Hello. Hello, everybody. Good morning. Please come on in, find yourself a seat. Just wanted to welcome you all to the 2025 Bloomberg Healthcare Investor Conference. Today you're in Room A. We have five presentations in this morning's session. The first presentation this morning is from Perimeter Medical Imaging. Joining us today is Adrian Mendes, the CEO. It'll be about a 20-25 minute presentation. There'll be time for questions at the end. Adrian, please take it away.
All right. Thank you. Good morning, everyone, and thanks for coming to our presentation. Perimeter Medical, I'm Adrian, and I'm the CEO. Lots of legal. All right. We didn't get it all. Okay. Our mission at the company is to make it so that no cancer surgeon ever has to say those words again to a patient. We didn't get it all. Right? The goal is to have a successful first surgery, first cancer surgery, and not have the patient come back again. We've heard from so many customers that that's the second worst phone call they have to have with their patients. The way you get to that point is through looking at something called margins. Okay.
Just to give you the basic background, what the surgeons try and do when they're taking a tumor out for any type of cancer surgery is to make sure that the tumor comes out fully encapsulated with a rim of healthy tissue, a few millimeters thick. If that rim is thick enough and there's no cancerous cells in that margin, then they call that a clean margin or a negative margin. That's a successful surgery. If there are cancerous cells in that margin, a couple of millimeters margin, that's called a positive margin. Once they've figured that out, that's when the patient has to go back for potentially another surgery, and it's not considered a successful surgery. Okay. That's the key: getting clean margins. That's the goal, and that's what we're here trying to solve. It's a very, very hard problem.
Even the best surgeons struggle with this. The challenge is that cancerous, especially microscopic cancer, you can't see it, you can't feel it. As a surgeon, you're trying to make those decisions based off of imaging that you had either preoperatively or imaging that you've got inside the operating room, like X-ray. The challenge is the resolution of those imaging modalities is not good enough to see cancerous cells. It really is at the cellular level that you need to look for that cancer in that rim, in that margin. The result of that is generally another surgery. The re-excision rates, the re-operation rates are really high across the board. Right? Double digit, 23% for breast cancer, which is where we're starting, but high across pretty much all other cancer surgeries. This is a big problem that just hasn't been solved to date.
I'll talk a little bit about what's different, what we've got, why it looks like we're on a clear path to that. This is just a little bit of data. Everything you see here is on our website, so you can download it from there. Basically, the left-hand graph is the top eight cancers by volume in the U.S. The dollars, the primary Y-axis, is how many dollars are spent in the U.S. every year just to do the re-operations, never mind the first operation, just the second operation. You can see for breast cancer alone, just in the U.S., it's almost $600 million a year wasted because it was not done right the first time. Close to 200,000 surgeries in breast cancer. It's a little bit why we chose that one to start with as the first cancer. It's across the board. Right?
You can see even down from that, although it's smaller, the impact can be greater in different types of cancers. If you look at the right-hand graph, even more importantly than the cost savings, the five-year survival rate of a patient that doesn't have clean margins on that first operation goes down, and it's not insignificantly. Right? The first bar on the left is breast cancer, where there's a little bit 90% to 80% survival rate after five years. The next bar is head and neck cancer, where the survival rate is down a lot versus clean margin versus not clean margin on that first operation. Not only is there a huge cost saving, but it's obviously a huge quality of care benefit to getting clean margins. It's across the board. Right?
From a patient standpoint, even if you do go back in for a second surgery and that second surgery is successful, there's a much greater risk of complications, 66% increase in the risk of complication, infections. You're going back under the knife again. Right? That's not a totally risk-free thing. Huge increase in complications. Any treatment that the patient has to have after the surgery, chemo or radiation, can't really happen until they confirm whether those margins are clean. That is delayed from the time that in standard of care, you're waiting for an answer back of whether you got clean margins. The cosmesis is affected for breast cancer. Of course, there's a waiting period to find out if you got clean margins, and it can be up to 10 days.
