Good morning, everybody. I'm Jane Mather. I'm the SVP of Corporate Communications for RadNet, and I'd like to welcome you all to RadNet's Investor Day 2025 here at the Nasdaq Marketplace here in New York City. We are thrilled to have so many here in person, as well as those joining us via the webcast. Please take a moment to silence your phones, and we're going to, as we get started. Before we begin, I'd like to note that today's presentations include forward-looking statements which are subject to risks and uncertainties described in our SEC filings. Can we pull that forward, please? I want to make sure everybody can read it. Okay. Oh, I have the clicker. I didn't realize it's the only slide I had. I didn't realize it. Is it this one?
That's the other one. Thank you. We have a full and exciting agenda ahead featuring presentations from our leadership team, our physician experts who will share our insights to strategy, financial performance, clinical innovation, and long-term growth initiatives. Additionally, we have two scheduled Q&As that will follow the presentations throughout the day. During our lunch break, at the end of each of the sessions, or at the end of the session, we will also have product demonstrations available to those on site to experience firsthand some of RadNet's latest technologies and digital solutions. Before we start the presentations, we'd like to share a short video that captures who we are and what drives us. Following the video, we'll be joined by Dr. Howard Berger, President and CEO of RadNet, who will officially kick off the day with his opening remarks. Let's begin.
Imaging. Waves of information caught, categorized, and visualized into an assortment of pixels assembled on screens. Pixels that can find cancers sooner and even predict the risk of disease. It can be easy to forget that these pixels are not just organs and bones. These pixels are people. People with a story. A story with laughter, with love, and with struggles, with walks through the familiar and steps into the unknown.
I came in, and I just figured it would be like every other mammogram. They offered an AI review of my scans called EBCD. The AI and EBCD with a radiologist found a very small DCIS cancer.
It all hits you at once. It hit me pretty, just eye-opening experience.
I wasn't in a hurry until I had a pain and did a self-exam and found a lump.
I think about the tech, like especially the ultrasound techs and the doctor that was there when I was getting my biopsy. I just can't even describe like how much I appreciate them and how much they made me feel so calm in such a devastating time. They went above and beyond to make me feel that I was also like maybe their family member.
I consider my relationship to RadNet a partnership, and I'm very proud and very happy of where that partnership has gone. We're able to do things like take care of our patients who have such extraordinary circumstances and get on that road to better.
From the time I walked through those doors, it just was amazing. Now I could cry with tears of joy because I made it. I'm here. I choose to live every day. You know, no one has a promise of tomorrow. I just choose to make every day count and be thankful for every day that I have.
Good morning and welcome to RadNet's first Investor Day. The question is, why today? What's different today than yesterday, or two years ago, or five years ago? The world's different. We've changed. We've transitioned. We've adapted to circumstances that probably none of us could have foreseen. That started five years ago with COVID. COVID was something that affected everybody on this planet, but it particularly impacted the labor force. People retired. People decided they did not necessarily want to work in healthcare or the premiums that were forced to pay to keep the good talent that we've had rose dramatically. The cost basis and the structure of how all of healthcare, and for that matter, perhaps so many of the industries that we think about, became and needed to develop a new model. Two years ago, or thereabout, something else changed.
Something else fundamentally that RadNet was determined to become a major player in this industry. Healthcare is under transition. Hopefully, that slide is up. If we could put that slide up. Oh, I have to use the button. Okay. Okay. So much for being a technology company. I've tried to list the items that are not necessarily unique to RadNet, but what all of healthcare finds itself exposed to. An increasing shift away from hospitals to ambulatory settings. Notice I did not put on there just outpatient imaging centers. You talk about surgery centers. You talk about physical therapy. You talk about laboratory. Places where people are looking to get lower cost and more efficient delivery of healthcare, particularly after COVID. People do not like going to hospitals. They did not like it in the beginning, and it became even more pronounced after COVID. Technology.
Technology is a driver, whether we want to embrace it or not, but it is fundamentally changing the way healthcare needs to evolve in order to address the numerous challenges that we face, one of which, again, is a very strained labor force, but other ways of making healthcare that much more affordable and effective in both getting to earlier outcomes and improved results in the diagnosis of diseases. AI is the core of that. RadNet wants to be the leader in changing from reactive detection to proactive prevention. That is what our EBCD program is, our lung screening program, our prostate screening program, our coronary artery CT scanning is about. How do we impact the lives of people to diagnose earlier and provide a better outcome? The core of that is technology. What happened a couple of years ago, maybe three years ago?
Some things that have become buzzwords today. ChatGPT. Cloud storage. Transitioning to cloud-native capabilities. Perhaps most importantly, using AI as a tool to improve diagnostic accuracy and efficiency. That is when RadNet decided to transform itself and use its platform as an IT company into a technology-enabled company. We have already demonstrated the various ways that we can impact the delivery of imaging and healthcare with things like TechLive, our EBCD program, and other tools that we are developing. You will hear from a cross-section of the RadNet leadership today, which is really what we want to focus on, that RadNet has the breadth and depth to become not only the dominant player, which I believe it is already in the radiology and imaging community, but to become a healthcare company. Many of you have heard the term that I have used in the past, that imaging is the gateway to population health.
Population health is the gateway to better health and better outcomes. RadNet wants to be that transitional leader to work across all the various disciplines that we have and use the technology, both from an equipment standpoint and from an AI standpoint, to make that difference. I'm proud of the team that has been assembled and that you'll hear from today, not only from the operational and day-to-day delivering of our imaging services, but from our clinical leadership that has the ability to guide RadNet into the use of these tools, which are profound. Having gone through that journey myself purely as a way to satisfy myself that early detection can be an improvement in both your mental and medical outcome, I'm proud to say that these tools work.
What we'll try to demonstrate to you today is that the success that we've had over the last four years in advancing the company's profitability, its improved efficiency, and things that have addressed the very challenges that have faced us are now something that we can embrace and bring to the next level. I'll look forward to seeing your reactions to the substantial amount of information that we've never presented before. While I call this our first Investor Day, it's probably the most important given the fact that we've not done this in the 40 years that I've been overseeing RadNet, and I think today puts us squarely in the middle of where healthcare is in transition.
I welcome your thoughts both during the Q&A and times that we have to come together, but I stand here proud of the organization that has risen to the level that we are today and feel extraordinarily confident that we can deliver value, as Mark and I talk about in each and every closing call, for all of our shareholders and our stakeholders. I'll leave you with one final thought. The biggest stakeholders that we have are our patients. It's not the people that own our stock. It's the people, as you saw in the video, that come to our centers every day and do not necessarily want to be there, but are welcomed with compassion and sensitivity that we're there to help them on that journey.
I'm delighted that you're here listening to this to be a part of that journey and hope that we can satisfy you with the confidence and enthusiasm to lead RadNet forward into a future that has unlimited potential. Thank you.
Let's see if I can get this clicker thing right. There we go. First of all, I want to thank everyone for coming today. It's a proud moment for the company, and I know you have a lot of things that compete for your time on a daily basis, and it means a lot to us that you're here supporting the company and listening to our story. I also want to thank you for your trust. You know, many of you are existing shareholders. Some of you may become shareholders in the future. We'd like to, we as management team, the board, the employees are actually the largest shareholders of the company. We stand next to you, and we're working hard, and I think you'll see that with the group that we've assembled here today to be good stewards of your capital.
We understand the trust that you've instilled upon us to shepherd your investments into the future. What I'd like to do first is just talk about some highlights from our third quarter results, as many of you may have seen. Yesterday, we had our earnings call. Sunday night, we put out our earnings release for the third quarter. It was another record quarter for the company, both in terms of revenue and EBITDA. Our revenue grew double digits like it has been for a number of quarters now. We grew 13.4% relative to last year's third quarter. Our digital health division, which is, as you're all aware, is the newer part of our company, grew 51.6% relative to last year's third quarter. EBITDA also had double-digit growth. We grew 15.2% relative to last year.
Our margin, we also demonstrated margin improvement of 26 basis points relative to last year. How did we do this? We have done it in many ways, and I'll go through it on the next slide, but procedure volumes continue to be very strong. What we're proud to show you today is that we're not just waiting for business to come to us. There are many initiatives across the company that are driving procedure volumes, driving capacity that allow us to continue to grow this business into the future. MRI grew 14.8% relative to last year on an aggregate basis. On a same-center basis, it grew 11.5% relative to last year. We'll talk about some of the things that we're doing inside of MRI to continue to build capacity and drive patient volume. CT also 9.4% aggregate growth relative to last year, 6.7% same-center.
PET/CT, which has been our shining star for a while, and Dr. Rose will be talking about it a little bit later in the presentation, grew over 21% relative to last year and almost 15% on a same-center basis. As Dr. Berger mentioned, we are well-positioned for the future, and that goes without saying. Our balance sheet is strong. We ended the quarter with over $800 million of cash and extremely low leverage. We are on a net debt basis, net debt to Adjusted EBITDA of about one times. We have a lot of capacity to continue to grow this business. Finally, on the strength of this third quarter, as well as the performance in the first two quarters, we increased our guidance, again, both for the imaging center business from a revenue and EBITDA perspective, as well as digital health. What drove this performance?
This is a great slide in the sense that what this really is is a setup slide for the rest of today where we're going to do a deep dive, both from a clinical standpoint as well as an operational standpoint, into many of these points that are on this slide. First, we continue to upgrade equipment and refine our operating protocols and clinical protocols to help reduce scan time and therefore increase capacity. Something that you're going to hear Steve Fortheuber and Norman Haines, our Chief Operating Officers, explain later, and we had a demonstration out in the lobby, is around TechLive and some of the technology innovations that are helping to increase scanning time and lower room closures.
We're going to talk a little bit later in the presentation about some of the predictive modeling that we do around scheduling, which has allowed us to make sure that all of our time slots are being utilized by our patients and that we have enough patients and overflow patients to fill what otherwise would be unutilized scan times from no-shows from some of our patients. de novo centers, we continue to build de novo centers. You'll see later on in the presentation a graph of what we built over the last few years. We've opened up over five centers this year. We've got several more centers left to go in the fourth quarter. We've got 11 de novo projects on the docket for next year. We are continuing to drive capacity to meet the demand that's out there in the marketplace.
Tucking acquisitions continues to be something that will be a part of our growth algorithm. This industry will continue to consolidate. We're well-positioned to be a consolidator. Scale is everything in healthcare today. We've demonstrated that we can be a low-cost provider of services, and scale allows us to be able to do that, and you'll hear more about that later on in the presentation. Another initiative that has helped us grow our revenue and be successful and grow our margins is that we continue to drive reimbursement discussions. Our clustered approach across the country has allowed us to have a seat at the table with commercial payers. They are recognizing more and more that we are their partner in trying to drive this business out of the more expensive hospitals into ambulatory sites of care that are much lower cost, number one, and better patient care.
I think that they're recognizing that on the reimbursement side. As well as in our capitation business, we continue to get nice pricing increases. We've been expanding joint ventures with existing partners. We'll have some things to announce in the coming quarters about new joint venture partners, potentially new geographies that will go in partnership with hospitals. Finally, last but certainly not least, and I think we'll probably be talking more about digital health today than the services business. Digital health continues to drive both revenue with third-party customers, but as importantly or more importantly, it continues to drive efficiencies within RadNet. You'll hear Kay, Shyam, and their team talk a lot about the solutions that will continue to make our centers more efficient, more productive, and increase capacity.
With that introduction, and you'll see me later on in the presentation because at the end, I'll close with giving you a little bit more information about a medium-term or three-year outlook on the business. I'd like to turn the presentation over to Dr. Greg Sorensen. Dr. Sorensen is RadNet's Chief Science Officer. He's a neuroradiologist by training, trained at Harvard Mass General, was a professor at Harvard Medical School, was the CEO of Siemens North America. What else did you do? Oh, he was the founder of Deep Health Breast. And I'll turn it over to you.
Thanks, Mark. Thank you very much, Mark, for way too many kind words. It's great to be here, great to see so many friendly faces. Thanks for joining us and giving us some of your time this morning. I'm going to just say a very brief introduction to our seven clinical speakers. We wanted to introduce these physicians to you over the next hour to have each of them talk about their specialty in depth. There are really three ideas that I hope you'll take away from the big picture here. First of all, I hope you'll get a sense of the clinical excellence that we have, the depth and the breadth of the skills that provide clinical care, and give me confidence to get my own care at a RadNet facility. Some of these physicians have taken care of me and my loved ones.
You can see why referring clinicians are comfortable sending more and more of their volumes to RadNet. Second, I think you'll get a sense for why, from an investor perspective, our advanced imaging procedure volume is growing. You'll get a sense and a flavor for the power of modern technologies and how these are providing better and better insights into patient care, and therefore there's demand for those advanced technologies. Third, I think you'll start to get a sense of how we are, to Dr. Berger's point, moving towards a population health by finding disease earlier and earlier in its stages. This, again, from an investor perspective, suggests a very broad total addressable market. We're looking at the entire populations, not just people who come to our emergency room or not just people who show up at a cancer center already with a disease.
Each of these, I think, will give you a flavor for why RadNet's growth that we've seen in the past years is going to continue and perhaps even accelerate. With that, let me turn it over to our first speaker, Dr. Susie Bash, a fellow neuroradiologist. Take it away, Susie.
Thank you, Dr. Sorensen. My name is Suzie Bash, and let's talk about advanced neuroimaging at RadNet. AI innovations like deep learning for image reconstruction and quantitative MRI are advancing workflow efficiency and early disease detection. It is really transforming population health into a data-driven, patient-centric, precision-based practice. RadNet is uniquely positioned at the forefront of this industry shift, really unlocking scalable growth opportunities. Deep learning for image reconstruction allows 50%-75% faster MRIs while boosting image quality. It will take a routine MRI exam slot from 30 to 40 minutes and basically cut it in half. This is what we have done at RadNet. It will take an image like what you see here. If you look closely at this image, and this is after you apply deep learning, a significant boost in image quality.
Thirty percent of patients have a severe anxiety reaction when they go in the MRI scanner. For us to be able to image these patients much faster, that is their number one—that's their best patient experience—is to get in and out faster. RadNet was actually the first major imaging enterprise to adopt this AI tool at scale. It enhances patient comfort, improves image quality, and drives brand loyalty. Here again is a routine image of the brain, and this is after we apply deep learning. You see the significant boost in image quality. It can even enhance things like detection of early brain metastases. In this example here, this patient has three little enhancing metastases. You can barely see them. I dropped an arrow on one.
When you use this AI tool, you can go 53% faster on the image acquisition, but now you can easily see these three metastases. This can make a difference in a patient's life. We're getting 30%-55% time savings per RadNet MRI exam, and that allows us to scan three to four more patients a day per MRI scanner. When you think that we have 389 MRI scanners, you can understand the magnitude of this positive return on investment. We also use an AI tool called quantitative MRI, and we can use it for a lot of different applications. We use it in dementia. We use it to assess for the side effects of medications for Alzheimer's, epilepsy, multiple sclerosis. We can use it for brain tumors, vascular disease, traumatic brain injury, and PET imaging. This is a look at one of our tools here.
This is Deep Health. Here's another tool here, but it calculates the volume of structures of the brain and lesions in the brain. So 7.2 million Americans have Alzheimer's disease. One in three of our seniors will die of dementia, and neuroimaging is playing a critical role here. We've noticed a big increase in our MRI brain volumes because these patients on treatment have to have at least five MRIs. They'll probably get two additional unscheduled ones. A huge increase in our amyloid PET volumes because we must confirm the presence of that toxic amyloid plaque before a patient can go on therapy, and a big increase in our quantitative MRI volumes. Of the 7.2 million Americans that have Alzheimer's, we think probably 1.5 are going to be treatment candidates, times seven scans. You're looking at 10.5 million new MRIs per year in the U.S. alone.
That is reflected in our volumes. 12% increase in our MRI brain volumes, almost 339,000 last year. Our quantitative MRI volumes are up 206%, over 24,000 last year. This is the Alzheimer's spectrum here. These biomarkers that you see in pink here can all be detected with imaging. Stage three and stage four is our treatment window, okay? Unfortunately, over 50% of patients are diagnosed too late. What we try to do, our main goal here at RadNet, is to diagnose the disease as early as possible with amyloid PET and also provide excellent disease staging. With our imaging initiatives, I believe we are leading the country in this initiative. The reason why it is so important is if you can diagnose a patient early, they do much better on treatment.
76% will have no cognitive decline, and 60% will actually have cognitive improvement if you catch them early on in that window. We use it for eligibility screening, diagnosis, and for treatment safety monitoring. For eligibility screening, you see this AI tool is calculating how much microvascular ischemic disease in the brain. It's also calculating and finding all the microhemorrhages, the little bleeds that can occur in the brain. This must go in our eligibility MRI reports, and this AI tool can save time doing that. This AI tool also helps us assess an Alzheimer's pattern from a non-Alzheimer's pattern, and it actually allows us to quantify the exact amount of that toxic amyloid plaque in the brain at the time of diagnosis. It even aids in Alzheimer's disease staging by assessing how advanced that brain volume loss is at the time of diagnosis.
We can use this for safety monitoring. This is a side effect that some people can get with treatment. Here is a patient of ours that did develop a side effect to treatment, and you see the AI tool is actually segmenting what we call edema in the brain here, and it is doing it all for you. It also automatically segments the number of microhemorrhages in the brain. It measures the largest lesion, and it can provide a report like this, which can give automatic grading. This is very important because the neurologist needs to know what the grading is because they might need to stop or suspend treatment until this resolves. Having this kind of tool can really save a lot of time. We can use it in multiple sclerosis to look for plaques in the brain.
It can automatically detect the new or enlarging or shrinking plaques. This is a type of image that we would get on our MRI scanner. This is Deep Health's product, which is calculating that white matter disease in the brain, and it improves our detection of disease activity. Actually, if you use this AI tool, it will jump from 24% to 76% increase in detection of disease activity. This would actually trigger a change in medication for the patient. It also improves our reliability of reading and allows us to read 62% more reports per hour when you use this AI tool. Now, we can also use it for things like traumatic brain injury, where the AI tool will actually find the different types of bleeds in the brain and color-code it for you.
Here's a patient of ours that had multiple areas of hemorrhage in the brain, and you can see this AI tool is finding that blood product for you, segmenting it for you, and tracking it in a hemorrhage report here. It tells you what's new and what's progressing. Similarly, for oncology, we can use this type of tool for preoperative and postoperative segmentation. You see here of a patient with a brain tumor. It will plot the amount of enhancing tumor over time and the overall lesion. We can use this for metastases. This would save a large amount of time. It actually finds all the metastases for you and calculates the volume. I believe the future of tumor imaging is volumetrics. This is the type of report you get. It will tell you the number of lesions, how many are new, how many have resolved, etc.
Meningiomas, our most common brain tumor. Again, it can find it, segment it, measure the volume for you, and you can track that over time. In summary, RadNet will continue to lead, grow, and redefine deep learning-based AI innovations in the imaging industry by enhancing workflow efficiency and, most importantly, patient-centric care. As Alzheimer's and neurodegenerative imaging accelerates, especially in the era of anti-amyloid therapy, RadNet is uniquely positioned to capture this growth and advance precision-based neuroimaging. Thank you so much.
Good morning, everybody. I'm Sam Hare. I'm the CEO of the HLH Imaging Group and a practicing thoracic radiologist, and I'm going to talk to you about lung cancer screening in the U.K. and beyond. It's really a poignant moment to be here during Lung Cancer Awareness Month, but we need to stop being aware about lung cancer and do something about it. Hopefully, I'm going to demonstrate to you how we can do that. Lung cancer is the commonest cause of cancer death globally. One in five of all cancer deaths are due to lung cancer. In 2022, in the U.K., around 70% of lung cancers were diagnosed at late stage. Now, I've been a chest radiologist for 15 years. These graphs, I'm sort of sick of seeing them because the lung cancer equation is rather simple.
All the graphs show is that if you diagnose lung cancer early, net survival and outlook, even cure, is possible. All of this is happening around us, despite the fact that we have evidence from multiple scientific lung cancer screening trials around the world, be that the U.S., so here, be that Europe, be that England, that lung cancer screening saves lives, and it saves lives by reducing deaths from lung cancer. Thankfully, England decided to stop beating around the bush and decided to move us out of the science realm into the real healthcare setting realm. How did they do this? What they said was, "Let's give the poorest, most deprived areas in England some money to set up so-called lung cancer screening pilots.
Let's try and demonstrate, does lung cancer screening work in a real healthcare setting, or is it just science? What they did was they invited ever-smokers by liaising with primary care physicians between the ages of 55 - 74 and invited them to come and have an assessment and a low-dose CT scan at two-year intervals to see if they had lung cancer. They devised a lung cancer screening protocol, which is a national protocol, and they did excellent communications and marketing around lung cancer screening to get patients to be aware of this. What they also did was recognize that radiologists, particularly in the U.K., were burdened with too much work already, were drowning in scans. Crucially, the English Lung Cancer Screening Committee decided, "We need to help our radiologists.
Let's mandate the use of an AI second read in lung cancer screening to help our radiologists be faster, more efficient, and more accurate. Now, I'm not a baseball fan, but I am standing on U.S. soil, so I want to just borrow a small quote from a famous movie, which says, "If you build something right, the people will come." That is what England and the U.K. have managed to do. That is demonstrated by our uptake rates in lung cancer, which are upwards of 40%-50% and growing. You can compare that to the U.S., where uptake rates in lung cancer screening are less than 6%. How has the U.K. done that? The U.K. has done that by providing a template for lung cancer screening for the world, be that Europe and even the U.S..
The way it's done that is by focusing on implementation and logistical challenges. For example, most of our lung cancer screening CT scans happen in the hearts of communities, in the hearts of deprived communities, and we achieve that by taking mobile scanning vans and trucks to these communities so patients can access CT scans easily. Some of the scanning does occur at traditional hospital sites and outpatient diagnostic centers, but the bulk of it is on mobile units. What is HLH? Whilst the U.K. Screening Committee established a lung cancer screening program, there was one question that remained unanswered. Who is going to report all of these extra CT scans? Because we've got enough work already. That's the same question that I asked when co-founding the HLH Imaging Group or Heart and Lung Imaging with my fellow esteemed chest radiology co-founders, of which we are four.
What is HLH? HLH is actually a network, a community of chest radiologists all around England that are connected up in a cloud-native cutting-edge IT infrastructure. What that allows us to do is think outside the box. We are no longer radiologists working in our individual silos at our individual hospitals, but we are working nationwide in cloud. All of our scans are AI-powered to help us be faster, better, and more accurate in our diagnosis. We now are over 170 chest radiologists reporting more than 30,000 lung cancer screening CTs a month. We have a national architecture, which means we can supply governmental bodies data to monitor the outcomes of lung cancer screening, but we can also perform national training, education, and research. The philosophy of the HLH Group or Heart and Lung Imaging, as we were known, is very simple.
