are operating at the edge, right? Now, the picture that you see here is it's giving you a view of a clinician or a researcher sitting somewhere in a hospital and they're trying to leverage Chris and run some analysis and then they take a look at the reports, right? Now, that brings us a question like, what's Chris? It's an open source project. It's a container platform for medical research, like medical image processing and compute. Who develops it? It's developed by Advanced Computing Group in the FNNDSC at the Boston Children's Hospital and we have a collaboration with Red Hat and New England Research Cloud and a few others. And there's the Boston Children's team. We from Red Hat, we work closely with Boston Children's from a Chris project perspective. I'm sure we have Jennings as well in the audience here. Hey, Jennings. Now, when it comes to architecture, we're using Ansible. I do know there's a lot of blocks over here and different different components, but let me explain in simple terms. Now, for us to be able to get Chris deployed in an edge device, all I do here is I go to my Git repo and I do a Git commit. And when I do a Git commit, I can actually do the changes for the specific packages on an operating system or I can also do the changes related to the application itself. And once I do the Git commit, the repository is configured so that it triggers the Ansible automation platform. And as you see here, we're using OS build collection and it's gonna do a number of tasks so that it's able to make us the updates available, which means think of it like this. The first provisioning, you will need the ISO itself so that you can boot it. And in this case, this ISO does come with the application. Now, after that, we will think of the lifecycle management, right? Like, how do we make the updates? Now, in this case, when I do the updates, once the image is built, all I need to do is I just go to my edge device and do RPMOS tree upgrade. I'm not gonna go into details about RPMOS tree, but that's a way of getting some updates in the context of edge. And so that's an oversimplified way of telling what the solution does, but there's a lot of work put behind this to make it all happen for the automation. Now, let's see a little bit in reality what it means. So, for the demo purpose, I just go here and that takes to the git repo here. Now, remember, I talked about doing some changes. Now, in this case, I'm not doing like any specific changes, it's all there, but just to kind of show how it works, I'm bumping the blueprint version. What was it before? Eight, I'm gonna do it as nine, one dot two dot nine. And I'm gonna do a git commit, right? Now, let's go back and see what it's gonna do in the Ansible automation platform. Now, these are multiple templates and if you go to the job, if you see, there is a couple of jobs that got triggered. Well, in reality, it's gonna take around 30 minutes for everything to be finished, but I just wanted to show you the template real quick. This one, it's still waiting on something. So, okay, so it's starting to be doing some things, right? So basically what's happening here is, it's gonna interact with the image builder and then it's gonna build the image with all the artifacts and then it's gonna copy those like a Tor file into a repo and that's when it's available for the device to be able to pull in those upgrades. Now, this is great from a technology perspective, right? If you talk to someone saying, hey, I have a ton of devices and how do I automate them so that I can deploy them at the edge? It makes them super happy. Now, let's think from a patient care perspective or a clinician perspective. For that, I'd like to show you a pipeline. Now, in this case, this is a leg length analysis pipeline. Basically, that can be used to identify some discrepancies. So think of it like this. Chris, when you're running Chris on edge and then you would actually use multiple plugins, which means Chris has the ability to do different type of analysis. Not just the leg length, but other ones too. But in this case, imagine, this is an animation, by the way. So it takes some time for you to go into Chris and run the analysis, but just to make the use of time here, this is an animated view and when I click on it, if you see here, it's receiving some data from PAX. If you're in healthcare, I'm sure most of you would know what a PAX is or PAX arrays. Now, once the data is received, it's uploading to Chris. And this is where it gets interesting, like finding landmarks with AI. It's using a specific Chris plugin that will be underlying using a train model, pre-train model, and that will be finding some landmarks. And once that's done, it's gonna be doing the measurement of lengths based off the X-ray, which means, so this is another plugin too. And see, it also has some data over there once it's done, and then it's pushing it to the PAX. So again, I'm not a healthcare expert, but that's the data. Someone like clinician or a doctor would be very interested, and it helps them to get an additional perspective, not just based on what they remember in their memory, but if you think of the data that's coming across from multiple research organizations, so this helps them to make some better decisions. So this is what you can do technically by being on the edge, meaning you don't really need to have access to the internet, assuming that the Chris is set up already in the box, and you would run this, essentially any hospital or a remote location so that it helps the doctors or the clinicians to get some better analysis, and also other value around how you could do automatically all this, like for example, using Ansible. So this is bothered, but I'm open if you have any questions. Yeah. Let me explain. So the edge device.yaml, so if you see there are multiple things here, the first part, if you remember the image I've shown, that's the blueprint. So the blueprint holds certain values. So for example, this is where I could have the version and all, but then there are certain things that let's say you have a new user, you can add the SSH key. So this is essentially where all the variables are being held, right? Now, once you have it, you're gonna specify this particular file, and if you go back to this playbook, so this is referring to this particular file that we just talked about, right? So we are externalizing all the necessary details that may need to go into the edge device using this edge device.yaml, and then we're using this particular playbook, and this is interacting with Ansible collection, always build, and that has multiple roles to do, like for example, how is it gonna be, like the image builder needs a blueprint, and once it gets the blueprint, it starts building the image, and then you have other things too that the always build collection does, and this is how it works, and you have a web hook over here. Let me see if I can show that. So you have a web hook that's configured with Ansible automation platform, so which is why basically when I, so I've said it in a way where on this particular branch, I'm gonna do some changes and I make a commit, and that's when it triggers over here, and then this is when the Ansible automation platform takes over the responsibility of everything that it has to do. So it's gonna help us both ways, like the first thing is, you can get a brand new image that you can put in any edge device. Also, if you want to make some changes, you still can do that. All you need to do is you just go ahead and update the variables over there in the edge device.yaml, and you just wait for few minutes till the update's already, and just imagine you would log in into some edge device, and then you do the updates as needed. Yeah, so thanks for asking. In this particular situation, we have used part-man to keep it simple. So this runs part-man containers on the device, but we also have the same work to be done using Microsoft so that we have two different versions of implementing the solution. Yes, to answer your question part-man. Yes, indeed. So in this case, I haven't have a real deployment that I can show you, but it's a work in progress, but you're right. So we would go ahead and go to cockpit, and you would be able to see the containers. I think I'm out of time. Thanks everyone.