That's super traumatic for any patient that's in there wondering if, did you get the cancer or not? Is it out? Am I clean? Am I not? It's not great. System strain, hospitals, constrained ORs, nursing shortages, all of these things are very impactful to hospitals. Not only that, if you look at breast cancer, it's not the most profitable use of those constrained resources in the hospital. If they can use those constrained resources for other types of surgeries, they can actually make more money. If they have to bring patients back in again, the hospitals get affected because some of the insurance companies don't pay for that second surgery, so it's on their dime. It's a huge impact onto the systems. The economic burden, which is mostly borne by the insurance companies or the payers, is massive as well. Right?
Anywhere from another $12,000 on average if the patient gets another lumpectomy to much more than that if it's a mastectomy. Okay. How do we solve the problem? Our approach is using a different type of imaging modality versus X-ray or ultrasound. It's called OCT. It's optical coherence tomography. It uses infrared light waves. It's similar to an ultrasound, but much, much higher resolution. Compared to ultrasound, X-ray, MRI, it's orders of magnitude higher resolution. The resolution of OCT is around 10 micron. Cells, cancer fat cells, are on that same order. X-ray, which is the next best, is 100 micron. How do you see a cancer cell if the tool you're using to look at the thing you're trying to look at has a resolution that's way bigger than the thing you're trying to see? You can't. Right?
That's the challenge right now with what technology is out there. Three key advantages of this using OCT, which is why we chose to go down the path. One is you can see that resolution is there. You can see that cellular level detail. The second is you can see into the tissue. You're not just looking at the surface of the tumor. Remember, you need to get those margins that are several millimeters thick. It doesn't matter if there's no cancer on the surface, but there's cancer in a millimeter. That's still problematic. You have to be able to see inside, and this technology can penetrate a few millimeters into the tissue and look at it. There's no injectables. The doctor doesn't need to inject some chemicals into the body, into the patient ahead of time.
It doesn't come in contact with the patient, so the safety risk is almost negligible. What we've done with OCT, okay, so OCT as a technology has been widespread use in the medical field for years, for decades. Ophthalmology, we've probably all had it done when we go to the eye doctor looking at the retina in cardiology. Those applications use a very small, let's say, a 1 centimeter by 1 centimeter image. Right? Your eyeball is not that big. Inside the vein is small. The challenge, though, is when a tumor comes out, it's bigger than that. What we've done is we've created something called wide field OCT, which allows us to image a much larger specimen. Our device goes up to about 10 by 10 centimeters.
It's big enough for a surgeon to take a tumor out, put it on the machine, scan the whole thing in one sweep, and be able to look at that entire margin. Okay. That is, let's call it technology set in our company number one. I'll get to number two in a moment. This differentiates us versus all else. This is a picture. Okay. Three pictures on this screen. The top picture is what the surgeon sees in the operating room in real time. What you're seeing here is an image a few millimeters deep into one slice, like one side of a resected piece of tissue. What you're looking for in there is you're looking to see what am I seeing? Am I seeing healthy cells or am I seeing cancerous cells? Okay.
If you can see, I don't have a pointer here, but sort of to the top left side, there's those white circles. That's cancer. You can see the right almost close to the very top, which is the surface. Right? The entire width of that's about 2 millimeters. Obviously, you've got cancer sitting in the 2-millimeter margin that the surgeon, when they remove the tissue, didn't catch it upon removal. They put it, the surgeon put it on the machine, scanned it, and we saw it. Most importantly, the gold standard, what the standard of care is, is take all of that tissue, send it down to pathology, and have pathology look at it under a microscope and then report back on whether those margins are clean. We have to make sure our technology correlates against what they would say.
The bottom picture is what they see. This is the same piece of tissue. As you can see in the bottom picture, that dark purple thing is the cancer that they would see, and they'd report back and say, "Hey, look, you have a positive margin. You need to go do something else with this patient because you weren't successful." The important thing for us is to be able to prove to the users, to the surgeons, that what they see on our screen correlates to what the pathologist sees after the fact. This is just an example of that correlation. It's very important for them. It does correlate very clearly. The system's made up of four key components, let's say.