If you take AI but harness that to not a general radiologist, but an expert chest radiologist, you have an unbeatable, unparalleled combination. Because what we've shown is these mandated recall rates from NHS England, we've managed to reduce them way beyond the mandated rates to an even lower rate. That is extremely helpful to government because it makes the lung cancer screening program even more economically viable because our recall rates are that much lower by taking an expert and almost boosting them and boosting their performance with AI. The journey of HLH or the HLH Group is intimately entwined with the lung cancer screening program in the U.K. I am going to take you through it. In 2019, screening for lung cancer was not recommended by our governmental bodies. In 2020, Heart and Lung Imaging was formed to deliver the reporting of these low-dose CT scans for England.
This was about the same time that RadNet acquired multiple cutting-edge leading AI tools, particularly a Deep Health lung AI tool, which is the most commonly used AI tool in lung cancer screening in the U.K. In 2022, RadNet acquired a 75% interest in Heart and Lung Imaging. In 2023, thanks largely and in part due to a lot of the work that we did by reporting all these CT scans, the governmental body then deemed that lung cancer screening is viable, and they recommended lung cancer screening as a national program for the U.K. We rebranded in 2024 to the HLH Group, and I'm going to tell you exactly why and the market opportunity from that on my penultimate slide. The U.K. should be considered as a global exemplar for implementation strategy in lung cancer. The size of the market is massive.
We're still in the red circle, so around 20% of the eligible population has been screened. Of that, HLH does 85%-90% of the market share of reporting. With planned 100% rollout by 2030, that materially means around 1.2 million CT scans for lung screening every year, year on year, all powered by AI, and that doesn't even include the population of the dissolved nations of Scotland and Wales, so the population size will be even bigger. How have we done this? I'm going to take you through our journey. We've done this through frictionless IT that moves scans from scanning vans or hospitals to our radiologists, and they can report them at scale from home, from hospitals. We have now garnered the world's largest lung cancer screening reporting experience with now actually over 800,000 lung cancer screening studies reported.
What we've built is a platform and scaffold infrastructure that covers the entire nation in terms of our frictionless IT architecture. What does that allow us to do? It allows us to report scans with resilience at scale. If you combine an expert radiologist, AI, and frictionless IT, you can report 30,000 lung cancer screening CTs a month, but crucially, you can do it fast. We've got an average reporting turnaround time at this program of 44.2 hours. That's reporting 30,000 scans a month in under 48 hours per scan, which is unparalleled and unheard of in U.K. radiology and probably radiology around the world. As Dr. Berger said, our stakeholders are our patients. Does this all actually make any material difference whatsoever? The answer is yes.
We are causing a seismic stage shift in the way that we diagnose lung cancer in England and the U.K., with over 70% of cancers now diagnosed at early stage, and actually over 50% of them are being diagnosed at stage one. Now, this is a space in lung cancer radiology that I didn't think I would see in my early career as a chest radiologist, but now this is where we wanted to be: true early diagnosis with over 50% of lung cancers being diagnosed at stage one and at scale and saving lives. I talked to you about why did we rebrand? We rebranded because this model for lung could theoretically be syndicated across other body systems: neuro, colon, cardiac modalities where screening is applicable and where prevention is key.
We, as the HLH Group, have solved the issue of bringing an expert reader, harnessing it to AI, and using AI from outside the box at scale across a frictionless IT across an entire nation. That means the market opportunity is that the HLH Group will morph itself into HLH Lung, HLH Heart, HLH Prostate, and that trimes perfectly with the ethos and vision of RadNet. Thank you very much.
Good morning, everyone. I'm Maxine Jochelson , and we are going to discuss our journey with breast cancer today. Breast cancer is the most commonly diagnosed cancer in the world, with 2.3 million cases a year and 700,000 deaths. Lifetime risk is getting higher and higher, and the incidence of breast cancer is increasing by 1% a year, but more so in women who are under the age of 50. We have learned, however, that we can reduce mortality by finding breast cancers earlier. Survival depends on the stage at presentation, and that is our job at RadNet. A sobering thought: my colleagues in breast imaging and I have noticed more and more that we are seeing 20-year-old patients. I saw two in one week, such as this girl who presented with metastatic disease. We really have our jobs cut out for us.
RadNet has major capacities to improve this space. What we can do is first identify which patients are at high risk. We can then choose the correct imaging tests for those patients and implement them. We then accurately interpret them, and we do so using artificial intelligence, which makes us smarter at reading them and more efficient. We can perform an image-based biopsy when needed. How do you identify risk? Over the years, risk was identified by using things such as family history, personal lifestyle. We had different models. The most popular one is a deep learning model called Tyrer-Cuzick. While this was a very good model, we have found that it's really not the best model.
In 2021, the group from MIT presented data on using the actual images of women's breasts to decide the risk that each woman had based on her own breast tissue. They showed that this was significantly better than using these models based on personal history, etc., than we were using for so many years. We know that we are more likely to find high-risk patients earlier on using this model. We at Deep Health, using our AI capabilities, are developing our own model to predict risk based on each woman's breast tissue. We can not only determine her risk, we can figure out which are the best imaging tools and how frequently we can use them so we can have a scientific approach to this. We have to provide the access to imaging.
A lot of this in what you're hearing today is about access. We have mammography as the single only test that has been proven to prevent death from breast cancer. It does this by finding lesions when they're small enough to be successfully treated. It is relatively inexpensive. It is relatively widely available, but there are still limitations that need to be addressed. Some of that is access to people in remote locations or to underserved people. Also, while mammography is great, the sensitivity and accuracy goes down in women who have dense breast tissue. How do we accomplish this? The state-of-the-art mammography units and state-of-the-art ultrasound units that we have at over 400 imaging centers provide 2 million mammograms per year. We have access to all the other imaging tools that help us do our breast imaging better.
We add the robust artificial intelligence that we already have. We really have a trifecta to getting these women diagnosed. The other thing that's important is access. We need to sometimes bring mammography to the patient because she can't come to the units. We have multiple mobile vans, and just in the last three years, we have demonstrated that we can find breast cancers in over 9,000 mammograms in California, and we can do this everywhere. You can also do additional screening on these vans, so you could do lung cancer screening and other types of screenings on these units. A very innovative idea, one that could be replicated in many other places, is you can bring mammography to patients where they shop.
This is what we have done using a super Walmart, and it's a great example of where we attracted women from different communities than you're usually getting to come into your medical centers, people from underprivileged communities. Interestingly, we were able to demonstrate that many more women having their first screening mammogram came to super Walmart than our beautiful breast centers. If it's going to get them there, then it's worth it. 25 million American women have dense breasts, and this is a twofold problem. Number one, if you have dense breast tissue, there is a higher risk of cancer death based on just the tissue and other aspects. In addition, it's harder to find breast cancers. Breast density is white. That's what breast tissue looks like, and breast cancers are white. These cancers are hiding within the breast.
Therefore, when they present, they're larger, and women are more likely to die of breast cancer because they've presented with a larger tumor. We, breast imagers, acknowledge that we need help with other imaging tools. We at RadNet have all the tools. We have ultrasound, which detects an additional four cancers per 1,000, and we have contrast mammography and MRI, which detects an extra 16-20 breast cancers per 1,000 women. Here is just an example of a woman with extremely dense breast tissues. You couldn't find a horse in this breast. Look at the MRI, which elegantly shows a cancer that is curable hiding in that dense breast tissue. Our breast MRI volume has gone up 20%, and I think that will also continue to grow.
We use our AI in many ways, sometimes to be more efficient, but also to try to do a better job reading these mammograms. You can use it because mammograms are digital, and this can enable the tool to go through the—it is a very complex pattern in so many breasts, and it can see through that complexity and find a cancer that I cannot see with my own eye. By doing so, it can also be integrated with the family history, with the genetics, and put this whole patient together in a much more comprehensive way. We have recently published an interesting study based on what we call the ASURE study. You have heard a little bit about the EBCD, the Enhanced Breast Cancer Detection.
What happens is women will all be getting AI when they get their mammogram, and often it matches what the radiologist reads, but occasionally we do not agree. We have a system in which a second radiologist can take a look and break the tie. We have published this data in over 600,000 women, and we showed that doing all of these things, we could find nearly 22% more breast cancers. I have talked a lot about breast cancer and the importance of diagnosing it early, but we have also learned something interesting that a mammogram can show you that we have not been thinking about. Breast cancer is not the leading cause of death in women. It is heart disease. One out of three women dies of heart disease.
What we started seeing is that on a mammogram, you can see calcifications in the arteries within the breast. These calcifications, particularly if you're seeing them in younger women, may very well be predicting that these are the ladies that are going to have significant cardiac disease. Now, it's important to quantify these calcifications, to characterize these calcifications, and this is something that we are also working on to do with our artificial intelligence so that when we get a 40-year-old woman walking in the door looking like an 80-year-old woman, we can go about and say, "This woman is at increase for developing cardiac disease. Let's get her to the right doctors, and let's get her the preventative things that she needs so she will not die of heart disease." In conclusion, RadNet performs over 2 million mammograms a year.
That's 5% of all breast imaging in the U.S. Innumerable lives are saved by finding breast cancers earlier. Many additional lives can be saved, and quality of life can be saved by finding earlier breast cancers and also recognizing those early breast arterial calcifications. We at RadNet continue to invest in improved technology, AI research, and resources to improve morbidity and mortality even further. Thank you for your time.
Thank you, Maxine. Women's imaging is a very robust discipline in radiology. When I trained, there was no such thing as men's health. The NCI currently spends twice as much research money on women's health as it does on men's health. I want to take you on a little journey about what we can do for men with prostate cancer screening. Everybody's heard about the challenges of PSA, and even the person who invented PSA said he doesn't want to get it himself. The reason was it was used incorrectly. Urologists would find a man with an elevated PSA, and the prostate was the only gland that would be biopsied without imaging. They would stick 12 needles without guidance into the prostate, hoping to find cancer by chance. Men didn't like that, and the data showed that the urologists were not very good at finding cancer.
What we're trying to do now is to use PSA in conjunction with prostate MR, and by combining those two techniques, we've become very, very good at finding breast cancer—I mean, prostate cancer. How can we reduce the mortality of prostate cancer? As has been mentioned by several of my other speakers before us, we've gotten very good at finding disease. Every week in my clinic, I find a man coming in for a prostate MR whose PSA is between 50-400, and they're metastatic at presentation. In other words, the disease has spread outside of the prostate. We can't cure those men. The only way we can reduce mortality is to encourage greater participation of men to get screened.
We want them to fight like women and not be afraid of the side effects of what I call I-squared: impotence and incontinence, because current treatment has reduced those side effects, so it's not a big issue. If men don't get screened, if primary care doctors don't order PSA and then an MR, if they do that, we have the ability of reducing mortality by 20%. The statistics of prostate cancer mirror breast cancer. It's really two sides of the same coin. 300,000 men and women are diagnosed each year of their respective diseases, and roughly 30,000-35,000 men and women die each year. The statistics are quite similar. One in eight men will be diagnosed sometime in their lifetime. For minority men or men who have genetic mutations or other risk factors, those odds go up. It's predominantly a disease of older men.
60% of the cancers we find are in men over 60 years of age, and it's the second most common death of cancer in men besides lung cancer. We should be able to reduce the number of men dying by 20% or save 7,000 lives. If we screen men between 45 and 80, we can do that. Our program is that if the PSA baseline is over 2.5, we're trying to encourage primary care to order our non-contrast rapid prostate MR, our enhanced prostate screening program. MR is very accurate, and I'll go over some of the data. If the MR suggests a suspicious finding, we're developing partner relationships with urologists. We don't want to do the biopsies. We want the urologist to do the biopsy using ultrasound MR fusion technique, which is the most accurate way of finding cancer.
If we can find the cancers with the PSA below 15, there's an excellent chance of cure and survival. What is the scope of the market? Right now, there are 70 million men in the U.S. between the ages of 45 and 80. Only one in three men in this age group get a PSA. The primary care doctors are confused because the guidelines on prostate screening keep changing from 2008 to 2025. The government, the U.S. Preventative Task Force guidelines recommended no PSA screening because the data showed it was not cost-effective. The urologists were not using it correctly, but they failed to incorporate the benefit of using prostate MRI. Now the NCCN guidelines are showing the benefit of this strategy. RadNet's prostate program, when I started training, there was no such thing as men's health or prostate imaging.
I was lucky I've been introduced to some of the world's experts in prostate imaging back in 2009, and Dr. Berger allowed us to start our prostate program, and now we have 16 years of experience, and we're performing more prostate MRs than any site in the country. We're doing over 1,600 prostate MRs a month on the West Coast and similar numbers on the East Coast, and we perform more of these exams than anybody in the entire country. The reason we get this volume is because we focus on concentrated subspecialty readers that maintain our accuracy, and the urologists have confidence that we do a good job. If you look at this chart, you can see the big vertical arrow is when PSA screening started to become available in the U.S., and you can see the reduction of death rate from PSA screening.
The stars are what happens when we add prostate MR to that, and it shows an improvement in reducing mortality. The current NCCN guidelines now say that in 2023, that MR is strongly recommended for screening if available, and now MR targeting is preferred. Before any man is scheduled for a biopsy, they should get an MR first. This serves several purposes. One, it allows us to tell the urologist where to stick the needle. More importantly, our negative predictive value of prostate MR is very high. It's over 90%. Many men with an elevated PSA can avoid a biopsy because we say the gland looks normal. For prostate MR, we can do screening, tumor detection, targeted biopsy. We can stage the tumor. There is now MR-guided ultrasound treatment, and we do follow-up studies.
For people who do not know what prostate cancer looks like, on a T2-weighted image, we look at the dark spot marked with the white arrow. Diffusion imaging is a measure of cell density or cell packing. The more aggressive tumors have more restricted water. We can measure that, and they also show increased tumor enhancement. The slide on the left side is an early cancer. This is the same patient, and we can actually measure the growth of this patient's cancer. PI-RADS is how we report prostate imaging. It mirrors what we do with breast imaging with PI-RADS, but what this does is it allows us to use a structured reporting tool, and it allows AI to work more effectively.
Because of a lot of data, this shows why MR first shows a proven benefit: that we can reduce 28% of unnecessary benign biopsies by using MR. We can find more clinically significant cancers with prostate MR than with the standard blind biopsy technique. We find fewer cancers that do not need to be treated, and we can do it with putting fewer needles in the men. Unlike breast cancer, low-grade prostate cancer does not always need to be treated. We can monitor those men on active surveillance. We do not do that in women at this time. If we add MR, we can triage men who would benefit from biopsy. We can reduce the number of men getting an unnecessary procedure with risks and harms.
We can find more clinically significant tumors, and then we can assign men to active surveillance who have small-volume, low-grade disease, where we can optimize their individual treatment strategies when we find disease that's gland-contained. This program has been proven to be cost-effective. If we compare what Maxine talked about with breast cancer detection with what we can do for men, we're actually doing a better job. As Maxine said, screening mammography tends to find five to seven cancers per 1,000 women screened. When a woman is sent to a biopsy from a mammogram, they only find cancer maybe a third of the time, but screening has reduced mortality from breast cancer. If we do prostate screening with PSA and MR, we can find clinically significant cancer in 70%-90% of those men that we send to biopsy.
Significantly better for men now than women, and we find more clinically significant cancer, so we can do a better job, but why have not we? The evidence is that we should even start screening younger men who are at risk. Age should not matter. Quality of life should matter. The PSA at age 45, if your PSA is less than one, your likelihood of developing significant cancer is very low for the next five years. What makes our program work is this data shows that we can do a fast non-contrast exam, and it is non-inferior to a full diagnostic prostate MR with contrast. That allows us to offer our enhanced prostate screening program with tools from Deep Health and our AI. This allows expert readers to even increase their accuracy, similar to what we are doing with our breast program.
There are not enough qualified prostate readers, so this will also help less experienced prostate readers get their accuracy to approximate those of expert readers. This allows us to be more efficient because we're talking about driving volume. We need to get these cases read, and the SAGE Deep Health Prostate Program has a streamlined reporting tool, and it allows us to do targeting. We can keep this cost similar to what we do for screening mammography. RadNet is providing this for self-referred men for $275. Here is a reference list to document the data that I showed, and I want to thank you for your time. Next up is Dr. Judy Rose.
Good morning. PET/CT tracer trending and future growth. PET/CT is RadNet's fastest growing modality. From Q3 2024 to Q3 2025, we had a 20.3% increase in PET, and our new advanced tracers, which are about what this talk is mostly about, represented 8% of our amyloid tracer. Alzheimer's was 8%, and the PSMA tracer, which was for prostate cancer, was 13%. Look at our volumes by tracer. We have the oncology tracer, the standard FDG, in that period of time from Q1 2023 to Q3 this year. We've gone up 33%. 140% of prostate images, PET scans, increased and 6,800% for our amyloid Alzheimer's tracers. To give you a little background, what is an FDG tracer PET scan? First of all, we're going to inject a tracer, which is radioactive glucose.
Because that goes to areas of high glucose metabolism, which most cancers have, the PET/CT scanner can see those cancers. The limitations are that from 2000 until recently, this was the only tracer. Every patient, every cancer, breast, lung, prostate, got the same tracer. FDG is not cancer-type specific, and it is also positive in infection and inflammation. What is our future opportunity? Cancer-type specific tracers. We had one tracer for everybody, and now we have new disease-specific tracers that could be for prostate cancer, such as Pylarify, neuroendocrine cancers, such as NETSPOT, and Alzheimer's, which is Neuraseq and Amivit. PET/CT has entered a whole new era with tracer expansion beyond FDG. If you can see it, you can treat it. These new tracers are driving growth and disease-specific treatment called something called theranostics. What is theranostics?
It's the pairing of a diagnostic PET/CT and a targeted therapy using the same molecular pathway. How does it work? You have a PET/CT scan, and it identifies the tumor using a specific cancer tracer, and then that pathway is used to deliver a targeted radioactive treatment specific to that cancer. Why does it matter? Because personalized treatment matches the therapy to a specific cancer. This diagram is a bullseye of all the theranostic companies currently looking for a therapeutic. Why does that matter to us? Because every one of those that's successful has to have a PET/CT tracer to make sure that they're a candidate for the therapy. What are the—Dr. Bash told you a little bit about this—what are the amyloid and TAU tracers? Amyloid and TAU are both hallmarks of Alzheimer's disease. Amyloid detects amyloid plaque, and TAU detects TAU triangles.
In our portfolio, in 2012, we did our first Alzheimer's research study, and we now have done 50+ phase II to three national trials. Those are for amyloid and for TAU, and we have to screen for inclusion because you can't get a therapy unless the PET/CT scan is positive. We have to monitor for anti-amyloid and anti-TAU therapies to see if they're actually working. The circles you see on the right are all the companies looking for a disease modification for Alzheimer's, and every one of those that's successful is going to need a PET scan for diagnosis and treatment. Here's one of the ones that's in play now, that's TAU. Amyloid gets all the buzz because it's most of the Alzheimer's patients, but 15% are not amyloid positive. They're amyloid negative, so they're not going to respond to amyloid therapy.
There is one TAU tracer approved, but this is one that's in experimental trials now. Okay, that's where we are now. What are the drivers of future growth in amyloid PET? The demographics, aging population, projected by 2060, 13.8 million Americans will have Alzheimer's. We're diagnosing it earlier. That broadens the eligible patient pool for treatment. We have blood biomarker tests, which will expand PET as a pretreatment exam, and the therapeutics. We've got Leqembi and Kisunla, which are the anti-amyloid drugs, and the drug pipeline includes anti-amyloid, anti-TAU, and neuroinflammatory inhibitors. All those patients need PET scans. The reimbursement is favorable. What's a PSMA tracer? Dr. Princeton told you a little bit about PSA, but PSMA is prostate-specific antigen expressed in all prostate tissue. 95% of prostate cancers overexpress PSMA, so now we have a cancer-specific tracer.
These tracers are approved for initial staging of biochemical recurrence. What's the driver of future growth for PSMA PET? Again, one in eight men will be diagnosed with prostate cancer, 313,000 + per year, new cases. The clinical adoption, it's already the current standard of care in prostate cancer management, and there's a large untapped market. The therapeutics, we just talked about scans and therapy. If you look at the two images on the right-hand side, all those black dots on the left-hand picture is a PSMA PET scan before the patient was treated with a theranostic therapeutic. That's Plovikto. After two cycles, you can see the marked improvement. Additional PET/CTs will be needed for therapy selection and monitoring, and the reimbursement is favorable. The next generation of tracers are entering late-stage clinical trials. We talked about the first generation, the new tracers.
We talked about the first generation treatments, both in oncology and neurology. Now the second generation of new PET tracers. There are so many I would have 10 slides, and you would not sit; my time would be up. They are all coming along. The next one you will probably hear about is FAPI. We are going to start that clinical trial this month. We are going to hopefully do the first patient in the United States. What is this? It is a fibroblast activation protein inhibitor. It is overexpressed in cancer-associated fibroblasts, and it binds to 30 different cancers, most of which are not FDG positive. The trial objectives are for gastric, esophageal, and pancreatic cancer.
What you can see is the FDG, the one we always had for everybody, is relatively negative, and you can see the FAPI next to it, and you can see how dramatically positive and what the difference is for those patients. We're very pleased that our academic trial partners for FAPI will include Memorial Sloan Kettering, Stanford, Cleveland Clinic, and many others. Here's the new patient journey with a new cancer-specific tracer. Here's the difference. The prior standard of care would have been a CT, which identified a 5cm pancreatic mass. That's the white arrow. They would have an FDG tracer PET scan, which you can see next to it, and there's no uptake where the white arrow is. They would go to biopsy, and then they would go to surgery. Now we have a new tracer standard of care.
This patient had a NETSPOT scan, which is a new advanced tracer. It's 90% specific for neuroendocrine tumors. This patient skipped the biopsy. The surgeon was confident, based on the NETSPOT exam, that this was limited disease to the pancreas. The patient went to surgery, and the patient was operated on this week, and he had only tumor exactly where that scan is, where the big white arrow is, and the big white dot, which is a cancer-specific PET scan. The question isn't if we can grow. It's how fast. RadNet today, 90,000 PET/CT exams a year, 67 PET/CT scanners, 5.3 scans performed per scanner per day, significant unused capacity so we can handle this volume, and large-scale success with the new tracers, Pylarify, NETSPOT, Neuraseq, Amivit, and a nationally recognized program. That's where we are now. What's the future? We have the tracers. We have the infrastructure.
