Enterprise Imaging 2030 and beyond —Elevating Care, Powering the Intelligent Future- Sponsored by AGFA HealthCare

From the RSNA, welcome to the Radiology Journal podcast. I'm Dr. Linda Chu, Associate Editor of the podcast program. The title of today's program is "Enterprise Imaging 2030 and Beyond, Elevating Care, Powering the Intelligent Future." And this podcast is sponsored by ACFA Healthcare. Our special guest today is Dr. Andrew Mugman, a seasoned healthcare IT professional with over 26 years of industry experience when it comes to health tech innovations. As the global chief medical officer, Dr. Ahmed leads the medical affairs activities, providing oversight to the pre and post-market clinical risk assessment of ACFA healthcare solutions. He has also published bestselling books on AI in healthcare and other evidence-based white papers and case studies related to medical imaging informatics and enterprise imaging. Welcome to our program. - Thank you so much for your time today and the opportunity. I'm looking forward to our conversation today. - Yes, and this year's RSNA theme is "Imaging the Individual." From your perspective as an experienced medical imaging and informatics expert and chief medical officer, what does that theme mean for both patients and clinicians? - This is really interesting time when we think about medical imaging in general and radiology specifically. I like to say this that, you know, they say a picture is worth a thousand words. And the way I believe it is that a medical image is worth millions of pixels of clinical intelligence. And that intelligence is actually out there for us to make an assessment out of it. So when I think about imaging the individual, what I would think is it really means is we have to look at the patient behind every pixel. If I'm a radiologist looking at the image, you know, the patient is behind the pixel. So that means the radiologist is generating those reports. And then the third important aspect here is the referring physician who turns imaging insights into care decisions. And I think this is where precision medicine meets the human connection. Because technology, I think, should help all the three work together in harmony, the patient, the physician, the radiologist, so that it can personalize their care without losing the empathy or the clinical judgment that defines us. So what does this mean? I think imaging is moving from, you know, population averages to becoming more personalized in terms of diagnostics. That's an important aspect that we see now. And the second aspect is how enterprise imaging enables this by integrating data across specialties. And that's something that we have been doing at ACFA healthcare with our enterprise imaging solution, how we consolidate multiple imaging service lines. So we enable that visual intelligence for the patient. And the third and important aspect here is that how the individual is not only the recipient of the care, but also the provider of the care. Because that's where I feel that, you know, that concept around imaging the individual and the patient and the radiologist kind of come together. Yeah, that focus on the individual is so important. But these days, as radiologists, we are experiencing increasing workloads and a lot of people talk about being burnt out. And we often feel like we are just report generators that are under too much pressure. So how can imaging become more human-centered, supporting radiologists as individuals as much as patients? This is a very relevant conversation because the way I, what we have learned from experience here at ACFA healthcare as well, is that radiologists don't burn out because they read too many scans. They burn out because they work in fragmented systems. And if you think about how, you know, the digital transformation in radiology has taken place over the years from film to digital, and that digital transformation brought about siloed solutions, whether these were the radiology information systems, whether these were the PACS systems, and then, you know, breast imaging solutions out there, you know, oncology systems. And then on top of that, we're learning, you know, bringing in AI as well into the conversation, which is another topic on its own. So this burnout is not as a result of, because they're reading too many scans, this burnout is because of the number of clicks or the fragmented systems that need to be integrated. So what we have done over the years is learned about how human-centered imaging design can, you know, work around the radiologist, and not the other way around, that in the past, radiologists have been trained around to work on specific technology and different solutions and forcing them to, you know, you have to click over there to launch this particular study. You have to press that button to load certain exam. I think the next decade of radiology won't be about speed alone. It will be about the flow. And that's when intelligence, empathy, and technology move in the flow together. And I think that's when radiology becomes truly personalized, sustainable, and transformative. For example, the way we build our enterprise imaging solution, we have kept in mind that complex reality of modern imaging networks, because it is becoming more and more about human-centered design, as I said, right? And there are three aspects of this. How a streamlined user interface design helps the radiologist or the users with fewer logins, consistency is the key here. The look in the field should be consistent around different sites where they