This all sits in the operating room, the cart itself with the imaging technology and the screen, the computer that does all of that. There is a specimen immobilizer. You use one per patient. You put the tumor inside of that. That goes onto the machine, and then scanning happens. Just to better understand, it's almost like a photocopier. The scanning happens from the bottom, inside of the machines, the camera, and the light source and the computer. Specimen immobilizer. Number three is our proprietary imaging atlas. What that is, is a library of images. We've had thousands of patients go through our machine. We've created this library. We use this library to train our internal users. We use a library to train our surgeons, the clinicians that are out there in the field, our customers. To number four, it's obvious, right?
We also use that library to train an AI algorithm. The device we have on market right now consists of those first three components. The device that's going through PMA approval right now adds the AI algorithm to it. Okay. Let's talk about what happens as a flow with our device versus standard of care. Standard of care is on the bottom. Surgeon takes tissue out, resection, closes the patient up, sends the patient home, sends that tissue down to pathology, and then waits 10 days for that piece of tissue to be looked at under the microscope and to get an answer back on clean or positive margins. If it's a clean margin, great. Everything goes on after 10 days. If it is a positive margin, the surgeon needs to do something next, which is generally a repeat surgery.
With us, surgeon takes the tissue out, puts it on the scanner, takes images, looks at it, and then makes a determination. Do I see something that makes me believe I need to take a bit more tissue? If so, the patient's laying right there. They take a bit more tissue. Does it look clean? I can close them up, send them home, take that tissue, send it into pathology, get the confirmation. What that does is it allows the surgeon, to some extent, get that information that normally they have to wait for 10 days within about 10 or 15 minutes while the patient's still there, which just saves that entire loop of 10 days and another week to get the second surgery scheduled, etc. The real-time aspect of this is key.
We have product out in the field with surgeons using it and paying for it. We're above 2,000 patients right now on the commercial side, hundreds and hundreds more through clinical trials that we've done. It's at a high growth rate. The nice thing is we're at a point right now where surgeons have had, our customers have had enough experience under the belt where they're now talking to their colleagues and saying, "Hey, look, this is working for me. I know you probably have the same problem I have. You should really look at these Perimeter folks and look at what they're doing. It's helped me. I think it can help you." We're seeing that information sort of that flow from the surgeons to their peers.
We're also seeing it flow up into the hospital administrators and then down to other facilities within the same hospital group that do not have the machine yet. It is a really great endorsement out there. We have done some, we have done some white papers internally and studies from our surgeons internally. I will move to a blinded clinical trial that we just completed recently. This is from one of our internal white papers from one of our surgeons. It was her first 72 patients, I believe, that she treated with our machine. With that, her re-operation rates were down to the 5.6% range. Compare that to an average of around 20% is a very big delta. Right? We have been able to see that on the device we have out there with a new, this is from when she started. Right?
With a new surgeon using it without the AI, remember, this is pre-AI, that technology already being able to see these types of results. If you peel that back layer to the next layer of the types of breast cancer, there are two key types of the majority from rather what they call DCIS, which is more an earlier stage, which is much, much harder to detect, and IDC, which is a later, more advanced stage of cancer. These are some of her results with the data set we looked at. Right? With DCIS, the re-operation rate for that is up to about 30%. If you cannot see it from the back, the dark bars, we did a big study of 30,000 patients across the U.S. multi-year looking at commercial insurance information as well as Medicare information.
From that, we're able to study re-operation rates across demographics and across commercial versus Medicare patients by age, by types of cancer. So 24%-30% re-operation rate for DCIS, she was able to get down under 14%. And then for IDC, which is again the most advanced version of cancer, where the re-operation rate is something around 12%-18%, she was able to get to 0%. So she was able to find all the cancer and get a clean margin in that first operation. Okay. So that's all based off our current technology we've got in field right now. The next generation technology, as I mentioned, takes basically the same hardware platform, everything you saw there, and adds a software layer on top of that. So there's two key benefits to this. One is it does make it slightly quicker to use in the operating room.