We're 100% outpatient, and we can leverage our PET/CT network to drive growth and revenue. Thank you. The next speaker is Dr. Micheal.
Thank you, Dr. Rose. Let's talk about cardiovascular imaging, all in eight minutes or less, hopefully. Heart disease. One in five people are going to die of heart disease. It's been at the top of that list for many years. I'm confident that at some time in my career, it's no longer going to be at the top, but it's going to take multiple steps to get there. What's really important to note is it's not just about symptoms. The vast majority of patients do not have symptoms prior to having a heart attack. It's not just walking around the streets of New York, going the 15 blocks that we walked yesterday to dinner, and saying, "You know what? I didn't have any chest pain. I must be fine." That's not the actual facts. In order to make accurate, good investments, you need to have accurate data.
If you were to Google the prevalence of coronary artery disease, most of the data you're going to see says 5%-10%. However, one in five people die of heart disease. How does that make any sense? If you look at the actual data, this happened to be in asymptomatic patients. Keep that in mind. About 50% of those patients over the age of 40 had coronary artery disease. If you were to include symptomatic patients, it would certainly be much higher. In fact, we looked at all the patients who came through our facilities over the last three months, and our most recent data showed numbers much higher than that. That includes screening patients as well as those coming in for symptoms. It is not just 5%-10% of patients affected. It is quite higher.
The old way of looking at coronary artery disease, diagnosing it, EKGs, getting blood tests, everyone in this room has likely had a lipid profile at some point, and hopefully it's in range. We need to move away from looking at indirect surrogate markers and actually looking at the disease process that we're trying to prevent. The old way of looking at things was not very accurate. We have much better technology these days that's going to catch it sooner, allow us to treat the individual patient, and have patient-directed care. Here's a coronary CTA for anyone who may not have seen one before. We can actually see the disease process that we're trying to prevent. We can take a look at whether our treatments are actually becoming effective.
Obviously here on the left here, we have pretty good-looking coronary artery, and on the right, it's pretty severe. We can actually determine how severe the disease process is and help guide management for the next steps. Maybe that patient needs a stent, maybe they need a bypass, or maybe they can just be managed conservatively. It doesn't make a difference. Here's some data from the Scott Hart trial showing that one single coronary CTA had significant impact, 41% decrease. One single coronary CTA. That's not following them serial. This is actually just one CTA in the beginning. The reason for that is multifold, but it's very important to note that the patients in this trial, when they had that one coronary CTA, their doctors were much more likely to be aggressive in medical therapy.
Showing the patients the images on this prior screen here, they're much more likely to be compliant when they can see their actual disease process. Not just saying, "You're higher risk for coronary artery disease," but here are your actual coronary arteries. Here is the disease that we're trying to help prevent from getting worse. It isn't the guidelines. Absolutely. It is really the best exam for looking at the coronary arteries. It is definitely the best for looking at the plaque that is there and monitoring changes over time. As of 2021, class 1A recommendation. It is now standard of care. Not everyone's the same, right? Some people might need to be treated more aggressively. Some can be managed conservatively. Some may not need any medications, yet some may need many.
We need to move away from simply looking at the group and lumping everyone together and focusing on the actual individual. That is where a coronary CTA-driven pathway can really excel. You'll hear often good medicine is good business, but you really do need a good business to provide good medicine, and that's exactly what we're doing here. What has been RadNet's approach? We need to increase access in all markets. There is high demand for this imaging. We standardize all the processes. You want to make sure it's scalable. You cannot just have a workflow that works for one, two, three centers. It needs to work for all markets. Any new market that we enter into, we can replicate our current processes when it comes to quality, when it comes to having subspecialty trained cardiac radiologists.
You're going to hear about TechLive more, using that to make sure you have the best, highest-trained personnel scanning complicated patients. This is really what's going to allow us to expand. Plaque analysis, FFR, great new technologies, also reimbursed CPT codes, very important for business because that is how you can really get the masses to have access to this great technology. We're also using technology to improve accuracy. I will always welcome a second set of eyes, particularly one that's very accurate AI to help us make those detections, as well as decrease turnaround time and to prioritize cases. Here is some of the AI that we use. It'll help us identify areas of stenosis. Going into the case, the radiologist can take a look at the AI. They can agree. They can disagree. It helps to highlight that area. Is it perfect? Absolutely not.
Neither is a single radiologist. Together, certainly more accurate. Let's talk about plaque. There's a lot of media attention to plaque recently. It recently got a category one code beginning in January 2026, reimbursable. We thank CMS for their fair reimbursement related to that, which is really going to allow us to expand in all markets for plaque, really focusing on that in 2026. Here is an example of plaque analysis. I like the colors on here, which is really great showing to patients because that is what referencing back to the Scott Hart trial. When you show patients their actual plaque, that's how you get them to actually be compliant with lifestyle changes and medication. What's really interesting about plaque analysis is you can track it over time.
You can take a scan two, three, four, five years ago and compare it to current scan and say, "You know what? We've just been trying diet and exercise, but your plaque's getting worse. We need to try something else." Or maybe you're just on a statin, and it's starting to slow down a little bit, but it's still progressing. Maybe you need to add in a PCSK9 inhibitor. Either way, you can focus on the individual and not just the group by following plaque changes over time, as well as not just serial scans, but in real time. If someone was to come into our office and have high-risk, non-calcified, low-attenuation plaque, you're going to want to be more aggressive with that patient. It is not just about the calcified plaque burden. It is not just about symptoms. It is certainly just not about lipid profile and surrogate markers.
Take a look at the actual disease that you're trying to treat. But does plaque actually change management? Yes. Looking at our recent trial data, patients who had a coronary CTA with plaque analysis, greater than 50% of the time, their doctors changed management. Greater than 50% of the times. I already said that a coronary CTA is the best examination for looking at the coronary arteries. Great when you add on plaque analysis, changing management 50% of the time, and we're certainly offering that in all markets. What about FFR, fractional flow reserve? This is a way to take the data from that coronary CTA imaging. It's not an additional scan or anything like that, and actually help to see whether or not that narrowing is restricting blood flow.
What's extremely important here is many of us have heard of a cardiac catheterization where they stick a catheter in your groin, inject some dye. It's invasive. Generally try to avoid that. You don't want to go to the cath lab and have someone say, "You know what? It was perfectly fine. It was negative." In that case, you probably shouldn't have been there. You probably should have had a coronary CTA with FFR first. In this situation, it's better for the patient, and it also saves the healthcare system money. You're going to avoid unnecessary scans. You might avoid a stress test, avoid the cath lab. When you do need to go to the cath lab, you're going to make it more efficient. You want to move cath labs from being a diagnostic arena to being one in which you have planned intervention.
When are we getting started? We already have. I could just say every RadNet state, but when I was making this slide, I kind of liked typing them all out, so I left it in here. Every office that has hardware and software capabilities, we've already identified, with many of them already live. Sixty-nine locations are live as of October 25. Now, some of you are probably thinking right now, "How many locations are possible?" We have 240 cardiac, well, 240 CT-capable locations. We are identifying all of those to look at the different regions, the market demands, the scanners that are there, so there is plenty of room for growth. Sixty-nine live right now, 240 CT locations current. I said we've already started. How are we making out? Not too bad. This is a great graph, but something that I really, really want to highlight.
This was actually just the foundation-building phase. When you're starting to build a program, the scale of RadNet's, you need to standardize everything. You need to make sure all the processes are working. Maintaining turnaround time, extremely important. You can't have multiple-day turnaround time for cardiac patients, particularly symptomatic ones. These are all the things that we've been working on. This is just the foundation-building. We are now moving into the growth-building phase. If I was to look back at this chart just a few months ago in terms of the amount of offices that are online, probably about half that as of December last year. We have been adding on new centers every single month. We have multiple new centers planning to come online very, very soon. Where do we go from here? If you can see the future, you can change the outcome.
We're going to continue to empower clinicians and patients. The journey is really still uncertain, but RadNet is going to be a part of it. We're going to continue to expand, assess new data, invest in people, and new technologies. Where do we go from here? We're going to go on that journey together. Thank you all. I appreciate your time. Next up, Dr. Peters. Here you are.
Thank you. Good morning. I'm Robert Peters. I'm the Medical Director for Cardiac Imaging at RadNet East. Today, I'll discuss how cardiac MRI, or CIMAR, is reshaping cardiovascular care driven by clinical need and artificial intelligence. I've organized this talk around three questions: why CIMAR matters, how it's growing, and how AI is accelerating it. CIMAR is a non-invasive, radiation-free imaging technique that evaluates structure, function, perfusion, and tissue composition using magnetic fields and radiofrequency pulses.
It offers superior soft tissue contrast and can quantify fibrosis, inflammation, and scar with precision. Looking at the diagram on the right, clinically, CIMAR solves diagnostic uncertainty around myocarditis, such as post-viral forms in COVID-related diseases, cardiac tumors, viability mapping after myocardial infarction or heart attacks, and valvular or congenital anomalies, especially when echocardiography is inconclusive. In short, CIMAR is a definitive tool for structural and inflammatory heart disease, complementing rather than replacing echo and CT. Why is CIMAR more important than ever? The American Heart Association 2025 data shows a major shift in cardiac disease burden. In 2010, ischemic heart disease accounted for about 75% of cardiovascular deaths. By 2022, it is reduced to 61%. Structural and inflammatory causes, heart failure, hypertensive heart disease, myocarditis have nearly doubled to 24% in that same time.
This reflects better ischemic survival, but rising chronic and inflammatory burden, exactly where CIMAR is most powerful. If we look back even further to the 1970s versus 2022, the change is dramatic. 91% of deaths were ischemic in 1970. Like I said just the last slide, today it is 61% with a larger portion of structural and inflammatory disease at 24%. That is a shift from sudden acute events to chronic complex conditions needing advanced imaging. The annual U.S. diagnosis underscores this change. Heart failure, 1 million new cases a year. Hypertensive heart disease, three-quarters of a million new cases a year. Cardiomyopathy, a quarter million cases per year. Collectively, that is conservatively over 1.5 million new structural or inflammatory patients annually, all potential candidates for cardiac MR. These patients benefit from CIMAR due to its unique ability to provide quantitative evaluation of ventricular function, mass, and fibrosis.
You can see from this slide, CIMAR represents the gold standard for volume and ejection fraction, scar and fibrosis, and perfusion and viability. Echo and CT remain essential, but for tissue characterization, CIMAR sets the standard. Over the past decade, CIMAR has moved from optional, which is nice to have, to guideline-mandated and multiple AHA, ACC documents. Just to highlight these quickly, in 2020, hypertrophic cardiomyopathy became a guideline. In 2021, class one status for chest pain. In 2022, heart failure guideline was added for fibrosis, amyloidosis, and myocarditis. In 2022, a heart failure guideline was also added. Key takeaway: CIMAR is now a core component of evidence-based cardiac care. Guess what? These guidelines are driving real growth for us. Since launching a dedicated cardiac MR practice in early 2023, monthly volume has climbed.
Initially at 15 cases a month in 2021, 40 cases in 2023, and now we are up to 110 cases recently, with the growth being shown here in this slide. To meet this demand and accelerate growth, we're focused on three objectives. We need to recruit cardiac-trained radiologists, roll out advanced AI-driven MR scanning platforms, and adopt AI reporting workstations to deliver rapid and consistent interpretations. What is an AI-driven cardiac MRI approach? We utilize an FDA-cleared platform that automates scan planning and quality control in real time. We find this process cuts sequence acquisition time by about 25% for each set of images, which translates to a reduction in a typical exam time from about 45 minutes down to about 20 minutes. It performs automated slice planning and parameter optimization, and it produces consistent image quality while reducing motion artifacts. This means greater throughput.
One patient per hour becomes three to four patients an hour and a more pleasant patient experience. Not only do we scan patients with AI, but AI is driving interpretation. On the reporting side, we use an AI-driven workstation, which automates segmentation, functional measurement, and structured reporting. This cuts our reading time from about 40 minutes - 20 minutes. It auto-populates key metrics like EF, ejection fraction, strain, and flow. It enables analysis of advanced sequences such as 4D flow, critical for congenital and valvular disease. The result is greater than 50% time savings and consistent high-quality reporting. Cardiac MRI is no longer a niche technology. It's a new standard for structural and inflammatory heart disease. In summary, CIMAR plays a critical role in the evaluation and follow-up of structural inflammatory heart disease. U.S. guidelines have elevated to core status across hypertrophic cardiomyopathy, chest pain, heart failure, and valvular pathways.
Advanced cardiac scanning technology and AI platforms cut scanning and reporting times by greater than 50%. CIMAR is growing in our RadNet practice. Our primary objective is to take our robust CIMAR clinic to scale. Thank you.
Thanks, everyone. We're going to take a three-minute break just because we're going to take the podium down, and then we'll play the video and get started again. We've got a three-minute break. Everybody, we're going to be getting started again. If you could please re-enter the room and take your seats. We'll get started. Everybody in the room, just so you know, we are going to present these slides on our investor page. You can pay attention in the room and not worry about getting copious notes that you need because you'll be able to get the slides.
Demand for imaging is growing.
The need for speed, access, and accuracy has never been greater. RadNet has built a connected network that serves patients, physicians, and communities across the country, expanding access to compassionate, high-quality care and building the capacity to meet tomorrow's demand. Behind every scan is a smarter workflow. AI-powered tools help technologists work efficiently. Radiologists interpret with greater confidence, and patients receive earlier answers and better outcomes. Innovation doesn't replace service; it strengthens it. A mobile mammography fleet, retail centers, and specialty programs bring quality care closer to more people than ever before. Partnerships with leading health systems and technology innovators continue to expand what's possible in diagnostics and care delivery. At the center of it all, people, technologists, radiologists, and teams dedicated to every patient every day. Through focused training and development, the next generation of imaging professionals is ensuring quality, consistency, and compassion.
Over 40 years leading radiology forward, now driving a ship from reactive detection to proactive prevention, making healthcare more accessible, more intelligent, and more human.
Norman and I are going to take a little bit of time this morning to give you some insights into the trends that we see impacting the imaging services portion of our business. We'll go through what we see are the opportunities emerging out of those trends and give you some examples of how we're executing against those opportunities. Let's start by sizing the market again. This is a very large, growing, still fragmented market that conservatively we think is estimated at $100 billion, but probably upwards to $140 billion. For some context, we're just below $2 billion in services here at RadNet.
If you take the other four largest chain operators in the country and you couple them up with us, we're probably only $6 billion or $7 billion of that $100 billion - $140 billion. The point here is it's a large market with plenty of opportunity ahead. Norman is going to talk to us a little bit about the trends that we see developing.
These are the trends that we see, which is creating the shift to outpatient imaging. If I can remind you right now, in the states that we're in, we comprise about 25% of the population, and those states are densely populated. What we see is an aging and growing population, which is a rising demand for screening and diagnostics, especially in our core markets. Workforce strain.
I'm sure you're all aware it's been mentioned multiple times that the shortage of technologists and physicians is really forcing us to increase efficiency and to create efficiency in our operations. Underserved communities. Many areas still lack access to affordable quality imaging. I think it's been previously stated, there's still 30 million women in this country right now that don't have access to screening mammography. That's a big number. Hospital backlogs. Backlogs in hospitals now with inpatients, okay, in emergency rooms that are filled with patients are not allowing them to service their outpatients. Those outpatients are now being pushed to the outpatient imaging environment. Advanced technology.
As you've heard today, new technologies, AI and radiopharmaceutical expansion is really driving our imaging and imaging research at RadNet is being able to provide us the ability to look at these technologies, look at these radiopharmaceuticals, and then put them into our facilities and scale them. Preventative care. The mindset, okay, is being proactive, proactive diagnostics instead of reactive imaging. This is really going to provide, as you will see in our presentation as we go forward, we'll be more proactive and we will increase the number of our screening exams in our fac ilities. Shift to outpatient care. Payers and employers are shifting their volume to lower-cost imaging, okay, and screening. We're getting the benefit of that in our outpatient imaging centers.
What we're seeing coming out of the trends that Norman just went over are really three big opportunities.
The first one is capacity. You have heard all morning about access. We want to create access so that more patients can take advantage of the types of treatment, the procedures that we are talking about. We need to create more capacity on each piece of equipment, from each room, from each technologist. We need to optimize the use of that capacity. We will hopefully give you some examples of how we are doing that. The second one, and it was alluded to first by Dr. Berger in his introduction, is the patient journey. The U.S. healthcare system is extraordinarily complex, very difficult for a patient, probably even us, to figure out how to navigate. We want to make it more welcoming, simpler, easier for a patient to navigate that system.
Likewise, we want to make it easier and more efficient for our team members to help them navigate that system so they can be more efficient and both of them can be more accurate. The accuracy allows us to do the right procedure the first time and make sure that we can compliantly turn those procedures into cash. Lastly, you've heard a lot about proactive care. We want to be very focused on raising awareness for the value and the importance of screening and preventative care. We want to be the leaders in healthcare to shift from reactive detection to proactive prevention. I really believe that these six targets here give us a distinct competitive advantage to create capacity through AI-powered innovation and technology, especially partnered with Deep Health. This gives us the ability and the operational expertise to scale it.
One of the big things at RadNet, which you've already heard today, is remote operations. This was developed by Deep Health, remote MRI operations with TechLive to expand operating hours and address staffing challenges. We've been very successful with that. Capital light investments. We invest in a capital light investment to drive utilization, manage durations, and increase our exam volume. Smart appointment management, smart scheduling tools to optimize appointment time use and strategic expansion. We have continued to grow through De novo, tuck-in acquisitions, and joint venture expansion. Advanced imaging mix shift of use, as you've already seen. We have clinical expertise, especially exam growth to shift mix to advanced imaging and team development, team building and development to enable continued growth. You've heard a lot already this morning. You've heard TechLive mentioned.
One of the best examples we have of creating capacity has been through Deep Health's TechLive tool. Here's what happens here. We take an expert MRI technologist who's typically working remotely somewhere other than one of our imaging centers, and we pair them with what we call an in-suite assistant, essentially a tech aide, who is actually located at the place of service. The remote expert technologist is 100% focused on capturing the best image for our radiologists to interpret. On-site, the tech aide or the in-suite assistant is 100% focused on the patient experience and patient safety. Typically, these remote experts are running up to three MRI units at a time. What this is helping us to do is address the staffing strain that we've talked about this morning. We've demonstrated some pretty impressive results, we think.
year-over-year, we've grown the use of this tool by 57% to the point year to date through September of this year, we've performed over 133,000 MRI scans with this remote technology. Perhaps more importantly, and this was based in our Northeast pilot market or co-creation market, we were able to reduce the number of unstaffed hours. Basically, the unit was sitting there with slots vacant because we couldn't staff based on the staffing challenges. We've been able to reduce that amount year to date in the Northeast by 41%, creating enormous capacity to drive more access, greater utilization, more revenue, allow for that shift away and into more advanced diagnostics and to drive margin improvement. We've further empowered this just last week or very recently, we announced the acquisition of Alpha RT. This is going to empower the process of using TechLive and remote scanning even greater.
What Alpha RT adds are these bullets on the left. They are primarily an MRI, a remote MRI staffing and resource management company. They also provide comprehensive training for the in-suite assistants and certification. We've been taking advantage of that long before the acquisition. We have a new tool called AlphaEye, which provides safety using AI and cameras, particularly since we have these in-suite assistants to make sure no inappropriate materials would enter the MRI suite. Lastly, there's a software tool to make sure we can optimize the scheduling of these expert remote technologists. Internally, this is going to benefit us first by allowing us to attack those most difficult to staff hours, the evening shifts, the weekend shifts that our technologists do not want to staff. Or if they are staffing, the minute something opens up during the day, they want to move into a more traditional staff.
Ultimately, the team at Alpha RT will be able to manage all of our remote MRI resources. Externally, we think Alpha RT will pair very nicely with our Deep Health team and TechLive so that if a customer wants TechLive, but they do not know where they are going to get their MRI techs from, we can provide the technologists. If they need training to have an in-suite assistant, we can provide the training. Essentially, we can provide them a turnkey solution and not just a technology solution. We have included a phrase coined by Dr. Berger, who he is fond of, that we have all become fond of, that we now have live tech for TechLive.
MRI equipment utilization. We have invested in capital-like technology upgrades, which increase our utilization capacity and enhance patient satisfaction.
This creates a situation where we manage our durations, we standardize protocols, we create efficiency in our operations. Right now, with our capital light investments in MR, we're seeing four to six additional MRIs per day per patient.
We have created all this capacity. We need to make sure that we use it wisely. One of the other things beyond complexity that we're kind of plagued with in healthcare and certainly in our outpatient imaging centers is a lot of patients wait to the last minute to either tell us they're going to cancel their exam or they just do not show up for their exam at all. In this environment of strained or short staffing, it's really important to us that we make every slot count. If there are 20 MRI slots open on Tuesday, we want to fill 20 slots.
We want to make sure that we can try and predict what slots might be going vacant so we can backfill them with another scan. We do not cancel the initial patient. We just shift it so that the contact center or patient scheduling on our portal sees that slot as vacant. What we have done is taken, again, in a pilot East Coast region that is in co-creation with our Deep Health team, we have focused on patients who have not confirmed their appointment. They are significantly more likely to not show up than patients that do. We have taken select procedures within that cohort. By trying to manage these appointments more intelligently, we have filled almost 51,000 appointments year to date that otherwise would have gone vacant.
Annualized, that will produce $17 million of additional revenue alone just in the testing we have done on the East Coast so far. Sean and I will talk a little bit more about this in our joint session to tell you where we hope to take this with the benefit of AI.
Our growth through Denovo's or new centers has been significant since 2022. We expect that that growth is going to continue. In 2025, we have seven new centers that are opening. We estimate in 2026, 11 centers, and in 2027, 10 centers. These are going to create more access for us, bring down our backlogs, and also payer contracting, okay, requires us in some markets to build new centers to be able to handle the patient volume.
Similar to the Denovo's, we've had a very strong track record of strategic and disciplined growth through acquisitions and consolidating our core markets. You can see in 2024, where we added 37 new imaging centers so far. Year to date this year, we've added another 29 centers. This helps us increase our geographic reach, increase capacity. One of the things, one of the skills we've picked up through all these acquisitions and consolidations through the year is the ability to rapidly integrate these new centers into RadNet with a term we like to call radnetizing. Typically, we can do this within a four- to six-month period where we're going to get these new centers on our IT stack.