work with and collaborate with, right? That's important. And the second aspect here is intelligent orchestration. How can it help reduce cognitive switching the way radiologists work every day and collaborate with colleagues or large studies? And the third is embedding, I would say, the well-being of the user into the workflow design. And I think that is going to be the next frontier of enterprise imaging, where human-centered design will be at play and be more empathetic towards how radiologists work and help them reduce their burnout and load. - I want to follow up on this discussion on workflow and orchestration. Even in my department, a lot of the biggest pain points, as you said, are not necessarily burnt out from reading studies, but from a lot of the inefficiencies or how the work is distributed. So how can smarter workflows and orchestration help reduce the cognitive load and give cases to the right radiologists at the right time? - This is the key for conversation in many contexts. The way I would describe this is that orchestration is really about matching the right case to the right expert at the right time. So it's not just about efficiency, it's about quality and fairness, in my opinion. So, for instance, what we have done is our orchestration engine, it is natively embedded in enterprise imaging. It is not at a bolted on integrated product. So it helps connect the people with their priorities and performance. So the enterprise imaging solution, the way we look at it, it should bring calm into that chaos because of the number of studies, the number of exams, the reading, the specialty, the subspecialty list. So it balances intelligently the workload based on credentials, the urgency of the case that needs to be looked into based on the clinical relevancy. So the radiologists stay focused on reading and not writing because it is not their job or even, I think, not the job of the PACS administrators or system administrators to look into manually assigning cases to the work list of particular radiologists. So there are five key takeaways in terms of how an enterprise imaging strategy should work. When it comes to intelligent workflow orchestration. Number one, I think, is balancing the workload and how it helps prevent fatigue. Number two, I think, is the way we have done it is how prioritization by urgency, subspecialty, credentialing and availability. Because this is really relevant in terms of how you can reduce that fatigue to my earlier point that I mentioned. And number three, how we create equity across distributed reading networks because that is also relevant to how radiologists would like to work on a day-to-day basis. The fourth aspect which is becoming more relevant in the conversation is subspecialty or peer collaboration. So how we can ensure peer review compliance and how it fosters a culture of shared learning and intelligently orchestrating those cases, intelligently into that use case. And in the important, the fifth aspect is the patience, right? Because by doing so, a patient will benefit because every case is getting the right attention by the right expert at the right time with the right specialty focus. And now I wanna switch gears a little bit and talk about AI. AI is everywhere at RSNA over the past couple of years and there's a lot of promise on how AI can help us become more efficient, improve our diagnostic accuracy and so on. But on the flip side, there are many radiologists who worry about just having to do more clicks and more distractions. So what does an AI co-pilot or sometimes referred as an augmented intelligence enable or look like in daily practice and how can it make work easier instead of harder? - Interesting, you mentioned augmented intelligence because this has been our philosophy from day one when we started our AI journey. And not many may be aware that ACFA healthcare started working, testing, developing, evaluating our algorithms more than 10 years ago. And it came naturally to us because our customers said to us, "Hey, ACFA, you built this enterprise imaging solution "that consolidates all imaging. "You have a lot of pixel intelligence now captured. "So how can we work on benefiting "from this pixel intelligence?" And that's where we coined the phrase around augmented intelligence. And I could say I even have a definition for it where I believe augmented intelligence is the intersection of machine learning and advanced applications where clinical knowledge and medical data converge on a common platform. So that's where we feel that AI should not replace intelligence. It should return it to the clinicians. And I think that's the missing link to date. A good AI co-pilot fades into the background if it is intelligently embedded into their routine workflow and it guides the radiologist quietly and not demanding any additional clicks. If we look at historically what we have seen at our SNA and various Congresses, you know, various AI use cases, what I've felt is that most of these AI startups and platforms, they have been focused on feature functionality. Whereas what we have done is we have created functional clinical packages. So for instance, we have a common framework that we have developed for AI with an enterprise imaging which we refer to as Ruby. So with Ruby, we have built five core pillars of our AI strategy. That irrespective of whatever the AI algorithm or the use case may be, those five common areas stay consistent. And what are those? Number one, intelligent triage. So the ability for the system to intelligent triage cases