Really, the most business-impacting part of this is it very much reduces the barrier to entry for new surgeons wanting to adopt the technology. The way to think about it is surgeons are cutters. They're not readers of, they're not radiologists. There is some hesitation when first introducing the product, "Oh, I have to learn how to read this image." We help them with that, and we get folks across the line. What that does, it does somewhat limit the adoption rate to those surgeons willing to take that on. With an AI assistant, now all of a sudden, the barrier comes way down, right? Because the AI is running through all the images, looking at all the thousands of images that get created in every surgery, and then marking and highlighting where it sees potential for being something that's suspicious, some cancer.
The surgeon only has to look in those maybe handful of pictures. That barrier comes way down. That is what we bring to market. We have a library of over 2 million images that we have gathered ourselves. I will say this. The value of AI in any industry, in any company in the world, is not so much the algorithms. It is the data set. The algorithms are pretty much all open source. They are relatively easy to get. The data set, if you have proprietary data, then you really have value in your company. That first technology node I talked about, the imaging device, that is ours. No one else has it. No one else can create pictures. No one else can create the data set. This is proprietary, it is non-replicable unless you have our technology.
This provides a great moat around us in terms of other people being able to undercut us by recreating what we've got. Anyway, we've got 2 million images. That's what we use to train the AI that we ran through the clinical trial. It highlights those areas of interest. The reduction of the training load for the surgeons is important. Here's how we're going to walk through it. Right? The product we have on market right now is the far left, the S-Series. That's everything you saw from a hardware perspective with no AI. That's what we've got right now. What's going through PMA approval with the FDA as of March, we submitted the application, is what we call B-Series, OCT plus AI 2.0. We had a first version of the AI.
We went through trial with the second version of the AI, and that's what's getting reviewed right now. What that does, you can see in the picture, it'll scan through the image and then highlight with those little red bars on top, "Look here, look here." What you don't see in this picture, which is very important, is actually the first screen generally the surgeons will tend towards is a thumbnail screen, which just has little squares of all the areas of interest. From that, they can either just make their decision based off of the thumbnail screen, or they can click into it and then get this and be able to inspect it a bit deeper. It makes it very, very simple to use. That's what's running through FDA right now.
Back at the ranch, we've already got 3.0 available where we've trained it with even more data. We've improved the algorithms. The hardware inside is all local inside the machine. Every year, the hardware is getting faster and faster. We can put more complex algorithms on there and really get that compute time very, very short. We'll get the B-Series through approval. The way we've done our application, we've done it in such a way that there's mechanisms for us to be able to upgrade without having to go through a whole new clinical trial with next versions of the AI. Very promising, better specificity, better sensitivity. Okay. This is a brief summary of our trial results. We had a press release on this a few months ago.
It was a 300-patient trial, 200 in the device arm, where we evaluated within patients how the margins were after standard of care. Then for a subset of the 300 patients, 200 of them, what the results looked like after they used our OCT device. It was powered for statistical significance. We hit the primary endpoint with a p-value of 0.0050. Just this past week was the annual Breast Cancer Surgeon Society meeting in Las Vegas, where the principal investigator, Dr. Alastair Thompson, presented the findings to all the breast surgeons in much more detail than this. We press released this morning. You can take a look at it with a whole lot of results out of that, which were very, very good. The long and the short of it is that we had 88.1% margin level accuracy.
We were able, of all the patients that went through with positive margins, where after standard of care, they had positive margins, we were able to affect positively 40% of those. Even more interestingly, above and beyond those where pathology said after standard of care, it looks like you've got positive margins, we were able to find another 12 patients where pathology said, "Your margins are clean." Then when they looked at the tissue that the surgeon removed after using our device, said, "Oh, shoot. Actually, it wasn't clean. It's a good thing you took that." Those are patients that would have gone home after the surgery. Pathology would have told them they're good. In reality, they wouldn't have been good because we saw something in us in using our device that the pathology missed. Just a quick soundbite on that.