We're going to have our uniform workflows installed, taking advantage of all the technology we've talked about today and allow us to rapidly deliver on the synergies that we built into the valuation models for acquisition. We also are continuing to grow the joint venture relationships that we have. We currently have 26 joint venture relationships throughout the company. These are very important relationships to us that we work hard at partnering. When you joint venture with these big health systems, it's very much like a marriage. You're in this thing for a while, so you better get it set up right from the beginning. What we're able to do, we're the managing partner in all those relationships. We're going to run those centers just like a wholly owned center. It's going to take advantage of all the scale and expertise that we've talked about so far today.
When we pair it up with a health system, we're now working with all the referring physicians that are affiliated with that health system. We are very focused together on making sure that there's no leakage of those referring doctors sending into the network of centers. We do that through a lot of education, making sure the service exceeds anything else available to these referring physicians. Additionally, we help the hospitals deal with a lot of the problems that Norman mentioned with the trends. The hospitals have tremendous backlogs. They're not very good at outpatient care. We help them optimize their past utilization, get rid of those backlogs. We infuse it with everything that we normally do. This is enabling the shift of the mix to more advanced diagnostic imaging that we talked about earlier. It's improving the negotiations of the relationships we have with payers.
It is also driving much greater patient satisfaction and referring physician satisfaction because they have greater access and can get the results more quickly. Of course, we use all of our Deep Health IT tools in these joint ventures. Again, they are just like the wholly owned centers. We build on that IT stack. All the IT tools that we have talked about today run through these joint venture relationships. We have shown, again, a very healthy track record. We believe there is a healthy continuing pipeline, including, as we alluded to earlier, inquiries coming in from new health systems and new geographies that we are not in today. Over the past four years, we have shown a 15.1% CAGR with volume within the joint ventures.
I think more impressively, a 19.1% CAGR in the revenue growth from these joint venture centers, which reflects some of the discussion about the payer relationships and the shift in modality mix to advanced diagnostic imaging.
The growth of our advanced imaging and this mix shift is driving margin expansion. From 2022 to 2025, we have a 12.5% CAGR. Advanced imaging is 12.5% CAGR, is two times the routine imaging CAGR. Advanced imaging as a percentage of overall exam volume is increased by 300 basis points. PET/CT. As Dr. Rose explained, we expect that there's going to be significant growth in our PET/CT volume. Right now, from 2022 to 2025, there's a 20.3% CAGR. I believe the increase in the number of tracers, the growth in PSMA, the growth in the Alzheimer's studies, which were demonstrated previously, is going to continue.
I think those volumes will continue to grow significantly over time. CCTA, you can see that the 2022 to 2025 is 62.7% CAGR. I expect that there's going to be explosive growth in CCTA, number one, because of the overutilization in coronary angiography. Number two, the payer policy changes, which is basically referring patients to CCTA prior to any kind of interventional procedure. Number three, the fact that reimbursement has been approved for these procedures. As Dr. Kord said, there are 240 sites that are available currently at RadNet where we can put CCTA in, and we're only in about 60 + of those facilities at this point in time. Right now, we're working on standardization, scaling, and introducing this procedure into those facilities. Breast MR, we show a 2022 to 2025, 17.1% CAGR.
My expectation is that the breast MR, which is more specific and diagnostic than any other breast procedure that we have, will continue to grow because the mammography volumes are going to continue to grow. More patients will avail themselves to breast MR for final diagnosis and diagnosis prior to surgical intervention. Prostate MR, 2022 to 2025, 26.8% CAGR. My belief is that prostate MR will continue to grow. Again, there is one in eight who get breast cancer, one in eight who get prostate cancer. As we educate the community, we offer more prostate screening. Men become educated on how important it is to get their prostate evaluated. My expectation is that this volume will continue to grow in a significant way.
We have talked a lot so far, Norman and I, all morning about technology and how technology is changing what we do.
The core of our business is always going to be the people. We have an incredibly talented group of people and a lot of depth, not only the folks that are in this room today, but the 12,000 other team members we have across the company. Given the strain of staffing that is out there, it is more important than ever that we invest back into developing our team members. We make sure that we retain the best and brightest of them, and we are attracting others into the organization as we need them to grow. We have instituted a number of things over the past few years to help us with this. One thing that we have done fairly recently is institute a loan and scholarship program that is primarily directed at our current team members who are not technologists to encourage them to become technologists.
We'll pay for this for them. This obviously helps us fill some of the staffing challenges that we have and creates incredible loyalty with the team members. We've invested over $1,500,000 this year alone in this program. We're also investing in the longer-term pipeline. We're actually very active in high schools in all of our states. We know this isn't going to give us a new team member next week, but it is going to fill that pipeline to refill the school programs, which went completely vacant during COVID. We have gotten very involved in a couple of programs like Win Next Gen, but also just directly into the high schools.
Out of that, we formed something at the bottom as a test case in Maryland called the Maryland Radiological Technologists Summit, where we took the lead and invited other organizations like Johns Hopkins, University of Maryland, other big health systems in the region, and our competitors to see how we could collaborate together to attack this problem. I am very happy to say that enrollment in the programs in Maryland is up 150% this year because of that, which will benefit not only RadNet, but everybody in the state of Maryland, particularly patients that need care. We have also developed a number of internal career pathways, which I think just make good common sense. We are taking our patient service reps, or our PSRs, and we are encouraging them to become the in-suite assistants we talked about with the TechLive program. That gives them a new skill.
When those same in-suite assistants become confident that they can deal with the patient, we're seeing them move on and wanting to get further upskilled to become an MRI technologist. Again, this helps solve a problem we have, but it's truly life and career changing for that team member in terms of the income stream and the career fulfillment that they can enjoy. I want to save for last for Norman to talk about an incredible program that we've run in California and that we're starting to expand to the East Coast. Norman, tell us about JVS SoCal.
JVS SoCal is a program that I'm very proud about. It's a workforce nonprofit. As it currently stands right now, we have graduated four classes in PSR, ISA, and Dex-O-Tex. Those employees are filling positions within RadNet in Southern California.
We also educate the students on their DEXA exam, which is a state licensure. Right now, we have 90% of those students who have passed their licensure. We have been very successful in Southern California with this program. We are now expanding into Northern California and looking at expanding to Maryland. In addition, in January, we start our first MRI class. We are taking 18 students from our SoCal existing employee group. Those 18 students will take an 18-month program, an MRI, and this will give them career growth and the ability to create a career as an MRI tech. It has been very successful.
It is a great program. This investment in our people is a very important component of creating capacity in our centers.
This is one of the, I think, ultimately, it is today, and it will ultimately be one of the most important things we do to be successful. I think it really separates us from the rest of the people who ultimately compete with us and the people that operate in our space. It is really improving the patient journey through workflow innovation. Workflows, in our opinion, is everything. It is not only important to our staff, but it is also important to the patient. Our guided workflow program is going to ultimately drive patient satisfaction, staffing efficiency, utilization, and collection performance. All these guided workflows are the essence of our success. We have to be able to design them, and then we have to be able to execute on them, and then we have to be able to scale on them.
It enables patients to be self-directed, removing staff burden and further enhancing the patient experience. Contact center deflection enables greater efficiency, improved patient experience, and more scheduling opportunities. These things are the essence of creating efficiency and also improving the patient journey. Radiologist reporting for faster results and continuity of care enhances referring physician relationships and improves radiologist productivity. A lot of these programs are already in place. They're being tested. We're creating an environment we can ultimately plan them, execute on them, and scale them. Right now, we have 95% patient satisfaction based on greater than 325,000 verified patient reviews and surveys, November to November, 2024, 2025. Our time of service collections, which I think is unheard of, 95% of patient responsibility collected at the time of service. We have a 2.8% call abandonment rate. That's a reduction of 45% year-over-year.
We have a 65% recurring patient population. These are some of the things that I think we are really proud of and some of the things that in our markets have established us as a leader. We spend a lot of time promoting proactive screening and prevention and raising awareness. We go into communities not only with our staff and our people, but also our physicians. We work with nonprofits, creating innovative ways to be able to get into those markets and promote our screening programs and our patient services. Somebody mentioned Walmart earlier. We created a situation in a very underserved community in Central California where patients were not able to get mammograms. If they had to get a mammogram, they had to travel great distances. It has been a huge success. We have patients getting a mammogram for the first time in their lives.
Patients breaking down and crying or in fear because they've never seen a mammography system. We are basically going into these markets and giving them the ability to get their screenings. The other thing that we do is our mobile mammal units. We work with a lot of nonprofits. We go into underserved communities where patients don't have access to mammography. Patients don't have access to get screening. We work with the nonprofits to go into these communities and actually screen these patients. Our physicians are very involved in the markets that we're in. They do a lot of presentations in these markets, okay, to promote our screening programs and raising awareness. They've been very successful. Now Steve will talk a little bit about the legislative advocacy that we're promoting in our markets.
Thanks, Norman.
Just real quickly, again, we want to stay proactive. This complex healthcare market we're in, we want both nationally and locally and the markets we're in to be involved legislatively, number one, just so we're aware and we know when to jump in to help steer something in the right direction, but also to make sure the causes that we believe in the most are getting the attention they need. An example in Maryland was for preventative care, particularly around lung and breast screening.
You'll see a picture here in the middle of our spokesperson, WNBA legend, Sheryl Swoopes, here with Maryland Governor Wes Moore, who we had a wonderful about 90-minute personal session with to talk about care in Maryland because we were able to introduce and write, find a sponsor, and ultimately get passed unanimously healthcare legislation in Maryland, which helped eliminate a barrier for these underserved communities. While women could go and get their breast screening or anybody could go get their lung screening, there was an economic barrier if something suspicious was found. Now they need diagnostic follow-up. Often the unaffordability of that care prevented them from going to the next step. What we were able to legislate in Maryland is that there would be no out-of-pocket expense for the patient to have the diagnostic follow-up care necessary, even through a breast biopsy.
We've been able to move similar legislation through here in New York. It's on Governor Hochul's desk for approval. In a recent encounter we had, she told Sheryl personally that she intends to sign that. We're going to try and hold her to that if we can. We've been very active at a national level too, again, fighting for preventative care. Dr. Sorensen has been very involved in helping guide the regulatory guidelines for how AI will be used within healthcare. Obviously, a very important subject for all of us. Important part of our proactive care. In closing, last couple of slides, the most prominent example we have of proactive care has been our enhanced breast cancer detection program. They've heard mentioned a couple of times this morning or as we like to call it EBCD. We believe through this process, we've created a superior mammogram.
We think it's the most accurate mammogram. What happens in a nutshell is when the scan, the screener is originally read by the radiologist, they have the benefit of AI if somebody's enrolled in the program. That's being done by a number of folks out there where AI is being used on the front end. The real difference is what we call the safeguard review that happens in the next three steps that you see is if there is a discordance between what that initial radiologist interpreted and what the AI saw, that image is then kicked out to a second expert radiologist who's looking at the scan again with the benefit of AI. If that radiologist also is in discordance with the initial radiologist, they're going to consult together and come up with the most accurate outcome.
All this is happening in the background before any report is created. Out of this process, we have created a 21% increase in the cancer detection rate. To put that statistic in some context, the last significant breakthrough in technology with screening was back when 2D transitioned to 3D. When that occurred, the industry saw a 9% improvement in the cancer detection rate. Through EBCD, we're seeing a 21% improvement in cancer detection rate. In fact, Dr. Berger spoke about the difficulty of interpreting a dense breast exam. We've increased those by 23%. Because of the incredibly diverse patient mix that we see, all these statistics are validated across all races, all ethnicities. These are just incredible statistics, I think, and a terrific definition of Dr. Berger's other quote. We quote him all the time.
Good medicine is good business." I think there's probably no better example of that. Just a reminder, this is a direct-to-consumer choice. We do charge $40 out of pocket. We collect that at time of service from the patient. You can see the growth. We introduced this in 2023 on the East Coast. We got it largely throughout all of our centers in 2024, with just a few left in the beginning of this year. Now this year, through nine months, we've had over 657,000 patients opt into the program, which represents an overall aggregate 45% adoption rate. Why that is important, going back to the 2D to 3D example, how 3D ultimately gained traction for coverage from CMS and other payers was its adoption. We charged $50 out of pocket back in those days if a patient wanted a 3D exam instead of a 2D.
With 45% adoption literally growing every single month, we believe in time this will be covered by payers. Hopefully, CMS and others. We are already seeing large employers who are building this into their plan design for coverage for their employees. We are seeing some of the very large medical groups on the West Coast build this into their capitation program. Very, very proud of the program and the impact this is making in the communities we serve. With that, we thank you for your attention. I think, Jane, we are going to Q&A.
Thank you. I want to thank everybody who presented earlier today. We have a packed house here, and we are going to take questions. We have two microphones in the audience. Please raise your hands. For appropriate responses, we will hand mics over for those to stand up.
Oh, and our CFO would like you to identify yourself.
All right. Thank you for taking our questions and the great presentations so far. Maybe first question go to Dr. Judy. You showed an example—oh, sorry, identify myself. Yuan Zhi from B. Riley Securities . Judy, you showed an example of FAPI to be adopted in cancer. Where do you see FAPI will be adopted most in different cancer types? The example you showed was a cancer in pancreas, but because of maybe the high background of FDG, it's not pancreatic cancer per se, it's a PNAT. Where do you think the FAPI imaging agent will be adopted most in cancer types? Thank you.
I think most likely we're going to see it adopted for those that we know traditionally are FDG negative.
That's why the first patient trial that they're going to do is for gastric and esophageal and pancreatic. I expect that it will be first. We'll probably—they'll walk through the other tumors as we see how that works. I think that's what you'll be looking at. We know that we're fairly successful with lung cancer. We're fairly successful with certain types of breast cancer. I think that's what you're going to see.
Got it. Maybe one more to Dr. Susie. Now we have different imaging agents for Alzheimer's, and so far A-beta is the most commonly used imaging agent. What about Tau and why it's still a small percentage in the overall volume?
I think, okay, for the second part of your question for Tau imaging, I do think there is a future role for Tau, absolutely, but right now we're not getting consistent reimbursement for it. That's why we're not really doing a lot of Tau PET. Can you repeat the first part of your question about amyloid PETs?
Yeah. Why is the amyloid a dominant usage right now, and what's the percentage of that?
Yeah. Amyloid—patients generally prefer to get an amyloid PET rather than a lumbar puncture where you have to put a needle in the spine to confirm. No one can start therapy for Alzheimer's disease unless they have confirmation of that beta amyloid plaque, that toxic plaque. Patients really want to get these amyloid studies. Again, you've seen here in these demonstrations, our amyloid PET volumes have significantly increased.
Really, RadNet is, I think, the natural place for people to get their diagnosis because these memory loss patients, really, it does not really make a lot of sense to go to a hospital and maybe have to wait more time or whatever to get your imaging. RadNet is able to get the imaging done fast. Not only that, but the big thing that drives REFERS is trust. We have done something very unique at RadNet at scale. Basically, our neuroradiologists—we have over 100 neuroradiologists. All of our neuroradiologists are trained in RA surveillance safety monitoring. That is something I do not really see because I do a lot of things on the national level with Alzheimer's. I do not really see that in other programs. We offer very consistent, high-quality reads. All of our neuroradiologists are also trained in quantitative MRI, which we use for staging.
All of our amyloid PET readers have had triple training for all three tracers as well. They are getting very accurate reads. We use the AI tool for quantitative analysis. We actually put the centiloid values in all of our reports. We know exactly how much amyloid plaque is in the brain. When that patient comes back at 12 or 18 months to make decisions about maintenance dosing, we can repeat that amyloid PET, driving more volume into RadNet and then reporting out the centiloid values after they have had treatment. We use that to inform maintenance dose therapy. We are doing—and not only that, but all of our protocols are standardized nationwide. RadNet has done a lot of unique things that I do not see available otherwise. It has really brought a lot of business in, and it has earned the trust.
Thank you.
First question for Dr. Bash, just to follow up. What do you think gets adoption ramping up for these tests in amyloid specifically, right? It feels like the drugs have been out there, the therapies have been out there, but it's taken a while for it to really ramp. Is this just a matter of reimbursement from the max, or how are you thinking about that?
Yeah. A big portion was initially when we started getting—oh, yeah, sure. I'm sorry. A big portion was initially when we started getting reimbursed for it. Another big thing was an educational component. We needed to inform our neurologists that we had the capacity to do this, that they could trust our reads, that we're providing training for our neuroradiologists.
They need to be able to trust that we can give accurate safety MRI evaluation because it's critical because they need to have something they can trust because otherwise they're going to suspend the dose. The good thing about ARIA is less than 3% of patients are symptomatic even if they have the side effect. ARIA E, the one with edema, essentially always resolves. The ARIA H with the microbleeds will stabilize. It sounds very scary, but this is a lethal disease. If we can diagnose the patients faster, which is what we're really good at at RadNet, and provide trustworthy reads, which we're also good at because of our training and our imaging initiatives, that's what's driving volume in. Again, access. We're completing the exams quickly.
I would just add, Brian, I think the memory disorder clinics took a while to get comfortable with the new therapies, and that's finally starting to ramp up. It is as much education of the primary neurologist as it is to—and then they decide, "Okay, I can put somebody on Kisluna." Then we can see it. I think it's a combination of our excellence and community comfort with the new therapies.
Sorry for not identifying myself. Brian Tanquilut from Jefferies. Maybe if I could, one more just for Dr. Judy. As we think about theranostics, what do you think is a realizable opportunity in theranostics, and how do you, as RadNet—I think one of the things that we hear a lot is that there's not enough access to PET scanners for companies that are developing these theranostics products.
Just curious how you're thinking this all plays out.
That's a complicated question, but thank you. I'll stand up. Okay, I'm so sorry. I think what we'll see is as each one—just like FAPI. FAPI, and there's other products coming, but like FAPI, it already has a therapeutic. Basically, they're coming out in parallel. As this develops, that big wheel with all those theranostics companies, every one of them is going to have a different tracer. That's how it's going to build. As far as the access is concerned, the access is probably more on the therapeutic side because the theranostics has to be—it's not a radiologist doesn't do that test. I mean, don't do that treatment.
Most of the legislative bodies in the different states have very tight controls because it's radioactive material and it's therapy, and the patient leaves and they're going to be exposing people in their homes or communities. What's happening is the access to get a theranostic treatment is really restricted. In California, like in Southern California, there's maybe only five theranostic centers. That's one of the problems, and that's why Pluvicto and Lutathera, which has been along longer, don't have as many cases because they can't find people to do it from a licensing standpoint.
Thank you. Yeah, Dave MacDonald from Truist Securities. Two quick questions. The first one, again, for Dr. Rose. You mentioned earlier, just to kind of pile onto Brian's question, just given the growth that you're seeing, you talked about having a lot of capacity.
Just any additional detail that you can give there in terms of how much running room that means? The second question I have is, when you look at either a dense breast tissue patient or on the prostate side, a patient moving from either memo to MRI or PSA to MRI, can you just talk about that conversation with the payers, kind of how that works? You're obviously moving to a different treatment, but just kind of that back and forth and how that transition works.
I am going to give the second question to my colleagues, but I'll answer the first one for you first. Many of our PET/CT scanners also do CT.
When you hear—when I gave you the number of 5.3 PETs per CT per day, depending on the scanner and the uptake rooms, we could be running that at 15 PETs a day. We just have to—we take those CTs when we're—as we're ramping that up because we need to do that, we'll move those CT scans to a standard CT scanner. We have that opportunity because we have lots of CT scanners. That's not in every market, and that's not in every center, but that's our way we're going to swing this so we don't have to buy 100 new PET scanners.
Regarding your question about prostate MR, there are approved CMS codes for diagnostic prostate MR with contrast. For most of the men who are eligible, there's a full reimbursable code.
What I think sets our program apart with the EPS or the screening program is we're allowing men to self-refer from 45 to 65 for a low-cost, $275 study for a non-contrast screening exam. So we're going after, with RadNet, a dual-market approach for both screening and diagnostic MR, and there's good reimbursement for both.
Hi, Andrew Mok from Barclays. This came up a number of times throughout the morning session, but it was noted that deep learning improved MRI scan times by about 50%. What's the realized volume yield on a productivity stat like that, and what are the friction points to fully realizing that potential?
The realizable volume gain through using those tools. What we're seeing is what we presented, typically at least four to six scans a day, and that's just on an average eight-hour day.
Typically, we want to try and run extended hours if the demand is there or run weekends. We are seeing, as Dr. Bash pointed out, better images in a shorter time frame. Four to six is the conservative range we typically look at.
Great. Sticking with the services.
Oh, sorry. I just want to emphasize that is per magnet times 389 magnets. The other thing I wanted to mention is I talked about neuroimaging. We use it for neuroimaging, but we use it for every body part. Okay? It is not just for neuroimaging. It is clearly, in my opinion, a win-win for everyone involved. The patients get the faster scans. Again, that is what they rate as their most important factor for satisfaction, is getting in and out faster. Radiologists get the benefit of the higher image quality. It makes us better doctors.
Again, you can see these metastases at an earlier stage. It also makes sense from the imaging enterprise perspective because you can scan three to four more patients per day per scanner times 389.
I think in MR, one of the things that we do, and we do very well—you see all these tools. You have smart scheduling. You have artificial intelligence. The other thing that we do very well is we manage our durations. We understand every single piece of MRI equipment that we have. We understand the software level, the type of technology we have, and we understand the durations that it takes to perform every single exam. Once you put that into the ability to predict how long it's going to take to do each patient, you can also measure your volume and your throughput through those scanners.
We've been very successful at it, and we're very good at it.
I'll just add to that because we are very inquisitive. We see what other people are doing out there. One of the things that's very different about us is when we find something out in a center or a market, and it makes sense, it goes through all 405 or 407, I don't know how many centers, where we are on any given day anymore. It runs through everything. You don't typically see that. We act as one RadNet. When it's good, it's everywhere.
Great.
I like that question. We're very good. We're very good at scaling what we do. We're very efficient at it, and we can scale very quickly. Once we do find something out, we scale it, and we scale it very efficiently and very quickly.
Great.
Sticking with the services side, I think during that presentation, it was noted there was a $17 million annualized net revenue impact from smart scheduling. That was during a pilot on the East Coast. Can you give us a sense for how big that pilot was or the revenue base so we can understand the yield or uplift from that initiative? Thanks.
It's just really been this year throughout all the centers on the East Coast. Sean and I, when we do a joint session later today, will talk in a little more detail how we're going to infuse AI to make it much more sophisticated than it is today with a two-fold example that you'll hear this afternoon to make sure that we can drive even more revenue, and then we'll push it around the company, but then we can make it more accurate.