based on specific findings or based on, you know, the use case so that radiologists, even before they open a particular study, they have a good view of, you know, whether they are resident, whether they're consultants, how that intelligent workplace looks like. The second aspect of this is a case may have been done for a different clinical manifestation. And another finding may have been picked up by AI. So that means intelligent routing and orchestration becomes very relevant here, which we, which is also referred to in the US sometimes as opportunistic screening, that the patient came for something else and something else was picked up. So how do you route those cases to the right specialist, right? And that's what we have built as the second core pillar of our AI program is intelligent orchestration based on specific findings that these cases need to be alerted to a specific group of clinicians, radiologists, or subspecialty reading program. So that's where the intelligent orchestration comes in. And the third important aspect of what we have done is embedding AI findings natively into the user's ecosystem. So if an Acfi Enterprise Imaging user is logged in, they will not need to click on an external viewer or an application to launch AI results. We will natively show those AI results into their ecosystem. The fourth aspect of this is automation of how the display should look like. And as you would know, radiologists, one of the pain points that we speak about and hear from radiologists is hanging protocols and automation. So the ability to show a raw image versus AI scanned or AI results side by side, or even if this patient had a prior scan and the ability to automatically show a comparison of the current and the prior. And for instance, in the case of if a nodule was detected and just CT, automated comparison, volume doubling time, and all these measurements can be automated. So that's something that we have done as the fourth key takeaway here. And the fifth and the most important one is, if AI generates all this metadata, so how do you convey this to the physician or the radiologist, either by automating their reports, by extracting key findings from the AI and then displaying those results in a structured manner within the context of enterprise imaging, not only with automated reports, but certain analytics as well, so that it helps them understand the intelligence and in terms of how the data is being beat. So few takeaways here. The AI should automate repetitive tasks. So whether it is measurements, annotations, they should be done automatically and this is how we embed them. AI should surface critical findings so that it can be compared automatically. It should help free time for nuanced interpretations and as I said, communication. And then again, as I said, augmented intelligence is not just about automation, it is about collaboration. - So certainly it sounds like AI is more a friend than a foe in the radiologist daily workflow. And then thinking more about the radiologist, we often are behind the scenes and are quite invisible and sometimes we feel the threat of the whole field being commoditized and we're being left out or we're being caught in the turf wars. So how do you see enterprise imaging making the imaging journey more visible to the patients themselves? So that we move from being behind the scenes to being part of their personalized care. - That's a really very important discussion because if we think about radiology, radiology has been at the forefront of digital transformation. And other ologies or imaging service lines are learning from radiology how this digital transformation has provided profound experience or profound, I would say, value to the other physicians and clinicians in the patient care pathway journey. So for us, enterprise imaging connects the imaging journey end to end. It is from acquisition to reporting to the patient engagement. And I think this level of visibility is what builds trust. Because think about this, when a patient walks into a diagnostic facility or a hospital based on statistical data, I think 60 to 70% of diagnostic imaging or diagnostic intelligence today for a patient resides in medical images. And as we said, medical imaging is also about pixel intelligence. When a patient walks into the hospital, a physician may refer them to either the lab or for histopathology or for radiology for some level of investigation. And now what happens as a result is that the patient receives three reports. There is a radiology report, there may be a lab result and there may be a histopathology report. And both the patient and the physicians have to make sense out of that report. And that's where with the power of AI automation and the data that resides in enterprise imaging on a common or a single platform, the patients will get more and more access to their imaging and reports directly because the reports and the results will be more, I would say intelligent to help them get better educated and be more compliant because I feel we believe that patient engagement improves compliance. So if enterprise imaging systems have those tools for better collaboration for radiologists, physicians and the patients, that's where the patients will get more directly involved with it. So patient-centered reporting is on the rise. And that's where intelligent enterprise imaging systems will need to be building those capabilities so that the patients need to interact directly. That's one. I think the second aspect of this is the structured aspect of explainable