 From around the globe, it's theCUBE with digital coverage of AWS Public Sector Online. Brought to you by Amazon Web Services. Hi, and welcome back to theCUBE's coverage of AWS Public Sector Online. I'm your host, Stu Miniman. Really excited always when we come to the AWS Public Sector show, it's not only governments, but you've got nonprofits, education, and lots of phenomenal use cases from the practitioners themselves. Really happy to welcome to the program Dr. Sergio Papa. He is a radio diagnostic specialist at the Centro, Diagnostico Italiano, of course, in Italy, if you can't tell by the name there. And Dr. Papa, thank you so much for joining us. Why don't you start with a little bit, your role at CDI. Thank you for your invitation. And I am the director of the diagnostic imaging department and the radiotherapy and nuclear medicine. We are a very huge institution in diagnostic area in laboratory and diagnostic imaging. I think one of the biggest in Europe. Excellent, and of course, one of the very relevant things to talk about, you have a project called Artificial Intelligence or AI for COVID. Maybe explain to us a little bit about what led to this and how this, what the goals are for this project. Yes, we, as you know, as you also are today, we were in the February March in the middle of the biggest pandemic we ever experienced in the past. So we were thinking about some new, new, new method, new, new methodology to give an help, to even hand for this important emergency. Central diagnostic with AI is working from three or four years around artificial intelligence in diagnostic imaging. In particular, we are working very hard in radiomics after if you need, I can talk, I can speak about the radiomics methodology that we are using. So we had the idea of applying radiomics method on diagnostic imaging, in particular, Tristics Ray, with the purpose not to have the diagnosis for this patient. It doesn't matter for us. We were focused on predicting the clinical outcome of this patient. I mean, all these, all the people already diagnosed with COVID-19 virus where they had X-ray examination. Then after we applied over this very huge number of X-ray examination, radiomic methods to understand what could be the clinical output of this patient. So I mean, dividing the people going well and then people, otherwise, that were going to a reversal of the illness, I mean, to critical therapy, even to the death. So we were trying to divide to groups of this patient. Yeah, absolutely. So important, of course, once you have that diagnosis to understand who needs the most treatment, making sure that hospitals can put the right resources in the right places. So really impressive to do something like machine learning on this in a relatively short period of time from when this whole pandemic has started. Help us understand a little bit, what are the underlying technologies? How does AWS fit into this whole discussion? The support of AWS is in many different areas. The first is we are trying to develop a platform without AWS that's useful for the hospitals, for the institution to store in a unique imagery set. All the images coming from different institutions. So we don't need anymore to send the images in anyway. This is the first aid that AWS can give to us. Second is the use of machine learning models to analyze these kinds of images coming from X-ray chest through AWS systems. The third, I think, could be an aid in generating the structured reports for this patient. And moreover, the identification of patterns, different patterns that we can find inside the images. This concerns the radiomic theory. I mean, inside the images, there are many more information than what the radiologists can get from. So artificial intelligence can help us. So AWS can help us to detect all these kinds of patterns that we want to collect for our study. And the fourth reason is, for the aid that AWS can give us is to share this kind of modality with other scientific centers or research center. And not only for this specific pandemic now, but also in the future. Maybe we will have... We do not hope this, but we could have the second wave of the pandemic. There are many signals about this in China, also in Europe. So this will be useful in the future to find the second wave of this pandemic. And also the final reason is that we will share all our results of this study with all the scientific community. I mean, we will improve an open access model together with AWS to share this information with all the scientific community in the world. It's wonderful that this information can be shared broadly across the community. So important for tackling this challenging pandemic. I'm curious, has your unit or had you used artificial intelligence machine learning before? I'd love to just get a little bit of background on how much you've used this technology, how accessible it is to be able to leverage it for new use cases like this. Absolutely, because I mean, I am a radiologist. If I check an image in a CT scan or in an XA, I can see inside that image, the maximum that I can see is 10, 15, the maximum different patterns. I mean the volume, the dimension, the growth, which is the way of taking contrast media or different sort of washing, wash out. But with my eyes, I can see 10, 15 different patterns. If the machine system examines the same image, it can reach out hundreds of patterns that I cannot see. So we can detect all these patterns in different images. We can collect this, machine learning system can work on this and put together all the similar patterns or they can be divided in different clusters. And then the system has to compare all this difference cluster of group or patterns with a very huge database that we built before. Comparing and trying to understand which patterns are linked to a different outcome. So we can say, okay, this image has 20, 30, 100 patterns that suggest to us that the destiny of deletion will be one specific while another deletion that for me is exactly the same with my eyes. Systems will tell us that the deletion are really different. Their destiny is really different. This is the radiomic theory and this is what we are applying in our study on X-ray chest in COVID passion. As I said before, we select only positive patient. I mean, all people that is for sure they are positive to COVID and in the first X-ray chest entering the hospital we try to evaluate from the first chest X-ray what will be the real destiny of the patient, better or worse and then we can also predict, try to predict obviously how many intensive care beds are necessary in that institution. We can set the therapy and adjust the therapy for the different kind, different group of patients. It could be a very big help to an institution, to an hospital, especially in periods like in our March or April when every day in every hospital in Northern Italy they were entering 200 person per hospital. It was a dramatic situation. Excellent. One of the other things that this pandemic has done is it really required some strong coordination between both public and private entities. If you could speak a little bit to that, my understanding is that AWS has also helped support this with the donation of computational credits. I believe it's the AWS Diagnostic Development Initiative. So help us understand how the finances and the partnerships between public and private help everyone really address this current challenge. Well, the support from AWS for us is very important because now in this way we can use a lot of computing systems much more than what you had in our institution. And moreover, I think that sharing our information with all the scientific community at the end of our study, it will be a very, very important thing to do. Now we are beginning to appear on our drawer, not on our websites, some effort to share information about COVID. Our study could be really one of the most important. Great. Final question I have for you, Dr. Papa. Give us your ideal vision going forward. You talked a little bit about how, you know, the importance of this to be able to watch and be prepared for a potential wave two. Where else is this research relevant and where do you see this project going forward? Well, our study is focused on pneumonias from COVID-19 but the methodology can be applied in every kind of intestinal pneumonia. I mean, this is one of the first effort that is made in radiomics to segmentate one full organ. Usually in radiomics we study the single lesions or little areas, I mean nodules or metastases or primary tube. This is one of the first, very first important studies where segmentation is dedicated to the whole organ. I mean, all the lung, both lung. In every patient we segmentate the right or left lung and in order to study diffuse pathology, in this case, pneumonia, pneumonia, intestinal pneumonia that is very different from a bacteria pneumonia. And this methodology at the end of the study will be shared with the scientific community and could be a very interesting advancement in our job. Dr. Papa, thank you so much for joining us and thank you so much for the very important work that your organization is doing to help attack the global pandemic. Thank you, thank you, thank you. I'm Stu Miniman. Thank you for watching theCUBE.