Right now, we have very few occasions where both patients show up. When they do, it's a little bit of strain on the staff, but we can normally finesse it because maybe it's one patient a day. The more we can make that sophisticated, more predictive, we'll eliminate even that so that we don't put extra burden on there. Hard to put a whole number on it. Some of these are market-oriented. I threw out a statistic of the 41% improvement using TechLive in the Northeast. Again, that's one market. There are unique conditions in each market. I would caution against extrapolating broadly, but we're seeing tremendous benefit. It would surprise you the amount of cancellations that we get in healthcare. It's a little crazy.
Okay.
Since there's no more questions, I thought I'd just put kind of a little footnote on what you saw this morning. The attempt was to give you a greater understanding of the breadth and depth of the RadNet teams that are here, not just operationally, but also from a clinical standpoint. I'm delighted on the number of questions that came to our clinical expertise because that was something that I felt was important to communicate. The seven people that we had presenting here are all of academic-level capabilities. We're proud of what they are leading us into. The underpinning of all of this is really why we've changed the moniker of RadNet. It's not just leading radiology forward. It's where healthcare is going. It's technology and innovation. Everything that was touched upon this morning and will be amplified in the afternoon really fits that description.
AI, as I talked about briefly yesterday on the earnings call, at least in healthcare, at least in imaging, at least in RadNet, is not a bubble. It is the future. It is here. It is here today. It is up to us to make it perform what we all know it is capable of, and that is better medicine, and that is good business. I do not think that the operations people also take enough credit for taking our assets and improving on them. One of the slides that you saw said how many PET scans we are doing per PET/CT scanner. When we first started on this journey of doing PET/CT, the program that we put in place was we would do PET/CTs in the morning and CTs for the rest of the day.
If there was two or three PETs, which is all we were doing in many of our centers, the rest of the day was spent doing CT scanning. That shift has created the opportunity for many of our centers to now be thinking about full-time PET/CT scanners. The challenge that we have is how do we go through those upgrades? How do we create that capacity? Because not only do we do more PET/CT scanners, but we have to have what they call quiet rooms for patients to sit for perhaps an hour, two hours, three hours before they actually get onto the scanner. Now, some of that is changing with the new tracers. All of this gets fit into the dynamics that Norman talked about, and that is logistics.
We've become, in some respects, a logistics company to try to manage as efficiently as possible all the demand that we have. I think we're in a unique place. Imaging is in a unique place. With the experience and the talent that we have, we want to be able to demonstrate to you that the performance that you've seen primarily over the last three years is something that is our responsibility to capture, not necessarily something we have to wait for the phone to ring. That is indeed the biggest challenge in all of healthcare. I don't know how many of you have been burdened by the difficulty of getting in to see a doctor to get your appointment approved when you need an imaging exam, particularly advanced imaging. These are the hurdles we face. The tools to do this are with us today.
It is up to us to be intelligent about how we implement them and ultimately turn them into a better quality business that everybody perhaps will look at RadNet as a way to embrace this and not necessarily fear it. I want to emphasize, at least in healthcare, at least in imaging, and mostly at RadNet, AI is not a bubble. It is here. It is today. It is real. For somebody that has been doing this for a lot of decades, I have never been more excited or proud of the opportunity that we have to demonstrate that RadNet is a healthcare company, okay, and not just counting scans. Thank you all. Joy.
Yep. We are good. Thank you, Dr. Berger. We are going to take a break. Yeah. There is lunch outside. We will be taking about a 45-minute break. There are demos.
Please, by all means, make sure you've heard a little bit about the technology already. You're going to get more information this afternoon. Spend time. We have box lunches to make it easier for you. Obviously, any of our speakers, please feel free to ask them questions as well.
We're at Nages. This is a five-minute call before we begin the second portion of the program. Five minutes to finish up your food and take your seats. We're at Nages. We will begin part two of your Investor Day in two minutes. Please take your seats. We will resume part two of your Investor Day in two minutes, so please take your seats.
All right, if everybody could take their seats, we're going to go ahead and get started on our second half.
It's going to be just as exciting and compelling as we had in our first half. All right, thank you, everybody. With that, we're going to start with the video, and then we will go on to our Digital Health Division.
Healthcare is run on information, but the systems that enable it rarely work as one. Most new technologies arrive with a promise: to make care better. Yet in practice, they can add friction. When technology is grounded in a deep understanding of the clinical reality, the experience changes. I don't know another radiology information system that has that level of expertise. They're working in the system every day and collaborating with your developers. No one else has that that I know of. It helps us as customers to know that because things have been vetted prior to us coming into it.
Most platforms aren't built with that kind of understanding. AI is too often designed as a standalone tool, leaving care teams disconnected and carrying the burden to bridge the gaps. Deep Health unifies the imaging experience, bringing clinical and operational intelligence together in one cloud-native system. Connecting teams, tools, and workflows with a personalized AI-powered experience. Solutions that reduce friction, increase efficiency, and raise confidence, making a new standard of care possible. I am able to look at my work list, look at the mammograms that may have a higher yield of cancer, and read them when I'm fresher in the day. That makes me much more efficient and also gives me a sense of confidence. From an AI perspective, make ourselves faster, make ourselves more efficient, and standardize our interpretations. Every deployment deepens the system's intelligence as real-world feedback becomes fuel for the next breakthrough.
A living, learning loop where innovation never pauses, offering new ways to scale and expand access. TechLive has transformed our practice by allowing us to create extended access throughout our entire footprint. When connection works, everything moves as one. Information flows where it's needed. The strain eases, and people move through the day with clarity restored. Deep Health was created by radiology professionals, what we do every day. Not a lot of them are, and you want to make sure that that's what they specialize in. At the end of the day, you want a solution that makes your life easier, and Deep Health ma kes our life easier.
Good afternoon. This is clearly on. I'll speak a little bit less loud. My name is Kees Wesdorp . It's a privilege to be here in front of you today.
Very exciting day for me personally, for RadNet overall, and for the Digital Health Division. As I reflect on that video, I just also need to tell myself it's remarkable to be in a position to work so closely with these physicians, professionals, technicians that hopefully you've met also at the demo station and to be able to co-create these solutions. I joined RadNet a little over a year ago. I've had a career in tech, different leadership positions. Before RadNet, I was with Philips leading one of the four global divisions overseeing diagnostic imaging and the informatics division. When I got the opportunity from Dr. Berger and team to join the team to think about the disruptive innovations we could bring to the market, I didn't have to think long. Sham?
Hi, Sham Sokka, CTO and COO for Deep Health.
Twenty years in the industry, all my life really in radiology, but on the technology side, building product. Started in early MR systems, ultrasound systems, and then moved to informatics later in my career. As Kees said, this is a really unique opportunity as a product builder to be in an ecosystem where you can work hand in hand with both operations folks and clinical folks to dynamically improve product. I think that's one of the really unique capabilities, and we'll talk about it more as we go on in today's session.
Sham and I are going to give an overview of the Digital Health Division and how we're delivering breakthroughs in care. What's going to be at the center of that is cloud-native solutions, AI-powered solutions, but also the importance of scalability and infrastructure, tech stack that's scalable.
Before I go into that, I want to provide an overview. This is a page that maybe some of you have seen on the course of the interactions that we had with the investor community. The overall statement is we see ourselves as leading the way in AI-powered health informatics. I very often get the question, do you have external customers? We have over 2,000 + external customers worldwide. The other question that I often get, but how does it work? Because those might be competitors to RadNet. It doesn't really matter. They're seeking for tested, scalable, AI-powered solutions. Actually, it's typically a plus that they're tested at scale in a service provider like RadNet. We have a global footprint with over 400 employees over three continents.
Very recently, we were very proud to add the See-Mode team, a team of 13 people strong in Australia. See-Mode was our acquisition in the ultrasound space. We're going to talk a lot more about it because there's quite some excitement, both in terms of the impact that we're seeing at RadNet, the deployment at RadNet, as well as externally. We're clinically validated, and we keep track, obviously, about the solutions that we provide to the market and the clearances that we get. We have the most comprehensive portfolio, including 22 FDA-cleared and 15 CE-marked solutions that we currently market. Last but not least, we have a track record of integrating capabilities. You might know, but I'll go through it. Over time, we've integrated different teams. For instance, Aiden's for lung. Remember that Professor Hare talked about lung screening in the U.K.
This is that solution that we deployed in Europe and will also deploy it in the U.S.. ConTIP for prostate and brain. Deep Health, the original company of Dr. Greg Sorensen in the breast space. E-RAD for wrists and PACS. Kyron for mammography, more focused on the 2D markets, in particular for Europe. See-Mode, I just talked about, and our acquisition of this year as well of iCAT. That is marketed under the umbrella of Deep Health. We call ourselves as part of RadNet, the Digital Health Division, and our brand name is Deep Health. It is one and the same. This page is already outdated. This morning, there was some news that I am very, very passionate about, and that is that we have acquired CIMAR. There is a whole piece of text that I can sort of talk to what CIMAR is.
There was actually already a person who explained that quite well this morning. I will again talk about Professor Hare because he explained that they had been able to scale, remember that bubble chart over the U.K.? They had been able to scale all those locations to do reading for the lung cancer screening program. That needs an image exchange platform. That image exchange platform is CIMAR. It is a cloud-native image exchange platform that powers AI solutions and interoperability solutions such that images can go from left to right to the destinations as needed, and it all works like clockwork. It is scalable. The unit economics work. We believe that that can scale up to further screening programs. That was also one of the pages that Professor Hare showed. We also believe that that is a model that can be replicated towards Europe. That is CIMAR.
If there are more questions about that, we're more than happy to talk about that. I will talk a little bit later on about See-Mode and iCAT. I can imagine that people have questions how far we're in the integration. There will be a slide a little bit later on that. Let me bring it back first to the market opportunity. We are operating in the AI-powered health informatics market, and we see very attractive growth. Now, overall, you can see on the right-hand side that currently we have a 2024 market of $5.1 billion, growing towards $7.7 billion by 2028. That's a pretty decent growth rate. What's more important is that we position ourselves in the segments that show highest growth. AI, cloud-native, outpatient solutions are the fastest-growing segment in that market.
This is really that tailwind that our solutions are feeling and that our deployments into the market are seeing. In that market, and Steve and Norm have already talked about that, there is an urgent need to navigate clinical, financial, and operational challenges. Let me call out a few and make them quite specific. On the left top side, disconnected patient engagement. The no-show or late cancellation rate is anywhere between 15%-30%. Massive issue for the operations that Steve and Norm face. That is not specific to RadNet. That is what the industry faces globally. There is a strained workforce, and it is projected to get worse. You can look at different sources, but by 2030, there will be further disparity between demand and capacity of radiologists by 15%. There are inconsistent clinical outcomes. A diagnosis can vary up to 30% from one to the other.
That clearly is a problem. Then we're facing fragmented IT systems. Currently, in any given health system, an IT leader in that health system or in an outpatient network is facing at least 20+ IT vendors to deal with. That's because there are many different AI point solutions or IT point solutions, and there are interoperability challenges. Again, there's a variety of reports about what kind of inefficiency this brings up. For instance, a McKinsey report, but there are also others. It's estimated that this has an impact of cost inefficiency of up to $25 billion in the U.S.. Obviously, the global perspective is even larger. We're set out to navigate these challenges. Now, why do we think we are uniquely positioned to do that? What I sometimes hear is, yeah, because you got access to data. Well, yeah, there's that.
It helps tremendously to have access to an enormous data set that is quite diverse across population groups to make sure you can train your algorithms and more. That is actually not all. There is a special formula that we've been able to crack. If there is one word that I want you to remember, I think Steve also mentioned it, that is co-creation. Between the clinical staff, operational staff, the engineers, the product managers across the services division and digital health division, there is deep, deep, deep co-creation. That vertical integration allows for rapid iteration. I do not want to steal the thunder of Sham, but he is going to talk about the Thyroid Suite, which is powered by the See-Mode acquisition. That went out of the blocks when we closed the deal like light speed. The deployment in RadNet was rolled out incredibly fast.
More importantly, the innovation cycle to improve the product, to really get it deployed in the right way such that we can capture the reduction in slot times was very, very effective. I'm not going to tell too much about it now because we're going to come back to it later on in the presentation. We now have clinical AI solutions in breast, chest, neuro, prostate, and thyroid. That's not where we'll stop. We'll show that roadmap a few slides later. We're clinically proven over 2,300 healthcare provider locations worldwide. We're starting to get very meaningful clinical impact. It was already mentioned in one of the presentations, for instance, the increase of cancer detection rate in breast by 21%. We're delivering at scale. 24+ million managed imaging studies per year and 1,80 0+ AI-powered cancer screening sites in the U.S. and Europe.
That used to say a much smaller number only six months ago. With the integration of iCAT, obviously, those numbers have gone up quite a bit. What we're going to talk a lot about is our Deep Health OS. Our strategy and our purpose is to bring clinical and operational intelligence together into one enterprise solution, the Deep Health OS. The Deep Health OS is really pioneering a cloud-native diagnostic operating system, connecting imaging, informatics, and AI across the workflow. We often use the word workflow. In this case, I'm talking about the overall radiology workflow. We'll talk quite a bit about how a patient, what the patient journey is from intake to center operations, to image acquisition, to image interpretation, to clinical collaboration and follow-up. In each of those domains, we have innovations to bring. These are not new products.
These are also products that are ready now. In each of those domains, we co-create and we deploy at RadNet. Sham, I'm going to hand over to you to talk a little bit more about that.
Yeah. Thanks, Kees. You saw us in the earlier movie. Let me set a little bit of the current state. If you come into any sort of radiology enterprise, even a radiology enterprise within a health system, you look at these tools. We color these specifically because in the real world, every single one of these colored areas is a separate tooling infrastructure. Patient experience, you have often portals and various tools that are, again, separate siloed kind of technologies. Center operations will be a whole nother set of tools, risk tools, oftentimes from different vendors than the patient experience tools. Image acquisition is modality specific. Again, whole nother set of tools. Image interpretation typically PACS, nothing to do with the risks or the EMR. In fact, completely independent systems in most places.
As you think about closing the loop, revenue cycle, again, different sets of tools. That is the system we're in. When Kees talked about workflow and workflow being a challenge, in order to deliver a seamless experience for a patient, behind the scenes, we have to integrate data from all these systems. Now with the promise of AI, and this is why you see some of these starts and stops with AI adoption, it is very hard to deliver AI into a system if you only see a piece. If you go into a radiologist workflow and you say, "Okay, I'm going to give you better detection," but you do not have AI tools to summarize and integrate data from the other parts, you kind of leave it flat. That is the world that we're in today.
What we're trying to do with Deep Health OS, fundamentally, is to bring the data in all these different steps together in one sort of environment. Then that data can be orchestrated to serve multiple different applications. The fundamental promise of Deep Health is this unification of data in the entire workflow so that we can operate on it, so different users can touch with different applications and really maximize their value. How do we then build solutions on top of this Deep Health OS? What we've done, because customers today are not even close to being ready to buy this promise of a universal system, is we have to package these systems in sort of different product scopes. The first area, what we call operations suite, is the next generation of RIS, radiology information systems.
These systems integrate the patient experience, do the scheduling, run the day-to-day operations. The goal here is, as we bring AI in, is more and more to automate operations. Think, for example, things like document scanning. Today, over 50% of our orders come by fax. We have teams of people, and not just us, almost all of the other radiology delivery organizations have teams of people that manually go through these documents, extract the order, and then put them into the system so we do the right test. If we do not get that right, we are not getting paid downstream. If we do the wrong test, all of this sort of flows downstream if you do not do it right. This is where now AI tools can come in and really automate some of these tasks upfront.
When we talk about automation, that's what we're talking about, automate operations. The second sort of family of tools, and we call all of these our suites, second family of tools are patient engagement. This is what now really drives the scheduling process, the registration process. If you haven't seen the demonstration, this is one of the things that we're showing. Today, more than 90% of our patients call in to schedule their appointment. Huge opportunity for improvement in terms of errors, in terms of making it easy for patients. We ask them, in some cases, 80 questions by phone, which creates a very elaborate process. What if we can actually digitize that process, the scheduling process?
When they come to the center, we have, again, the majority of patients need to talk to a person to be able to then be registered, hand over their final set of paperwork. What if we can actually automate that as well? Both of those we are showing, and we are starting early demonstrations of that in our centers and starting to scale that up. The goal there is really to drive a guided patient journey, but more and more, again, automate the complexity, the number of steps. One of the reasons that patients do not do digital scheduling is we have way too many manual steps. If we then think about what are the AI tools that are really driving this area, the left and why we kind of put it into this infinity shape is everything on the left is operations.
We really are really focused on bringing agentic AI tools. Agentic AI, again, is leveraging LLMs, large language models, visual language models, Gemini, ChatGPT, think all those kinds of tools, but to then automate tasks. When we create these automated tools, those are the agentic AI. We think about, over time, hundreds of these agents working to do things in the background, pre-authorization, document scanning, automatic order selection, making sure it's stat versus non-stat. All these are separate agents that we start to bring in as we think about the operations side. On the right side is the clinical pieces. The core of that are two, three large families of solutions. First is the clinical AI suite. These are all the clinical AI tools that we've talked about: breast, prostate, lung. These are not just about making radiologists faster.
Although we want to make radiologists faster, we want to make their jobs easier, it's a lot about this notion of state-shift disease. It's about better outcomes. You saw from Dr. Hare how the program together is shifting outcomes in lung cancer. Steve talked about the improvement in cancer detection rate, 21%. That's where value is, making radiologists faster while we enhance clinical outcomes. Now, adjacent to that, and almost all the AI tools are delivered through what we call the diagnostic suite. Diagnostic suite is the integrated experience for the radiologist. This is their one cockpit. The idea there is really about automating more and more of their tasks, the mundane tasks, things like measurements. We talked about plaque characterization, again, things that, again, radiologists find it difficult to do if you don't have quantitative tools behind that.
Also, to do that at a much greater velocity, to have the ability to diagnose, report faster. The one thing that I always sort of am constantly reminded of by our team is we talk a lot about visual AI, tools that help find cancer, these things earlier. If you take an average reporting time, so if it's an X-ray, it's a couple of minutes. If it's a CT or MR, maybe 10-15 minutes, 70%-80% of the time is spent on creating the report, not actually reading the images. It's all the documentation. This is where we bring the tools in to make the reporting easier, auto insertion of text, auto insertion of measurements. Can we create 90% of the time first-time report that the radiologist is reviewing and adding maybe some final remarks?
I mean, that's where we're going on the diagnostic suite and as we bring report automation in. Lastly, if you look on the acquisition side, this is where multiple folks have talked about TechLive. The goal of TechLive is really to increase our capacity of our systems, doing more with our imaging technology by making it easier for our techs to do more scans from remote situations, but also when we do not have enough technologists to deliver services across our entire fleet. I'll talk about that in a bit of a Zoom. All of these tools are powered by both agentic AI, so things like automated agents, as I was talking about earlier, but also clinical tools.
One example of a clinical tool that we haven't talked about is as we think about TechLive, how do we make it easier for a technologist to make sure they're doing the right scan? Today, it's remote, but we also want to bring tools that help them look at image quality. If there's artifacts, they're notified so they can actually immediately retake the exam. Tools also to make sure that we have the right positioning, the right orientation of patients so that downstream, the radiologists aren't giving us yellow stickers where we tell them the image quality wasn't good enough. We can reduce the amount of recalls and, again, get the first-time right imaging there. Those are kind of the agentic and clinical AI tools that sort of drive those clinical tools to be more specific. How do we enter the market?
Because if you think about this, this is a large space. There are multiple companies creating hundreds of millions of dollars of business in one of these verticals themselves. We approach the market with those tools in sort of three larger market spaces. One is what we call enterprise operations. This is the clinical record, if you will, the clinical record, the operations. Think in the hospital world, EMR. Our solutions for the outpatient imaging center are essentially the EMR for the outpatient imaging center. In fact, it is a certified EHR from Meaningful Use and so forth. There, we compete with folks that are really core risk folks. The differentiation over time is going to come from these agents. It is an EMR, but on top of it, can we automate more and more tasks? I talked about that a little bit earlier.
We're doing that with operations suite. In the middle is the traditional, let's say, the radiologist experience market, so the PACS market. Again, in hospital, I don't want to say in hospital, health system PACS is that have to handle all kinds of different complicated flows. In a hospital, a PACS is not only serving a radiology department, it's serving an oncology department. It's serving a neurology department. It's serving all these specialties. You need additional functionality than in an outpatient imaging center. What we've now built in diagnostic suite, we're extending some of that functionality. We want to enter that market over the course of the next 12 - 18 months. We're already very deeply rooted in the outpatient imaging market. We're bringing that market. We're modernizing that market with cloud, enabling it with AI.
That is how we are now differentiating in that enterprise imaging market, that middle market. The last vertical that you see is that emerging AI market that you see. I think this world is quite different from the other tools. You wonder why competitors from one world do not come into the other. This is class two FDA medical devices with evidence. You need to do clinical trials to get devices approved. You need to have clinical outcome data. You really need to have the results. It is a different space altogether. Now you have seen multiple companies generate revenue in this space. Some of these, if you look toward the bottom, are more specialized. They have their single areas. Think HeartFlow in cardiology. Think Viz. They are in neuro. Single sort of area spaces. As you go up, you are seeing players that now do multiple clinical applications.
That's the space that we're playing in, multiple clinical applications across multiple domains. Kees talked about 22 FDA approvals. The world is evolving where you have specialties that are highly reimbursed, and then you have folks that can do breadth. That's the space that we're really starting to gain momentum on. That's how we start to build also momentum with our customers. Now, one of the critical things, if you're going to be an integrated player, is you have to have a tech stack that can rapidly bring these technologies in and create efficiencies, really make it an enterprise solution as opposed to separate point solutions. What I want to talk you through is a little bit about our technology strategy, our organic technology strategy, about how we actually do that.
How can we play in all those markets, build solutions, and have a technology stack that can address all these markets? Because today, each of those players that we compete against are monolithic. They have special technologies in those areas. How do you actually build a technology stack that can work across all areas? As I talked about, we talked about these sort of three large markets. The fundamental of that is you need to have an integrated cloud platform. This is very difficult to do if you're not on cloud because there are so many more tools, things like security, infrastructure, scaling infrastructure is much more difficult if you're trying to do that on-prem. The first is a strong foundation on cloud. On top of that is the strong data layer.