summaries that can help improve patient understanding. When a patient receives radiology report, sometimes they do not have an understanding of what they're reading. There may be a final impression and some recommendations. But that's where there are requirements now that are being built by luminaries, sites and customers how they can engage referring physicians and patients with some of these explainable or patient-engaging summaries. So radiology will become part of the conversation, not a hidden service, by supporting this multidisciplinary care. And that's where, you know, whether it is MDTs or tumor board and some of these multidisciplinary conversations, that's where we are seeing the shift towards radiology coming more out and to the forefront. And then I think it is all about extensible engagement. What does that mean? Like imaging data that patients need access to with clinicians and they all learn from each other and kind of collaboratively work to foster that level of collaboration. - And earlier you talked about trust, which is critically important in the, in the doctor-patient relationship. And with all these regulatory frameworks that are emerging, how do we build trust that individualized imaging insights are safe, unbiased and clinically meaningful, given the fact that not every AI algorithm is regulatory cleared for the same intended use in different countries? - This is the most important aspect of, you know, building, not just integrating AI, but also building safe systems and eco environments. One of the prime reasons we shifted from PACS to enterprise imaging by building a solution from ground up was this particular fact, the safety and security of patient data. And being the chief medical officer, I have to look into each and every product, you know, Linai Trump, the feature functionality that is being developed from a patient safety perspective as well. And beyond the quality and the regulatory perspective, I think trust is earned through transparency, governance, and as we say, safety. We cannot have, you know, one size fits all AI because there is a lot of regulatory diversity, as you mentioned. What I mean by that is when we look into the regulatory framework in the US with FDA, where I am based out in Canada, there is Health Canada and in Europe, there's the CE. It reminds us that the validation of AI locally is very important, that's the key, how it's monitored continuously and how patient safety needs to be prioritized. And what do I mean by that? I'll give you an example. A chest X-ray algorithm in Canada or in Europe may have a regulatory clearance for, let's say, 40 or 50 different findings. Whereas a similar solution by the similar provider may have regulatory clearance in the US for only four or five findings. So in Europe, they might be able to show where those specific findings are on that X-ray. In the US, the same provider may not be able to show where those specific findings are in the X-ray. They may be only regulatory cleared for triage and modification, which means show yes or no, or false or negative or something like that. Which also brings into perspective how we, as an enterprise imaging solution provider, need to be careful about how we integrate these multiple solutions from one market to the other. 'Cause we have the same enterprise imaging solution for the entire market, international, with respect of US, Canada, or Europe, or Australia. But the algorithms that we will be integrating are different. So their intended use is actually different. So that's where there are four key aspects that need to be kept in mind when it comes to deploying AI into the clinical environment. Number one is the continuous performance monitoring post-deployment. Because that's where we have seen challenges where clinicians had a different, or radiologists had a different expectation from how this AI is going to work in their environment. And when they went live after one month of use or two months of use, they were like, oh, no, this is not performing the way I expected. And there may be multiple reasons, because I mentioned at the very beginning of our conversations. The challenge is that I've seen in the industry is that AI is marketed for its feature functionality, not for its value that it is bringing. So it is very important for the end users to understand what is the intended use of the application. That's number one. What is the regulatory clearance? Because that's where they will build, okay, expectations around this. So that's number two, the vendor accountability. So that's where vendor accountability also comes in. Before you start onboarding, you need to understand what is the intended use of that application and what is going to be the expected behavior of this application and the audit trail around it. And the third important aspect is the communication with the end users. And that's where we have seen that what we have done is developed a kind of a collaborative framework with our vendor partners, the AI partners and developers, with our customers. And we have created a mechanism or governance around before go live and after go live. So how AI is tested and evaluated before the hospital goes live, so that they have a better understanding of how AI will perform. So that's all. I think the fourth key takeaway here is explainability, because explainability builds confidence, both from a regulatory and clinical perspective. But I think that's how the enterprise imaging solution provider, like what we do at ACFA, the customer, and then the AI partner, they need