When the pathologist does sampling of that tissue after the surgery, they're actually only sampling 1% of the entire surface. That means 99% of that surface they're not looking at. We're able to image 100% of the surface. This is a known problem actually within the industry. There is an opportunity there for us to really drive that message home and help the whole medical profession sort of understand that this is a way to get a little bit more confidence on whether those margins really are clean versus just sampling a small, very small section of things. To some of the economics, there's about a 20% re-excision rate, 20% + depending on the type of breast cancer across the U.S. It's about $17,000 extra cost just for that operation itself.
If you just sort of extrapolate those numbers out, that's where you get to the $600 million or so of wasted cost in the system. If you just do the math, right, if you're able to take that 20% down to the numbers we saw in that small little white paper of 5.6%, we're able to just deliver immediately from that almost $500 million worth of savings just from where we are today, never mind however we get better, and then just in breast cancer alone. On top of that, or as important of that, is the 29,000-some-odd patients who don't have to go through a second surgery, get back to their life much quicker. Right?
This is a huge, huge benefit both to the insurance companies as well as the hospitals as well as the surgeons, and of course, as well as the patients. It goes from there. Right? Those numbers I've talked about are only in breast cancer. The tech, there's nothing special about breast cancer that the tech doesn't work across other cancer types. We've already started to collect images on head and neck cancer, which was the second bar in that graph, skin cancer, lung cancer. We're starting very shortly here with a surgeon down in California. We see the line of sight for the breast cancer business, and we're able to sat in our to expand it.
If you can expand the business beyond just breast cancer, all of a sudden, your TAM goes from that $600 million + or a billion plus to much, much greater than that. This is kind of the roadmap for us. We are in 2025 right now where we are S-Series. That is a product without the AI. We are expanding that while the FDA is reviewing the AI version of the product, building that base of customers, building that base of users, doing UI improvements on the device we have right now, which we will carry forth onto the B-Series, getting clinical data out there, really building the groundwork for the B-Series, which hopefully will be approved later this year, early next year, and be able to bring that into market. As we get into next year, that is when commercialization of the AI product starts to scale.
We can start to not only do that, but also start to ramp up some of those other cancer types that we talked about, how we're starting to collect some images there. This is the team. Andrew Berkeley, who's a co-founder, is here with me as well. The company was founded in Toronto. We now have about a third of our employees in Toronto, two-thirds down in the U.S. In the U.S., we're headquartered in Dallas. Andrew's up here. Most of the technical team is still up here. We have a broad set of experiences within the team. Touched upon the IP. The IP is very well protected. We have a number of patents, both U.S. and Europe, across the key hardware components. There is a lot of trade secret as well in terms of making the hardware work.
As you can imagine, at that sort of 10-micron resolution, manufacturing process isn't easy. We outsource it. There is a lot of work there to calibrate it precisely, to be able to transport it from the factories to the hospitals very, very precisely. It is very well protected from that perspective. On the software side, the image library, it's not patented, but it's trade secreted, of course. On the software side, we've got a lot of protections around the company as well. We're on the V. We're on the OTCQX. That's some of the details. The major shareholders are Social Capital and Master Holdings. That's Chamath's firm. We're covered by those three firms you see up there, Leede Financial, Paradigm, and Raymond James. Just to summarize, we've got it's a very big market.
It's a relatively high gross margin business that we've got right now that we continue to see growing in the future. Once we get the AI out there as well, then there ends up being some sort of software revenue model that we can bring on top of it, which has got more reflective of those types of gross margins as well. Even without that, the 75% gross margin is kind of where we're seeing things right now. Very positive reception from our current customers. We've got the B-Series coming through. That will open up a whole bunch more customers for us. Hopefully, before March of next year, we sort of anticipate about a year approval process. Big question around the FDA, what's going on there. Our review team's intact. They're highly engaged with us. We're on email exchange with them every week.
Hopefully, fingers crossed, that continues the way it is. That's it. Any questions?