Remember, I talked about core of the default OS is bringing all the data together. We are bringing both the clinical data and the imaging data. We need to interoperate with other systems. For example, the clinical data, we need to interoperate with EMRs. In the image world, we have to operate with other PACS because in many environments, you are operating with other PACS systems, other image sharing environments. Think of that as an interoperability layer on top of our cloud stack. We have what is called an orchestration layer. It is a complex word, but the simple thing an orchestration layer is doing is taking the data and serving it to the various applications.
Think, for example, if I'm a technologist, now rather than looking at two, three different tools to get the information to do the right scan, the orchestration layer is taking the data from the RIS, taking the data from the imaging information, and then giving me a universal view of what I need to know to do the right exam for that patient. That structure is what we mean by Deep Health OS. It's the foundation layer of all of our solutions. Every single area, whether it's enterprise operations, enterprise imaging, or any of our AI tools, sit on this stack. Now, why is that so important? If I have one common foundational stack, I have one service infrastructure.
It makes the margin of running a software business much more straightforward because I do not have a service team for enterprise operations, a service team for enterprise imaging, a service team for AI, which is why a lot of these companies are not able to succeed because the point solutions all have different service infrastructures. I have one security layer. Think about if I had, again, separate monolithics, I have to secure each one with a different security infrastructure. The third thing is regulation-wise, everything runs on this. When I get this through regulatory bodies, I just repeat the documentation every other time for every other solution. It creates the scale on the product side, mirroring what Steve talked about earlier as RADNETizing. This is essentially the way we scale as we build our assets as well as acquire our assets.
That's also why I mentioned scalable in the beginning. Sham, for the audience, you might want to point out if you click one back where CIMAR fits in.
Yeah. The recent acquisition of See-Mode essentially is this layer for the U.K. screening programs. What they have done in the U.K., because you have these legacy PACS systems in all the hospitals, is they have created a virtual data layer. Then we have run our AI applications on them. We can now start to scale AI on top of, let's say, legacy systems. That model, we think, is going to be a really critical model in state-funded health systems because it costs so much money to change the infrastructure at each hospital. What Sam talked about earlier with scaling out lung, breast, that is what now the NHS is doing on top of the scale layer.
We will come back to that when we talk specifically about those programs downstream. Now, on top of this Deep Health OS, we have created our intelligence layer. Why I talk about it as an intelligence layer is when we build our AI assets, we build it once, and it can be used in multiple ways. Think about what I talked about earlier, like a document scanning application. When I build it once, I can use it if a patient is scanning a document as they are registering on their phone, it can be used there. If they come to the site and we have not done that, I could use it there. I could use that same document scanning technology to scan clinical information that our physicians get via fax as well, by the way, and put that and give that information to them in the PACS environment.
Those tools, I build it once, and it can be used across all my solutions. Same with the clinical AI tools. I build them once. It can be used in reporting. It can be used in AI applications standalone. It can be used in other areas. Now, I do not want to pretend that we are only using our tools. We are also integrating in third-party tools. We have deep collaborations with the hyperscalers. We are building models with Google in the LLM space with other players, and we are integrating those kinds of tools in. We are also integrating third-party AI tools, other in X-ray, for example, where we do not today have a portfolio. We are integrating other people's tools as well. Those are injected into our application layer. Typically, in our world, it is viewers. Think viewers as Netflix for us, things that images go through.
Reporting is where we create reports, whether that's technologists creating tech notes or radiologists creating their reporting notes, work list and workflow applications, and then communication collaboration. How do we disseminate information between the service, but also to our referral base, to patients? If we build those once, then I can assemble these parts for those various pieces. That's how we're now creating that speed and velocity on our organic innovation as well. Now I'm going to walk you through some of those, all built on the same tech stack. The first, what you see here is the diagnostic suite. This is the radiologist cockpit. This is where radiologists are starting to—we're now starting to roll this out, and we'll talk a little bit more of the rollout at RadNet. We're starting to roll this out.
Today, in the radiologist workflow, in many cases and many institutions, each of these things is a separate vendor. Think something like viewing. Viewers are typically separate solutions almost everywhere than the reporting. For example, Nuance is in the reporting world. People like Visage do the viewer. They do not both do the same thing. Today, in a natural environment, these are separate tools. In many cases, work lists can also be separate tools. Of course, AI is always a separate tool in today's world. You come over here with AI and analytics, again, different tools. What we have done in our radiologist cockpit, which is our diagnostic suite, is we create one unified experience, one tool set. As we bring AI in, it can be in every single part of the workflow. Take something like I have a MAMO result.
We talked about that earlier. There's a finding. On the work list, I can flag medium, high, low. The radiologists know right away when they jump onto the work list, is this a case that has a finding? Is this a case that's a normal? They can plan their day. They can sort their schedule for the day. People have different patterns on how they interpret. Some folks will actually read cases with findings early in the day when they're fresher. Those kinds of things they can start doing on the work list. That same AI result is in the viewer. When the viewer pops up, it goes right to the place where cancer may be, and people can scroll up and down, view the cancer. Measurements, findings, for example, density measurements in the breast case, auto-populated in the report.
They do not have to dictate it. As we build more and more of these AI tools, the report will be auto-generated. I want to just give you one example where we are doing that already at massive scale, which is thyroid. In thyroid today, 94% of the reports are auto-generated. The radiologists do not touch them at all. The radiologists will look at the images, confirm the diagnosis, click accept. No more dictation. That is where we are going with these kinds of tools, to be able to really enable that kind of speed. Speed in our world used to be about how fast is your viewer. We are going from how fast is your viewer to how fast can I actually do a full report. I mean, that is really the name of the game.
Doing that with high level of accuracy, high level of consistency, high level of fidelity over and over again, especially when these measurements have to be done multiple times. You have to be able to compare them. The second area for a bit of a zoom is TechLive. TechLive is out, again, something that you guys can see, same as the diagnostic suite is outside. It is also going to be available after the session. TechLive is essentially our cockpit for our technologists. On the left, you see the work list. The work list, again, similar to what you see on the radiologist side, is their daily workload. They can click on the patient. They can find the patient information. It is integrated with the risk information today as well. They can run as many scanners as they want.
Typically, in MR, they're looking at two to three. In ultrasound, they're looking at more because it's more of a supervisory approach. They can really literally run these exams. Now, why is this different from Teams or some other type of video tool setup? First of all, we have compatibility to almost every vendor in MR, almost every vendor in CT. They all do some things differently. Some of them have two screens. Some of them are 2K. Some of them are 4K. Some of them use different keyboards. They're different mouses. In the process of our FDA approval, we have to validate all those different use case scenarios because this is a diagnostic device. These are devices that are being used to acquire diagnostic images.
We need to have the fidelity of that sort of image quality, the fidelity that when I touch something, that the exam acquisition is going to be done right. That is quite a bit of the pieces. On the right, you see here the video cameras have been sort of reduced to size, but you see patients can see, the centralized operator can see into the room, can see other support devices, so they can run the scan with full efficiency from the centralized location. On the right is a little bit of what we actually install. Each of the MRs, for example, have this edge device that is attached to the MR system. That is what connects the video stream, the keyboard controls, the mouse, etc., so that we can run them locally.
The business model essentially is a one-time investment on this piece of hardware and a subscription per system that sort of drives this whole online scanning approach. To be clear, the edge device is ours. It's something that we innovated. Yeah. All this is, again, requires FDA clearance. It's also something that, as we do drive improvements, we got to maintain the quality system. We got to maintain compatibility. Now, as we go forward, as you can see, we're getting the images as we're doing the acquisition, live streams of that. We can start running AI tools on them. One very classic example is in mammography. There's a standard in all the position views, etc., that you need to have. We can do that live check as we do the scan so the technicians know that they've actually acquired a high-quality exam.
That is really useful for, let's say, junior operators who may not have the full experience to assess image quality. The last sort of area, and I want to just jump into them, is our suite of clinical AI tools. Each one of them has its own story, and we will talk about them later. I am not going to go into each one of them, but you saw some of the clinical stories earlier. I want to dive into a couple of areas and why, let's say, our AI tools might be a bit different from what other folks are doing. What we have done, and I will maybe highlight that in the breast case you see there, is a lot of tools run in the background. They take images, they run the AI, and then they create the results, and they push it into some viewer and some reporting tool.
What we've started to do is really put those together into a package. In the breast suite case, we've integrated viewing, reporting as a single solution. Half of the breast market, if you will, is integrated workstation market. It's not a pure use the workstation on a PACS. You have dedicated breast workstations. Essentially, what we've done is created a complete breast solution for those practices that need all the tools into one comprehensive suite. Now, why is that so difficult? Each mammogram is about 1 GB in size. In many cases, we're looking at five priors. We can do up to 12 priors. Think about streaming 5-12 movies in a second. That's the level of speed you need to have to be able to interpret these images at the speed that the radiologists are going at.
There is a very high demand for the IT infrastructure in addition to the AI. We have started putting those solutions together so that you can drive that value as well. Steve talked about earlier EBCD. A key part of EBCD is not just the AI, but that whole second reviewer workflow. That second reviewer workflow is quite complicated if you do not control the workflow pieces underneath because the primary radiologist will create the report. That report has to go to a second radiologist to read. There has to be a tool set for the second radiologist to annotate it, put pictures in, tell them what they missed because they are not just talking on the phone. They are digitally communicating. They are documenting that, and that information is going back to the primary care, the first radiologist, and then they are accepting that or not.
That whole workflow tooling is built into the breast suite. These tools have depth. They're not just the AI pieces. They have multiple workflow elements, IT elements, AI elements all together into what we call a suite. We do that across all the areas so that you can get the maximum value of that AI tool in the workflow. I'm going to conclude. I'm going to pass on this organic. One of the things that I really wanted to get across in this section is the depth of organic innovation that we're doing and how we've created a stack so we can bring new technologies on board. What you see here is really where we're going over the next three to five years in terms of our AI vision.
The rationale for why when Kees showed you that AI market growing so fast relative to, let's say, the core radiology informatics market. The reason for that is we're on this path as an industry, and I think we feel that we're going to be driving that leadership toward the sort of autonomous or semi-automated report. If you look at where we are today with prostate, lung, mammography, with thyroid ultrasound and X-ray just starting, we're at the infancy of coverage. Meaning, if you look at typical outpatient volumes today, the AI tools that are available today cover a very low percentage of what we're actually doing. I mean, I love the fact we're doing prostate and neuro, but we're doing something like together, maybe 80,000 cases. We're doing 1.1 million MSK MRs. Completely white space right now for AI. How do you automate tools there?
How do we automate degenerative disease, arthritic findings? How do we help people figure out when's the right time to do a knee replacement, hip replacement? Huge opportunity space. How do we automate reports to get there? As well, moving into abdomen and other areas so you get broader coverage on the MR landscape. On CT, we've started in lung. Lung's about 250,000, let's say, in the RadNet context, but we do many more in abdomen and pelvis. We're looking at tools there. We're actually starting to partner with some folks to bring abdomen tools as early as the beginning of next year into the RadNet workflow. In mammography, we have tools for detection, but detection is not the only piece. People focus on screening as cancer and no cancer.
Our average mammography report has, if you look at all of our mammography reports, we have 50-100 additional findings that we put in these reports: pacemaker, lines, clips, etc. If we were going to go to automated reports, we need detectors for all those things to create an automated report. We are starting to work in those areas. Calcification is one of them. Risk assessment that was talked about so elegantly earlier. The idea there is also to move toward autonomous. Autonomous here means normals. The majority of what we do are normals. Can we actually filter those? We are very good at it right now. Can we actually filter those with the FDA and actually have an indication for an autonomous read? That is what you see there on the mammal space.
In ultrasound, it's the progression from thyroid, which we've talked about quite a bit here with 240,000 cases. Breast is 500,000 cases. We're actually right in the middle of submission for our breast device, moving on to kidney, pelvis. If you put this together, ultrasound is our largest volume as a modality. In fact, the bottom three make up 70% of our volume. Advanced imaging is super sexy, large revenue, really high growth. What we want to be able to do is free up our docs to do more of that volume while we start to drive automation in the routine imaging. X-ray, obviously, is a huge opportunity in that space: chest, musculoskeletal. Just those two, we do almost 2.2 million studies at RadNet, chest and musculoskeletal. We get coverage there. You start to get toward that 85% coverage.
Then moving on to autonomous, again, they're normals. Something like 50%-60% of our chest X-rays that we do at RadNet are normals, in fact, in the outpatient space. Can we start to, again, go toward autonomous normals on the chest X-ray space? That is the evolution of the AI. As we bring the AI tools on top of our unified IT stack, we can start to deliver that integrated experience to more and more users. That is the transformation in radiology that we're driving fundamentally with our tools. I'm going to hand back to Kees. He's going to go a little bit deeper on some of our inorganic investments.
Yeah. Thanks, Sham. I think this also answers the question around, are you going to invest further in your clinical AI portfolio? The answer is yes.
Widen it and deepen it in the propositions as per this page and potentially beyond. See-Mode and iCAT, I first want to talk a little bit about our approach to it. You have heard the term RadNetize. We cannot really say Deep Health analyze because it does not quite work. We have, let's call it, an approach of rapid integration where we obviously build out the capabilities that we acquire. We are in a growth phase, in a build phase. We are desperate of strategic capabilities, whether it is in the engineering space or in the regulatory space or otherwise. Deep technology fit is what we are seeking. We then want to figure out how do we deploy really, really, really fast. The thyroid example will come back a little bit later. Obviously, the unit economics have to work.
I will say that we're one of the very few or only AI platforms where the unit economics actually work. We can go very, very quickly to break even because we have access to that skill. Both for See-Mode and for iCAT, the integration is complete. Their teams are embedded in the organization. Now, what that means, for instance, is that the product teams are working on one roadmap. For those who are close to these kinds of integrations, that's always the golden question. Are they working on the same technology stack, on the same roadmap, and so on and so forth? The answer is yes. See-Mode now live at 240+ RadNet sites with a high volume of scans, 14,000 scans processed. Obviously, what's very important is to figure out, does it have an impact? Yes, ultrasound slot time is reduced by 30%. You'll say, so what?
That will come a little bit later. That means that we can do more ultrasound exams, for instance. There's also something else that I can't really talk about or we'll say it here. There's reimbursement. It gives additional reimbursement for the services business as well. It's proved to be a very, very smart acquisition. iCAT, cost synergies ahead of plan. Revenue synergies, 2026 is going to be the big year where we show that the installed base allows us to upsell both in the clinical AI domain as well as in the informatics domain. For both companies, the underwriting financials of the standalone business are proving to be very much spot on. Customer traction. I talked about the customer base of over 2,000 clients. Maybe to dissect that a little bit more, that's across U.S., Europe, and other territories.
You will see that our customers have the major hospital systems as well as the outpatient networks. We sometimes get that question, do you only have outpatient customers then? The answer is no. In particular, in the clinical AI domain, we have a very, very broad base now also of hospital networks. That is obviously a fantastic opportunity to also start upselling the new propositions that we have that Sham was talking about in the diagnostic suite space or TechLive as an example. We are building great momentum. This is sort of a sample, not the complete list, a sample of the logos of recent customer wins. Think of 12-18 months. Again, it sits across hospital systems and outpatient networks. I have used a quote here. You can read it for yourself. What comes back and comes back is this point around, we want an all-in-one solution.
Now, don't get me wrong. That doesn't mean that they buy the entire Deep Health OS from the get-go, but they want a partner where they do not have the headache of interoperability and fragmentation of yet another AI vendor, but where they can actually pick the modules that make sense in their IT landscape. The other point is obviously the testimony around realizing efficiency and seamless workflows. Let me round off with the last slide before we head over into the section about how digital health and the services divisions are a unique combination. The ambition towards 2028 is greater than 30% revenue growth. Remember that market is growing, let's say, in the segments that we operate 20%-22%. This means that we're definitely set out to gain market share as we should.
As a SaaS business, the entitlement towards greater than 20% Adjusted EBITDA is towards 2028. If you then think about what's under the hood, that means that our recurring revenue, annual recurring revenue, is today 60%, will have moved towards 80%. External sales today is already 55%. I mentioned that earlier. It will then have moved to 78%-80% simply because of the new customer wins. Customer mix today is more biased towards the outpatient segment. Our ambition is to make that more balanced across hospital and outpatient segments. In terms of geography mix, we are with roughly 80% of our revenue from the U.S. today, 20% from outside of the U.S., mainly Europe. That will be more balanced between the U.S. and rest of the world.
Last but not least, and this is a little bit more dependent on the pace at which we can see our clients adopt cloud-native solutions. Today, we have 10%-20% of our solutions delivered in a cloud-native way. Yes, we also have significant installed base that is still on-prem. By then, that will have migrated if we get our things off in the right way with our customers towards 60%-80% of our solutions deployed in a cloud-native way. The last point is actually strategically important also because ultimately, then we can really help our customers scale and make their IT infrastructure economically viable. There is going to be time for questions, but not just yet because I believe we are going to the next section. Steve and Sham, take the controls.
Yeah. Thanks, Kees. Hello again.
Sham and I together want to tell you a very important story of how seamlessly the services division and the digital health divisions work together. I'll show you an excellent out there in the industry. I want to start by going back to what Kees was talking about as the challenges that digital health and Deep Health are trying to address. Every challenge is an opportunity. We're going to look at these from an opportunity standpoint and how they drive value within the relationship in here within services. The first is to improve the patient engagement, as we talked about before. Terribly complex, simplify it, make it more welcoming, as I said earlier today. That is going to help us drive patient referral growth and drive revenue in the company.
We'll give you some examples of how we're doing that already, where we think that can go from here. Same with the strained workforce. Common comment throughout the day, we need to drive more productivity from our team. We don't want them to work harder. We want the old work smarter, not harder routine, but we've got to improve productivity. Do it at the same time that we're improving satisfaction, letting them drive their careers so that they stay with us. They want to be with RadNet. We'll give you examples of how we're progressing there. Lastly, with the clinical AI, how we're driving outcomes. Consistently, Sham spoke to that already. A number of those codes are reimbursable. We're driving new revenue already. We're certainly getting better, and we hope earlier detection of disease. We're making those outcomes more consistent and reducing variability.
As we said, those are all going to help us increase revenue per piece of equipment. They're going to allow us to do more of the specialty imaging we spoke about this morning with our seven expert radiologists, furthering that shift to advanced diagnostic imaging and also generating additional new revenue out of the same procedure. In addition to the base reimbursement for the procedure, the additional AI reimbursement. Shem touched on this in the prior discussion, how we'll bring down the infrastructure cost because we're going to be a one common platform where all the data is easily accessible to allow us to continue to drive new improvements throughout. All of those unlock new revenue opportunities, cost savings, drive margin. Turn it over to Sham to explain why that's so unique internal to us.
Yeah. Thanks, Steve.
I think when I joined RadNet two years ago in 2023, I think I felt like a kid in the candy store because being a lifelong product builder, it's really hard to build product when your customer or your collaborator is far away. It's even harder to build AI products when you're really far from the workflow. One of the really unique things, I think, and really the collaboration that we've set up with Steve and Norman and the broader operations team and all the radiologists is this ability to innovate in this sort of flywheel, right? I'm going to walk you through that a little bit, right? As we build something, we build a first kind of first-of-a-kind product.
A lot of it, as we've kind of built that tech stack that I was telling you before, that modular tech stack, we start assembling different pieces together, right? We say, "Okay, Steve, here's your problem on scheduling. How does this work?" We put that in what we call co-creation, clinical co-creation. Actually, our patient registration tool that some of you might have seen outside is in that co-creation phase. We put it at a couple of different centers. We test it out. We get feedback. People tell us from every part of the workflow, right? The patients give us feedback. The operators give us feedback, right? We're able to digest that. We do our product improvements. Within weeks, we can put that back in, and we can iterate on that process.
Once we say, "Okay, it's ready," we can start scaling it. These guys over here are the masters of scale, right? Once we give them something, the staff's loving it. We can scale it 240-300 sites overnight. As we in scale form, we're continuously still learning, right? Some solutions like TechLive are really in that scale and adaptive learning phase. Some solutions like our patient registration piece, they're more in that clinical co-creation and product improvement stage. The other unique piece of RadNet here, and I can't emphasize it enough, is the clinical data, right? We have enormous amounts of clinical data end to end. If I'm a clinical AI builder, right, most people are looking at image and report pairs. That's it. I have access to other clinical information now. I have access to patient demographic information.
I have access to data of prior images and so forth. I have longitudinal medical data. Once I put that device in, I can start to assess where I'm failing. That's the insights, the continuous insights that I'm getting. I can then improve the model in the areas where I may be failing, right? I'm great at chest wall tumors, really bad at finding disease near the nipple, for example, in mammo. I can feed more data in that and start to iterate. That combination of clinical data plus the live interaction is a key differentiation as we build out products and then scale out outcomes in Steve's part of the business in terms of the imaging system. What we'll do, and you're going to get tired of this slide because we're going to repeat that a few times.
What we're going to do is we're going to go through different parts of the workflow, and we're going to talk about some of the solutions that we've deployed at RadNet already. Steve will walk you through the value that we're starting to see, and I'll walk you through a bit about what we're doing next so you get a sense of both the roadmap, but where we're delivering value already. The first is sort of in that patient experience and center operations area. Steve, if you click over.
We've mentioned a couple of these before. Today, most of our appointments are scheduled by patients calling our contact center, going through the process, the complex process we've talked about to make an appointment.
Part of the opportunity here is many of those calls, up to 50% of the calls that are inbound, have nothing to do really with scheduling an appointment. It could be, how do I get directions? What's my prep for the procedure? All sorts of questions. The first thing that we started to do in the co-creation process so far is to see how we can deflect those calls away from the contact center so that the contact center agents can really be all about scheduling actual appointments. So far, we've seen a 10% improvement in the call deflection rates. We know we've got a lot more runway there. Sham will talk about how we attack some of that. Along with that comes an increase in staff productivity. We've been able to bring down what's called the average handle time of each call.
We're obviously not handling some of these disruptive calls that are not about scheduling. We're already seeing staff productivity there that makes for a better patient experience. Steve, we do 60,000 calls a day, just to give you some scope. 10% is a lot of calls that we're deflecting. Thanks for putting that context.
It's a big deal. It's a big deal.
Shalom and Adrian would be thrilled with that. We've touched on the smart scheduling earlier to try and deal with the no-shows and the cancellations. We've shown some pretty impressive progress so far. I think, again, we're just at the tip of the iceberg because we've not done it in an overly sophisticated manner. Now that we've started co-creation and the first pilot of smart scheduling, we're seeing even a further increase in the number of open slots that we can fill or backfill.