to team up and work together to build that trust and confidence together. - So far enough discussion will talk a lot about what enterprise imaging can do for us today. The title of the podcast is about 2030 and beyond. So if we are to look five years ahead, what will the individual radiology report or care pathway look like, and how will enterprise imaging help realize that vision? - Oh, interesting, because the way we see the industry moving forward and my feedback here is going to be based on not just what we are doing at ACFA healthcare with our enterprise imaging solution, but also what we are seeing the shifting trends in terms of modalities that are becoming more and more advanced. So that is one perspective, and as we see innovations, so AI, last year, did we hear anything about AI agents? No, this year, any other conference that I've attended, everyone is talking about AI agents. And I would say enterprise imaging already had built in agents, which in certain cases are now being referred to as AI agents, because when it comes to automation. So I'd like to say this, that the future is not more about technology. It will be about intelligence and how we use that intelligence. So that means by 2030, radiology reports won't be static PDF documents or reports that are published. I think the reports will be more multi-media, multi-omics, they will be dynamics. They will be data-rich narratives tailored to each patient and the clinician. So the concept of digital twin is going to become more and more realistic. Radiology reports will become more personalized. They will combine imaging, pathology reports, and genomics. So think about this scenario where a mammogram may have been done for a patient with subtle microcalcifications. So today, radiologists then need to decide whether the patient needs to be called for a follow-up or further investigations, MRI. But if that particular exam also provides the patient's genetic profile and red flags this patient for a particular genetic mutation, so the risk profile this patient changes and the radiology engagement and follow-up also kind of gets more personalized. So that's one perspective, the personalization of how reports are going to be done for the patients. Then I think they will be embedded AI context for disease trajectory prediction because prediction is something that is missing today. I think clinical care or radiology care will become more predictive. There will be more, I think, platform consolidation. You would see that these marketplaces that you hear about in the industry, they will either become part of an enterprise imaging solution or they will become more, I would say, seamless how data flows today. Because today, there are certain challenges in terms of integration frameworks. That's number three. And I think, as I mentioned, the fourth and important aspect is embedded intelligence and automation and how we use it is going to reflect upon how enterprise imaging or radiology will look like towards 2030. - And finally, what advice would you give to young radiologists at RSNA who want to embrace individualized imaging while also protecting their own well-being? - I would say they need to stay curious, stay human, because technology will keep evolving, but empathy, context and judgment will always be their differentiators. Because this is what is missing when it comes to the technical aspect, right? So I think as we move towards 2030, the future of radiology won't be defined by AI or AI agents or automation. It will be defined by this augmented intelligence as we speak about. So the radiologists and their insights, the referring physicians, their judgment and the patient story, they will be brought together through this connected intelligent imaging ecosystem as we define it today with enterprise imaging. So they need to learn about, there are certain aspects that they need to be careful about. They need to learn about AI and be more AI literate. So it will become part of their clinical fluency, if I may say. They need to advocate for smarter systems, not just faster ones, because that's where integrated versus embedded user experience comes in. And they need to protect their well-being. The most valuable diagnostic tool, I would think is clear mind. And that's where they need to open up about how, what they learn. And then if we keep in mind the imaging, the individual framework, I think it starts with the self being of the persona. So they should not forget, why they chose medicine. And I think radiology as a practice will become more profound. And the, I would say the experience and the knowledge of radiologists will become more profound because they will become the consultative powerhouses of diagnostic and data intelligence. And I feel that other relevant clinicians and physicians will be consulting radiologists. So I'd say it's not, you know, it's understanding how humanity imaging and they all come together into that framework of automatic intelligence. I think that will be my advice to the young radiologists. - Well, thank you very much, Dr. Ahmed, for sharing your insights on enterprise imaging 2030 and beyond and how we can leverage AI and these other technologies to help improve workflow efficiency, patient-centered imaging and also our wellbeing. Thank you very much. - It was an absolute pleasure. And I look forward to seeing some of you at RSNA as well. - Thank you. Well, this concludes our interview. Please subscribe today on Apple Podcasts, YouTube Music or Spotify. Come back next week for our new episode. Take care.