We are seeing a 16% reduction in the instance of winding up with a double booking, which makes this much more comfortable to the imaging centers, and they are much more able to adapt to it than otherwise before. With digital registration, tremendous opportunity here. I am sure, like everybody in this room, you travel a lot, you do a lot of things. Everything I do, whether it is the train up here from Baltimore, it is an airplane, it is a hotel, it is where I am going to go to dinner, it is all done on my phone. It is all super simple now. As we said earlier, healthcare is a lot more complicated than any one of those transactions. Unfortunately, we could have 80 questions or whatever it is. Right now, 51% of our patients are eligible to do everything online for registration.
We're not coming anywhere close to 51 just yet of doing that. That's a big opportunity. This 80% of patients that are eligible for self-scheduling or being self-directed, like we said earlier, today, 9.1% of our patients make their appointment through the portal. 80% of them are eligible for doing that. As we start moving through the co-creation phase, I think there's a lot of upside here, not only to drive more revenue, fill more slots, but just to make it a much more welcoming experience, just like it is for me to book my train back to Baltimore. Simple, very easy to do. That's what we want the patient to feel when they're trying to interact with our facilities.
Yeah. As we get into 2026 and beyond, the roadmap, which you start seeing some of there, the first piece is taking this from co-creation.
From some of these things that are only in a couple of centers and then scaling that out. That's the first piece. We're also working on additional tools to drive various types of productivity. In the center, for example, we're starting to build productive analytics applications. Just simple things like, same day, do I know what's happening in what center? How do I load balance? Can I predict the schedule out a week, two weeks from now? It'll help the operations folks predict their schedules and their patterns better. I talked already about automated document processing, but also things like coding, right? Medical coding can also be automated. We're bringing agentic AI to do some of those pieces.
The last bullet you see there is as we now start to bring some of this functionality and move the risks, the operations suite onto cloud, the ability that now that toolset can be available anywhere and rapidly deployable. Just to give you some example, when we go into a new location, we often have to find a data center. We have to put the system in, deploy the IT, and then Norman and Steve's teams take over. Now, if I have a cloud-based solution, it's much easier to turn on. I do not have to find a data center. It is much more scalable as we do acquisitions. The moving to cloud also is going to help our implementation speed as we do acquisitions on the services side.
On the patient experience side, this big problem of how do we get more of patients to schedule themselves, do their own guided journey. We are building agentic workflows to drive that. For example, how do we take someone that has to do 80 questions, have them do five? That's why people do not schedule. That's why only 9% schedule out of the 80%, because we ask too many questions. We do not often take advantage of the data that already we have. These agents now are collating that, auto-filling some pieces of data for the patients so they do not have that laborious experience when they schedule. Autonomous patient outreach is really also meant to drive that 80%. How do we outreach to patients after they've scheduled to make sure they're prepared, right?
Because many times patients are given lots of instructions when they come to the site, but they come in and they did not do their contrast agent prep, or they did not do their various forms that they had to fill out before they had to come in. Driving some sort of proactive outreach as we bring patients in is another area to really much more and more do that digitized guided journey that I was talking about earlier. The next area, as we start moving around, the next area is productivity and image acquisition. That is both what we are seeing with TechLive deployments and also with Thyroid Suite deployments.
Again, we have talked about TechLive quite a bit. This is more at scale beyond co-creation. Like we mentioned, we have 400+ active remote scanners going at this point.
We've already talked about the reduction we've seen in the Northeast on the room closures, all the capacity that we're creating. One of the things we didn't touch on earlier is how TechLive can be used as a mentoring tool as well, where maybe we do have the technologist, but the technologist isn't as experienced doing a certain exam. We've seen this fantastic growth in coronary CTA, but not every CT tech is comfortable or should be doing that procedure. We can have a CT tech, an expert in coronary CTA, remote in so that we can get more of those CT units that we talked about this morning to be eligible to provide coronary CTA, creating way more access. It is much more convenient for the patient.
We can do it at the same high standards because now we have the expert remoting in on the right piece of equipment to do this. We do that with prostate MR or breast MR, which are not as common necessarily. Maybe a newer MR tech does not have the experience to do that. It opens up so much more capacity, so much more access because this center that may be a little bit more remote or in a center that is here in the middle of Manhattan can use evening hours and weekend hours, again, where we may not have our most experienced tech to do these more complex exams. It is a great, great benefit.
I hope what we'll start to see through some of what Sham will talk about for the future, but everything we just talked about there is that TechLive, instead of being kind of the exception to what we do, will really become the norm for how we run MR equipment. It will not be about just, do we have a tech or not have a tech. It is just a better way to manage our MRI system. We want to move that to become more the norm and less the exception.
Yeah. What's next for us in this space is, obviously, we want to expand this. Not all of our MR systems are connected. We are going to connect more MR systems, more CT. The first line should also have CT there in the first half of next year.
We're also building different use cases for the other modalities: ultrasound, X-ray, and mammography. Take X-ray, for example. Steve and Norman already started in some cases, but are on the move to creating X-ray hubs, right? Sites that only do X-ray because we do most X-ray walk-in. If I create an X-ray hub and I have multiple systems, do I really need an operator for each one? Why not actually consolidate that with one TechLive system? Maybe that one TechLive system is operating three X-ray systems at once. There's an opportunity there to create more capacity, better utilization of our staff as we think about that. We're working on things like mammography. Today, for diagnostics, there's a mammography diagnostics.
is a very integral workflow between the radiologist and the technologist to make sure the images are of right quality and to interactively look at them. Imagine now we can, with TechLive, pop open the live mammography system view in their radiologist experience. They do not have to go to another location to go in and give that feedback. It allows us to do different types of workflow enhancements. We are working on that over the course of the next year. I have already talked quite extensively about integrating AI solutions for positioning and quality control. The last one, and I think something that we are excited about, is the acquisition of Alpha RT, which is our live tech service. Now we can go to market with technologists paired with TechLive and actually start to build a tele-acquisition business opportunity.
That actually will have huge value at RadNet because, obviously, we have our own tech shortages, but the idea is also to bring that externally to the market as well.
Before I leave this slide, I want to touch on this last point because I'm noticing Sam in the back, who's one of our expert tech leaders. It made me remember that I skipped over the no compromise on quality, and it's too big of a point to skip over. Sam can either shake up and down or side to side, but I think a remote scan is a better scan. It's a better quality scan. There's not only no compromise in quality, it's a better scan.
Yeah. Steve, in September, we did 20,000 fully remote scans. I think we're getting a good momentum in this space.
All right.
Likewise, we're seeing a lot of growth in our Thyroid Suite. We've talked about See-Mode before. We now have 240 sites automated with our See-Mode ultrasound or thyroid AI tool. We're doing 14,000 thyroid studies per month. Hopefully, by the end of the year, that number will be bigger. We've got two more markets that we're pushing this through. We're essentially at scale today, not all the way there, but essentially at scale using it. We're seeing some pretty remarkable numbers here. We touched on, I think, when Kees and Sham were talking about the 30% reduction in scan time. We have two issues going on with ultrasound. One, it's another one of those areas, very tough to find good, experienced ultrasound techs who want to work at the pace and the quality that we're accustomed to and asking of them.
More importantly, it's kind of a build it and they will come. I think Sam mentioned that earlier with the Field of Dreams. Ultrasound is kind of that way. If we open an ultrasound room, it seems to fill up. Very important that we take advantage of, use those slots very wisely. Picking up 30% slot time there, fantastic. The information that's being fed from the AI to the radiologist is also fantastic. They're accepting greater than 90% of the AI-generated reports as the report of the information that's given to them. As we mentioned earlier, we're actually benefiting from additional reimbursement for these AI tools. Again, we're getting reimbursed as normal for the thyroid ultrasound, and we're getting additional reimbursement over 50% of the time using this tool today.
Yeah. Just to hammer this point home, right? This acquisition was closed June 1.
It wasn't set up for our workflows. So we brought the product in, tried it out in a few sites. We adapted it for our workflows here, put the tool into practice, and it's now scaled in two-thirds of our sites. We're in October, right? I mean, that type of innovation and speed is really unheard of. It's one of the real amazing things about the flywheel that I talked about earlier. We're looking to expand this in 2026 to get that last one-third through the course of the early part of next year. As I said earlier, we've fully completed the FDA study for breast, and we're in the process of admission. The government shutdown kind of maybe delayed us because we couldn't actually submit our FDA filing. Anyway, we're kind of running with good pace there on the breast ultrasound.
Of course, we want to deploy that very quickly like we did with thyroid. Because it is on the same stack, I am going to—it is my month, right? It is on the same stack. It is going to be much easier to deploy because it is exactly the same solution. It is already plugged. It is just about getting the physicians comfortable with the results, and we can really start to drive scale around the breast ultrasound. We have various other high-volume applications, whether it is pelvis. We do almost 900,000 pelvic studies, renal studies. We will start to really bring the toolset into these other areas as well as we go forward.
I threw in a couple of Dr. Berger quotes earlier this morning, so I will throw in a Sham quote because what Sham calls this with thyroid, I think b reast will be the same.
It's a triple threat because you're going to get improvement for the technologists, improvement for the radiologists, and you're going to get new reimbursement. It's a pretty nice combination.
Yeah. The final two areas, so bear with us for a little bit longer. The next area is in image interpretation. This is the Diagnostic Suite, the suite of portfolios that's really aimed at the radiologists. We're just starting to roll out. That's in co-creation phase. We're going to talk about some of our early results on reporting and cloud viewing and so forth.
Yeah. Our new FAST viewer, that Kees and Sham again talked about, is really improving reading and viewing. Almost all of our images are able to be viewed in less than a second. Sham commented on how much data there is.
Even the larger data sets, the breast tomosynthesis, are being acquired and ready to be viewed in less than three seconds. I think the radiologists in the room probably, I hope, appreciate the speed that's happening there. That's at scale today, I believe, Sham. We're somewhat between scale and still co-creating and reporting in a lot of ways. I think we've got more to go here, but we're already seeing between a 5%-10% improvement, which means it's an increase in the reporting productivity for the radiologists. We talked about the technologist strain. There's a radiologist strain. We want to make sure that, again, we're retaining, attracting the best, the brightest. We're fond of saying our radiologists are really, when you get down to it, their most important asset.
We need to make sure that they feel that they're being surrounded by all the right tools and technology and personnel so that they can be as productive doing the things, reading the exams that they want to read the most. The same thing with the new AI tools, particularly with prostate and thyroid, where they're AI-assisted, and it's automatically putting in the volume and the lesion measurements into the report. We're seeing a 30%+ reduction in reporting time for the radiologists using those t ools.
Yeah. In 2026, the goal here is now to take those early sort of wins and expand it across the enterprise. We're expanding reporting to all radiologists. Just to give you that context, I was telling you earlier, reporting is where radiologists spend a lot of their time. To me, this is the hardest part, right? It's like replacing your phone.
I'm basically telling folks, "Okay, you've had this iPhone. I'm going to give you a new phone. And trust me, it has all this AI built in." It'll have its challenges as we start to do it, but we have a methodical agenda to bring that in because we do truly believe, and the early data is that it's a better reporting experience. In addition to that, we're cloud enabling. I talked about cloud enabling the risks, which is a bit easier because it's clinical data. Our entire risk is maybe 10 TB of data. We generate about 2 PB, 2-2.5 PB of imaging data. Moving all that to cloud, really getting a cloud-based experience is a really big goal for next year.
One of the values that will come out of that is the fact that today, East Coast radiologists cannot read West Coast cases. West Coast radiologists cannot read East Coast cases. Now, as we move the data to cloud, right, radiologists can read cases across any different zone. That will also help us load balance and really gain productivity. Think almost like internal teleradiology sort of opportunities that we can now create with that as well. In addition to that, cloud also allows us to, when we now, again, do acquisitions, bring radiologists on board very rapidly because I do not need a data center again to go and build out an IT infrastructure. The infrastructure has value as well. That is one of the key things that is why we are moving to cloud, our imaging part of our stack as well.
Obviously, we want to orchestrate more clinical AI. I talked about that earlier. We're doing that both with Deep Health tools, but also external third-party tools to improve reporting speed, consistency, and prioritization. Obviously, we want to be able to scale out this footprint as the team, as I said, does more acquisitions as well. It will be an enabling capability for improving acquisition speed as well. The last area, and we'll just hammer it home here, is the population health clinical collaboration sort of space. We're going to talk about EBCD a little bit, some of our results and how we're going to scale that out. We'll finish off after that.
Good. I think you've heard a few times about EBCD today. Indulge us one last time.
We'll just remind you again of the adoption growth we're seeing at the end of the year, 39% in aggregate for the company, up to 45%, keeps growing. I looked at it this morning. We're at 49% for all of the East in aggregate. I'm hoping we'll be at 50% before the year's out. The adoption just keeps getting better. The more we educate, the more patients get familiar with it coming into the center. It does take a lot of education. We have some regions as low as 29%. Some of that burden's on us to continue to educate our own team members in terms of the value so they can be good ambassadors talking to patients and referring physicians about the value. Because often, the patient says, "You know what?
This kind of sounds good, but I'd like to talk to my referring physician." And then they come back maybe the next year, and they adopt it. We have work to do there. Again, we have regions now over 60% as of this morning, but when we put the slides together a week or two ago, at 59%, awfully impressive. We're seeing the same thing. You may know we've gotten into the OB-GYN space a little bit. Within those centers in the OB-GYN office, we're seeing about a 57% adoption rate there because they're closer to the referring physician relationship. It is a little bit more natural. All of these, we should emphasize again, are generating $40 for the patient to adopt.
Part of our strategy is to take these preventative tools directly to the consumer, whether that's the patient or the employers that are often the ones paying for healthcare, and let them make the value decision. Ultimately, the payers will start to come on board. Going to the consumer has proven to be a very good strategy for us in the interim.
Yeah. Maybe this is one thing that I think we're also proud of. As we've done this work, we've also really taken a lot of the science from this work as well.
In about a week, the early results from this in 600,000 patients, the largest screening study of its kind, will be published in Nature Health, the inaugural issue of Nature Health, with the results from the first 600,000 and really the impact of what EBCD has done in terms of cancer detection rates, but also in terms of the population itself, right, covering dense breasts, covering different types of women at-risk populations as well. We are really proud of not only the business part, but the science part, the outcome part that we are driving. I think it is a paradigm change, Steve, because I think we are talking about a new way to do screening. As we reinvent mammography screening, I think we think about also adding image-based risk, breast artery calcification, and really expand this to other potential domains as well.
That last area that I'll, if you advance, Steve, I'll kind of conclude here. Dr. Herrer spoke eloquently about lung cancer screening, right? Really, this notion of stage shift, this is the story of stage shift, right? England, historically, before such a screening program, the majority of disease was found in late stage, 70%. Now, the majority of disease is being found in a treatable stage because of these sort of programs, right? It's that sort of triple layer of the program. It's the radiologist interpretation with AI, with an infrastructure that can now allow you to scale that. What we want to do from that is now bring that technology also into the United States. Lung cancer screening is really horribly done here. We want to be able to bring that type of outcomes into the United States.
We want to extend that program in the U.K. to other areas. We recently were selected to bring our prostate screening in similar pilot form in the U.K. We are now in pilot form. I do not know if you remember when Dr. Herrer was talking about it, there was a pilot study, and then it moved to broader adoption. We are in that stage now in the prostate with HLH driving the interpretation piece. With See-Mode, the acquisition that we talked about, the infrastructure piece, we are also bringing on screening services that the NHS already does today, but now enhancing it with AI. For their breast cancer program, there is an opportunity there also to bring these kinds of tools into the U.K. for breast cancer screening. What is super interesting is, based on the NHS experience, Germany has now started a lung cancer screening initiative.
Italy has started a lung cancer screening initiative. There is really large opportunity in Europe based on sort of this model to start expanding in other parts of Europe. I do want to make one point, which is I think sometimes we forget. The majority of screening workflows in Europe is a double reader workflow, right? If you talked about what Sam and his team have done, it is the only screening workflow where you do not have a double reader. You have a radiologist with AI. AI is actually acting as the double read. Now you are able to scale this because if you were doing 6 million, as Sam was talking about, reads in England alone, you would need 12 million radiologist reads, which is crazy if you want to scale out a program.
This idea of being able to take double read screening workflows and really bring single read screening workflows into Europe is a tremendous opportunity with AI, not just in the U.K., but across. I think that's another opportunity that we're really looking to grow in. It's one of the reasons why the AI market is so interesting as we start moving it forward. Where does that translate value-wise? As we think about it, as we walked you through these sort of four areas, these different solutions are in, let's say, different stages, right? With image acquisition, we're at scale. We're really already generating the value to some degree, and we have value coming over the course of the next year. Things like reading and report generation, we're very early, and so value delivery will happen more in the 2027 timeframe.
You see that rough roadmap here of when we think the value is going to be delivered. Over the next three years, we think in the imaging services business, now, this is just the imaging services business alone, we would see an impact of 100-150 basis points on the margin itself, right? That is kind of what we are tracking and looking to sort of deliver into the business. With that, I think it is a perfect time to transition to Mark, who is going to give us kind of a closing financials.
Thanks, Sham. Thanks, Steve. We have thrown a lot of information at you today. This morning, you heard from our seven clinical leaders about each of his and her individual specialties.
You heard from Norman and Steve this morning about operations and all that they're doing to drive efficiencies, to drive growth in many of the specialties and in other areas. This afternoon, you heard from Kees and Sham about all the innovation and the technology that we're developing with Deep Health and what the implications are on our operations. You are probably wondering, when you put all this together, and there are a lot of moving parts, what does this mean for RadNet over the next few years? I think the best way to address that is really to set the stage for that with, how have you been performing? Is this sustainable? Can this accelerate? If you look at the past four years, you've seen that we've been very successful in driving double-digit revenue growth.
We've compound annual growth rate over the last three years at 12.3%. Our EBITDA CAGR was 17% over this three-year time period, meaning 2022 was the baseline. By definition, we've had some margin improvement. We've improved the margin about 200 basis points over the last four years. That is through all of the specialty work, all of the innovation on technology, and all of the operational efficiencies that we've gained. What I think we're really proud about as a company is that we've done this and had to overshadow a number of headwinds over the last couple of years, particularly around Medicare reimbursement. We've absorbed over $35 million worth of annual Medicare cuts over the last four years. That has been difficult. Every year, we've started off in a new hole.
We're happy to tell you that next year, we actually have a benefit from Medicare in the area of about $4 million-$5 million. This is the first time we're not working from a new hole when we're going into the following year. I think even more profound than that has been the amount of labor increases that the business has absorbed. It's a credit to Steve and Norm and really everybody in the organization where the shortage of labor, the shortage of technologists, which we've talked about a lot today, has caused us to, just in a market where there's supply-demand imbalances, to pay a lot more for our staff, and particularly in the area of technologists. We've absorbed over $100 million of same-center labor increases over this four-year time period.
I think we're very proud that we've been able to overcome some of these challenges. We're hoping with some of the things that you've heard with regards to the technology that we're bringing forward in the next few years. We think that we can slow the curve, maybe even reverse the curve in terms of some of these challenges, particularly around labor. The question is, with this performance, do you think it's sustainable? Do you think it's durable? The answer is yes. Not only do we think it's sustainable and durable, we actually think it can accelerate with all the things that you've seen here for a variety of reasons. One is we've got positive industry trends, okay? Imaging is in the sweet spot of the healthcare delivery system.
With a focus on preventative medicine, non-invasive medicine, it's shown that if you make the investments and you diagnose disease earlier in the disease process, you have far better patient outcomes and outcomes at much lower cost. That's not changing. It's only growing. There will continue to be innovations in technology in our industry. As you've probably heard a lot today, we're a technology-driven business. There's going to be continued improvements in equipment, in post-processing software, in AI, in radioactive pharmaceuticals, contrast materials. All of this will lead to more and more clinical indications for ordering the types of diagnostic tests that we perform. The shift that we've been benefiting from during this four-year period of procedures from the much more expensive hospitals into the lower-cost ambulatory sites of care, that's going to continue. You're seeing it, as Dr. Berger mentioned, not just in radiology.
You're seeing this in all specialties. We think that when you look at the industry, as Steve showed in his slide, of being a $100 billion+ industry, today, about half of all the diagnostic imaging exams are still being performed in the hospital. When you talk to these health systems and hospitals and you look at the mix of that business, they'll tell you that at least half of the business that they're continuing to do today is still performed on ambulatory outpatients. There is $25 billion or more of imaging revenue that can and should shift out of the more expensive hospitals into the ambulatory sites of care in the future. Obviously, there is a lot of pressure on the insurance companies and the health plans to continue to lower cost. Managing site of care is top of mind for those. We're going to continue to build sites.
I think it was Steve and Norm that showed our De Novo strategy over the next several years. We've got 11 De Novo sites in various stages of development and construction for next year. There'll be more to follow. We're aggressively trying to build capacity and expand geography. Tucking acquisitions, we'll continue to be doing Tucking acquisitions. As big as we are, we're still a very small part of the delivery system in outpatient diagnostic imaging. It's believed that there's over 6,000 imaging centers nationwide in the U.S. I think Steve mentioned this. If you put the major chains together, the top five chains together, maybe we're 15% of all that volume. Scale is becoming more and more important. Consolidation will continue. We hope to play a role in that consolidation. Reimbursement. Good news on the Medicare front for next year.
Obviously, all bets are off after that. Medicare is the one thing we've learned about CMS and Medicare is it's highly unpredictable. Medicare is 23%-24% of our business mix. The rest of our business mix, we have a lot more control and a seat at the table. Almost 60% of our business mix is with commercial insurance companies. We've built the scale. I think we've demonstrated to them that we're an important part, a party, and a partner to them in terms of helping to drive this business out in the more expensive hospitals. I think that they've started to recognize that, which is indicative of the fact that we've been successful over the last few years in getting price increases. We think that that's going to continue going forward. Joint ventures will continue to be a big part of what we do.
We have 26 joint ventures today, as you saw in an earlier slide, totaling about 152 locations of our 407 locations. Today, we're up to about 37% of all of our centers held within these health system partnerships. We think that that can, in the coming couple few years, will be over 50%. Hospitals who are feeling the pressure, who are continuing to lose outpatient business, are more and more looking for an outpatient strategy. Many of them recognize that they don't have expertise in managing effectively ambulatory outpatient businesses at scale and at lower reimbursement. They're looking for partners. We hope to be in a position by the end of the year, maybe early next year, to be able to announce some additional exciting joint venture relationships. You've heard a lot about digital health today, I think, in two respects.
One, in terms of what that business can do to grow and prosper with third-party external customers, but also the major impact that we think it's going to play in terms of creating efficiencies, cost savings, and improving both the patient journey and the clinical outcomes. Finally, we think we're going to be successful because we're well-positioned from a financial standpoint, from a capital structure standpoint. We're sitting on over $800 million of cash. We've got very, very low leverage and a balance sheet that can really accelerate the growth in this business. Here's what we're going to talk about in terms of our outlook. With respect to procedure volume, I think we're—and before I sort of get into the numbers, I think the point I really would want to make here is this is not an aspirational slide.
This is what we think is very, very realistic and very achievable. I think I would challenge any of my partners in the management team. I think that each one of us believes that we can do better than this. I think this is not—this is a bar that we're setting, and we're hoping to achieve it and exceed it. We're very comfortable in assumption around advanced imaging, MRI, and CT growing in the mid-single digits. We're giving Dr. Rose some extra pressure to help us grow our PET/CT volume and growth rate in the double digits. I think we can feel very comfortable in growing it 10-15% over the next several years each year. Routine imaging, we expect to grow with population and with our expansion, kind of the 1%-3%. I think MAMO could grow faster.
X-ray and ultrasound, I think we're in that range. Things like See-Mode could accelerate that, where creating that capacity with the demand that's out there, I think that this is more than achievable. Where that calculates, we think that we can continue to grow revenue in the double digits, 11%-13%. Note that this assumes no substantial acquisitions. It assumes small tucking acquisitions, which we do in the ordinary course of our business. If we were to put—and I think we're all hoping and aspiring to put a fair bit of this $800 million of capital to work over the next several years, we think that that can accelerate this revenue growth tremendously. The shining star in our portfolio, the digital health division, we're expecting 30%+ growth out of Kees and Sham.
I think that that's not—oh, he's putting the pressure on Sham. Sorry. From a margin perspective, we're expecting continued margin enhancement. I think that's a function of all the initiatives that we've made you aware of and did a deep dive today, particularly in the areas of how our growth or our investments in technology and innovation and digital health will impact the core business. Free cash flow, we expect to continue to be strong and continue to grow in the area of how we'll grow our EBITDA. From a balance sheet perspective and from a capital structure perspective, we're comfortable with, obviously, with the balance sheet today. We'd be potentially willing for the right transaction or transactions to lever up a little bit more from here. We wouldn't want to see, given today's interest rates, we wouldn't want to see leverage go above three times.
Clearly, we're at one times today. We're really trying to manage leverage and manage our free cash flow. From a CapEx standpoint, at the size of the business today, we typically spend about 3%-4% of our net revenue on maintenance CapEx. That's the money that we believe we need to spend each year in order to keep the revenue and the EBITDA in the door over the long run. We will spend significantly more than that as we see continued opportunities for expansion and growth in all of the areas that you heard today. From a pricing standpoint, I think across the board, given our revenue mix, we think that we can impact pricing on the positive side, 1%-3% over the next three years. We're very excited. I think I've been here for over two decades.
I can't imagine—I know that there hasn't been a more exciting point in our history. We really very much appreciate all of you coming here and staying and absorbing this information. I think what we'd like to do is open the floor to any questions about anything you've heard today, whether it's clinical this morning or Deep Health, financial, or otherwise.
Hi, Andrew Mok from Barclays. On the digital health side, you reconfirmed 30% revenue growth with greater than 20% EBITDA margins. I think your digital health.
margins today are about 17%-18% this year, which includes $20 million or so of investments and some dilutive acquisitions. Can you walk us through the assumptions underpinning that margin target, and what revenue base or scale do you think you need to reach a mature margin? Thanks.
Yeah, thank you. I think we're today at roughly 17% indeed. We are going to see in the near term a slight dilution of margin because of the investments that we're making, both in go-to-market resources as well as in the product platform. What we expect as we start to grow the top line is that we cover the fixed cost accordingly towards 20%+ EBITDA. That's more for in the range of towards 2028. It's more in the outer years that we're going to see that 20%, within the near term a slight dilution of the investments that we're making. On the M&A side, we will expect to see quite quickly the turnaround of profitability for iCAT. iCAT, as you might remember, was loss-making by, let's say, $4 million-$5 million when we acquired it.
Somewhere mid next year, we'll have that turnaround to neutrality, and from there onwards, it will be starting to be EBITDA positive.
Great. Just a follow-up, Mark, you mentioned that these targets are realistic rather than aspirational. Where do you see the opportunities to outperform these targets? Does it accrue to one side of the business versus the other? Just curious to hear about the potential for upside here.
Yeah, I think there's opportunities to exceed this on both sides of the business. I mean, even in digital health, I think we're being conservative with margin. We're giving Kees and Sham some flexibility over the next 36 months to continue to make investments. Because, as you're probably aware, and I think maybe you intimate in your question, that software margin should be higher than 20% once mature. And we agree with that. That might be in 2030. The focus on Deep Health over the next 12 months is going to continue to get this penetrated and implemented solutions with inside of RadNet so that we can start saving on the imaging center side. On the imaging center side, there's a lot of opportunities for tuck-in acquisitions. I think that could accelerate growth. There are larger acquisitions available to put capital to work.
I think as these digital health solutions get implemented, we'll have more data to share about how that can impact margins. One of the things that we talked a lot about today was the shift in our business mix from a modality standpoint. You saw a several hundred basis point shift over the last four years to today, where over 27% of our business mix is MRI, CT, and PET/CT. If that continues in the future, and we believe that it likely could, I think that's another area of significant potential margin enhancement and growth.
Mark, David MacDonald from Truist. Two quick questions. First, just on capital deployment. When you look at the efficiencies that the combined business is driving, especially into the imaging centers, I'm curious at a high level, does it change how you think about capital deployment, M&A around the imaging center business and just the incremental opportunity because who you'd be acquiring isn't running as efficiently? The second question on capital deployment is just on Denovo's, when I look at your current footprint, you guys aren't in what I would consider very developmental-friendly areas of the country. As you end up expanding the footprint, could we see the Denovo development activity pace accelerate as you move into states that are candidly more business-friendly in terms of development?
Yeah, on the Denovo side of the business, I mean, there's opportunities in all the markets that we have, and I'll let Steve talk about that. I mean, we're not in any CON states yet. There aren't any conditions to us making investments, buying machines, growing machines. There's a lot of room for us to continue to grow the company in all the markets we have. We have a toehold in Florida. There's a lot of expansion opportunities in Florida. There's a lot of expansion opportunities in Texas. We're just starting in Texas as of last year in Houston. There's growth everywhere.
Yeah, we try and get some balance between all those opportunities. We look at the core markets where you have the economies of scale, and we've got the consolidation opportunities, the conversations with payers. There is a lot of benefits to continue in those markets. We try and weigh out those, whether it's a Denovo, whether it's an acquisition, whether it's a joint venture. I will say we're getting more incoming inquiries from standalone or from health systems in states where we're not today. Those opportunities are certainly there. We just need to choose wisely. I've been here almost, well, I have been here two decades, I guess, because I founded my predecessor company. Time flies. I've never seen not only a more exciting time to be here, but more opportunity.
I think our biggest challenge as the executive team is to choose wisely and to weigh all the strategic and economic benefits between whether it's Denovo, whether it's acquisition, whether it's joint venture, whether it's current core market, whether it's a new market. If we do go into a new market, I'll go with one more quote before the day's out. We will be not a buy-and-hold, we'll be a buy-and-build. If we evaluate a new market, we're not only looking at that entry point, we're looking, okay, if we can come in, what does it look like down the road? We came in here 12 years ago, something like that. We bought Lennox Hill Radiology, which at the time was six imaging centers. We have over 100 imaging centers.
Not every market has the population of New York City, but that's the kind of roadmap we like to look at before we go into a new market. If we go in with four, five, or six, where do we see the path? Is it a joint venture partner? Is it another acquisition? Can we consolidate? Is it underserved? We could do more Denovo. There is a lot that we look at to evaluate that.
I think the first question we did not address, which is, given the efficiencies that we can provide with Deep Health and the operational capabilities, does that change our view on value in terms of acquisitions? The answer is, I think we have all talked as a team that bigger is better because of that very thing, where if we can make this impact on our existing 400 centers, wow, what would it look like if we could do it on 800 centers? There is no question that we would like to grow in the imaging center segment significantly, but it does not mean that we are willing to overpay just because we have a lot of capital. We have still been disciplined.
We want to work with motivated sellers, meaning people who want to be part of this type of clinical operation and have the capabilities and who want to be part of the RadNet network. That is self-selecting. We want to work with quality groups. We want to provide quality medicine. We want to go in, if we are going to go into a new market, as Steve said, at scale.
Yeah, we won't lose our discipline, but unlike some of our competitors, we might be able to stretch a little bit because it's really that effective multiple. If we know we can layer these synergies in, whether it's the operating synergies or maybe even just laying in our rate schedule, quite frankly, we may look at that a little differently than somebody else would look at it because we know what the synergies are going to be. We can RadNetize very confidently and very quickly.
Just one last one, if I could. On the slide, when you talked about scaling AI clinical suite and outpatient volume capture, how should we think about that in terms of how much of that will be internal development or what you can do within the four walls as opposed to additional acquisitions? I would assume that there's some combination, but just any detail there would be helpful.
It's definitely a combo across three. It will be organic. It will also be acquisitive, but it will also be through partnerships. And maybe, Sham, you want to elaborate on that.
Maybe just to add, right? Areas where we have capabilities, like we have capabilities in MR because we've acquired in prostate neuro space, most likely that's going to be organic, right? In areas where we don't, there may be inorganic opportunities. That's the way to think about. In ultrasound, for example, we've acquired See-Mode, so we have a capability there as an example.
Thank you.
Yeah. Brian Tanquilut Jeffries. Hey, by the way, thank you for hosting this. This has been really great. Mark, maybe as I think about the CAGRs that you gave here, obviously three-year outlook, 2025 to 2028, a lot of moving pieces as it relates to revenue CAGR and the margin outlook. Just curious how we should be thinking about the cadence of these gains as we think through the contribution of See-Mode or Deep Health or some of these things that you guys are rolling out. Without giving guidance for 2026, how should we think about the ramp to get to these goals?
If you look at our historical performance and there's.
I think it's off. There you go.
There you go.
Now.
Yeah.
That one's out.
There you go.
Technology.
If you look at our historical performance, and there is a chart in our investor deck that you can pull down from our website that has 15 years of historical performance, it has been pretty consistent. Particularly in the area of our imaging center business, because it's a much more mature business. While we can't predict how many acquisitions we're going to do next year, we don't sit around as a management team and allocate capital for the following year. Our business is a steady grower. I think that there is going to be more ratable growth coming through. Obviously, there will be some acceleration in the digital health from third-party external revenue in the out years. We'll be giving guidance for 2026.
We have a budget, and we'll be giving guidance at the end of February of next year, which will at least give you one year of assurance around that. I think you'll see some pretty steady growth.
Mark.
I think the question was also related to the RadNet programs.
Yeah.
We put a slide in there that showed the four domains that we talked about of RadNet impact, of the digital health impact. Some are at high maturity level right now. Some will come a little bit later. TechLive, high maturity level, context-centered operations will come a little bit later. In theory, if you would model that, that gives us quite a nice ramp-up over time of the benefits of the digital health program. It is not going to be that lumpy or sudden or 26 is bigger than 27 or otherwise. It is actually quite a nice ramp.
Kees, since I have you, my follow-up question would be, there was a slide where you put clinical AI growing from a $5 billion market to a $7 billion market. Just curious, given the reimbursement outlook here in the U.S., at least for clinical AI reimbursement, where it barely exists, how are you thinking about the progression of that? Or what needs to happen for that to be realized? Thank you.
It's the combined market, just to be clear. The $5.1 billion to the $7 billion+ is radiology informatics, so the traditional RIS/PACS business that will go cloud native, and then the AI business growing. There is now evidence that certain AI domains will actually get reimbursed. There is also evidence that reimbursement is not de facto the reason that you get value for money for AI solutions. Even in the absence of AI reimbursement, you would still have AI solutions brought to market that your customers would pay for. It's the combination of the two.
Just to clarify that, the AI business is going from about a little over $1 billion - $3 billion. Right? It's the integrated market that's going $5 billion-$7 billion.
Yuan Zhi from B. Riley Securities . Maybe a couple of follow-ups. First, can you quantify the AI's contribution to your revenue and Adjusted EBITDA over the next three years? Just more specific. We saw the 11%-13% growth on revenue. Can you put more details in there?
I think you could do the math, right? Because if you took the run rate of the digital health business going into next year, it is about $100 million of revenue, maybe slightly more. If you compound that over three years by 30%, you're going to have a $30 million benefit next year and so on and so forth. If you look at the EBITDA, by achieving 20% EBITDA margins, our EBITDA will almost triple over the next three years, despite the fact that we're not at a mature EBITDA margin in that business. You'll see it go from about $15 million- $17 million this year to about $43 million just by doing the math. I'm not giving three-year guidance, but you all can do that math as well.
It's not going to contribute because we're talking about a $100 million revenue business versus a $2 billion revenue business. It's not going to contribute on the top-line growth tremendously because it's so small. When you look at the EBITDA, it's going to be a real contributor on the EBITDA side, not to mention the cost efficiencies and the EBITDA it's going to create in the imaging center business for Norman and Steve.
Got it. Maybe switch gear a little bit. CMS is in the process of removing prior authorization for most or many processes, including the diagnostic imaging. I wonder how will that impact your top line?
We do not really give it a try.
We are seeing a number of our payers, at least, backing away from the pre-authorization process, which is going to be a huge cost saver. We spend an embarrassingly enormous amount of money on authorizations. Obviously, it helps simplify the whole process that we have talked about today to get people in more quickly. Any relaxation of prior authorization requirements is good. It is good for healthcare. Maybe it was impactful and beneficial in the early years to drive out some waste, but it now is just a barrier to care right now and an unnecessary cost.
Got it. Thank you.
Hey, this is Matthew Gillmor from KeyBanc. I wanted to ask about the JV pipeline. You all, at least to my mind, sounded relatively bullish with respect to joint ventures. I wanted to see if you guys could characterize sort of what you're seeing from a pipeline perspective. I was also curious if that is contributing to some of the positive rate dynamics you mentioned with commercial payers and just how you leverage those health system relationships.
All right, I'll take a crack at that. That was a question that actually came up in a conversation. Most of all of our joint ventures are driven by what is an inexorable movement of outpatient imaging away from hospitals. Some people think that that's just because of the better pricing. What it really is, is the hospitals are just not capable of keeping up with the demand that they currently have and growing for imaging. Imaging is not just growing with what we've demonstrated in an outpatient sector. It's growing throughout the entire industry. That movement is going to continue. The hospitals that have already been joint ventured with us, as well as new ones that we're getting inbound calls from regularly, recognize the need to be able to have a market approach to radiology.
Part of that market approach and what I believe we're getting some traction on is creating an ecosystem that no longer differentiates between where the procedure is done. Clearly, what you can see that we're trying to do with our 400+ centers and more is put them on the same platform. As I think either Steve said or somebody said, we will, in the very near term here, be able to read cases from the East Coast on the West and the West Coast in the East. We're doing that to some level right now, but we'll be able to do it much more efficiently if you look at what the opportunity is to bring the best reader to the procedure, regardless of where it's done. It doesn't matter whether it's done in California or New York.
The quality of what we want to be able to produce that will allow us to continue the momentum of generating more volume is about putting the best person in place. That is the same thing as in hospitals. What we are trying to educate because it is a big lift. Part of that big lift is that you have CTOs at the hospital levels that are unwilling to give up their domain, if you will. You have radiologists that do not appreciate the level of technology that we are capable of delivering and are concerned about their own self-preservation. These are things that are going to change. From a fundamental standpoint, it should not matter where the procedure is done, but where the best reader is. The best reader is not always the one at the highest quality, but who can get that turnaround time.
We have to think in a more broad sense here that as RadNet continues to grow its footprint, the opportunity to create real change is not in the conventional way that we're used to doing business. All right? As we get more and more into the very fabric of these hospitals, and I can tell you that we're having some of these discussions that go way beyond radiology. We're talking about with big systems, what is their approach to population health for screening for lung cancer, for not just breast cancer, for prostate cancer? They're all looking for ways to drive better quality and outreach into their communities. That's not done in hospitals. That's done in the outpatient sector.
If we can create this flow of information where the best reader at the best time is taking care of the clinical side of it, that's the best outcome for everybody. What's additionally coming, and I challenge all of you to kind of think more broadly, is that healthcare is going to change also. We haven't spent any time here in this meeting. We will subsequently. When we talk about doing outpatient imaging, we all think because we're in the outpatient business, it's done in imaging centers or at hospitals. There's more imaging done in doctors' offices, in urgent care centers, in OB-GYN offices. Those are the places where fundamentally we could create change from the way that imaging is delivered right now.
We're not concerned necessarily about all of those cases, particularly routine imaging, which are not as profitable and which tend to clog up our imaging centers. We can bring these, particularly with hospital systems that have medical groups that are now part of their system and can be part of the delivery system. It does not have to always be in their hospital. The fundamental changes that we're talking about, particularly with artificial intelligence, with autonomous reading, can make such a dramatic change in the way healthcare perceives imaging to be the gateway to healthcare, is that I would imagine in our next investor day, whenever that might be, we will be talking more and more about these kind of opportunities that reimagine imaging and reimagine the delivery of healthcare because there is nothing that is more important in managing a large population than imaging.
Maybe that sounds very self-serving, and obviously I'm biased, but we see it. We know it. There is a dramatic change that has occurred, not just what RadNet has adopted, but which has occurred at all levels of the health system, and that is going to change some of the fundamental unit economics. Things that we, I'm going to give you a simple example, and the reason why I'm excited about not just TechLive, but LiveTech. The shortage today, it's easier for a hospital, it's easier for an imaging center company to write a big check for a technologist that they know is sitting there generating revenue. TechLive, while it's a great tool, is looked at as another cost. People don't look at healthcare necessarily with physicians and technologists simply as a cost. It's an existential need.
I think what we're likely to see, both for technologists and radiologists, is a bundling of AI tools along with staffing needs that will make it a lot easier for this to not only be a variable expense, but one that they can titrate to the facility's needs. All of these are what is being driven by a technology-enabled company, which RadNet is, and which others are working on today, but not with the same approach. It will be a pleasure to stand in front of you in some future day to say that we're making strides on this with tools that we don't have right now, but which we envision. Part of my job, as long as I maintain it, they don't kick me out, is to try to have that vision come to reality.
Every day we're learning something new about how we can reimagine this vision to be delivered. You'll be hearing, to the extent that some of you may come to the RS&A, which is in three weeks, encourage you to come there. It's going to be quite a show for us and our booth and other announcements that we're going to make. Working hand in hand with the OEMs is now much more of a reality than it ever has been for a company that they're now recognizing, understands better the delivery of imaging, not as somebody who's building a piece of equipment, but it's about a service. It's about workflow. It's about the changes that can come about and make the entire system that much more capable of delivering better healthcare, better outcomes, and lower costs. That's the challenge that we've taken up.
We'll do that for ourselves and do it very successfully. I think that the projections that we give, if you do your modeling, none of us would have sat here two or three years ago and believed that we could achieve. The purpose of this was to show that what we're doing is sustainable. It's not aspirational. Can we achieve better than this? What we have every year for the last, or let's say for the last two years at least, we've been meeting and exceeding our guidance. Somebody might say, "Well, you're kind of sandbagging because you know going," we're not sandbagging. It's just that this growth, this transition is so rapid that it's very hard to predict from one quarter to the next how much of that adoption is taking place. It has been consistent.
The team that we've built, and particularly adding digital health and this extraordinarily seasoned and knowledgeable team, makes the future unlimited. Will we achieve better than what we're expecting? I hope so, but not just for RadNet, because I think it's good for healthcare and I think it's good for this industry. I'll end my comments because I think it's been proven out here that this was our coming out party. We've changed our moniker from leading radiology forward to advancing imaging through innovation and technology. I think everything that we tried to demonstrate to you today really fits into that characterization of how we'd like you to look at RadNet as a healthcare company, but one that is going to expand the horizons of what we traditionally have felt about the delivery of healthcare. This is not just something that RadNet is on a mission for.
This is something that has to be done with a growing and aging population where the cost of healthcare has just gone up every year, and it's not sustainable. There needs to be this innovation. There needs to be this technology. It's there. We should embrace it, and we should welcome it. Any more questions? I didn't want to.
We have time for one more question.
One more question from Mark.
You reminded me of something when you mentioned the RSNA. I want to invite everyone to come through our booth. Andrew Mok at Barclays and Brian Tanquilut at Jefferies, our research analysts, are taking tours through the RadNet Deep Health booth on Monday, Tuesday, December 1st and 2nd in Chicago.
If any of you or anybody listening on the webcast have further interest in doing a deep dive into our products and services on the digital health side, we'd love to have you.
Yeah, Martin Ji from ClearBridge Investments. My question is on the digital health external sales. When you are making the sales right now, typically what are the pressure points you're seeing from your customers that they need to overcome before you can make a deal? Also, what kind of accomplishments do you run into in that process?
Sure. I'll let Kees take this one.
Yeah. It depends a little bit. Remember that Sham presented three business domains, so enterprise operations, enterprise imaging, population health/clinical AI. It really depends a little bit in which of the three domains we go out.
In the clinical AI domain, you can have a technical discussion around who has the superior model. We try to avoid that because our model is fantastic, but our integrated solution approach with, for instance, viewing and reporting at the same time is differentiating in the market. We try to emphasize the value add we bring with the integrated offering. For enterprise imaging, we currently are going towards markets with, for instance, a reporting tool that's a little bit early to report out on. We are engaged on it. We have a batch portfolio that we sell in the outpatient networks that we competitively position and similar on enterprise operations. It depends a little bit on what segment you are, and it depends also a little bit on what module of the Deep Health OS specifically is the request of the customer.
Kees, if I can add as well. Given those three segments, we see growth differently. If you go look at the RIS market, a lot of our growth is selling additional solutions to our existing installed base because acquiring somebody to change their RIS is a massive undertaking. Right? We are getting those new logos. We are announcing those, and we are getting those, but those are much slower where we get document scanning as a service, $500,000 a year, that kind of thing. That happens within an existing customer base. If you go to AI, that is where acquiring new logos, because these solutions, health systems, are buying them. Outpatient imaging folks are buying them. That is where you see new logos.
In the middle of the enterprise imaging space, that is where we are kind of gaining momentum, both seeing new logos and then converting our existing installed base from on-prem to cloud. Right? That is literally how the three businesses sort of break down.
Thank you, everybody, for being here and joining us and joining us online as well. For those who are still in the room, the demos are still outside. If you did not get a chance to see them during our break, please take time before you walk out. We are very grateful for your attention today.