I am so excited that Soner is here. He runs PONS Tech and we're going to talk a lot about it. His company has done some real good innovation work and more importantly, it's not just the innovation, you know, people can innovate, but his is real practical implications on improving patient lives.

Again, Soner, thank you very much for coming. I really appreciate it. Start with your story. Who are you and why the hell are you doing this crap?

Yeah, sure. And first of all, thanks for having me.

Like it's really fantastic to be here. So yeah, I mean I'm, I'm sonrs or one of the co founders at PONS, with my brother. We are twin brothers actually, so we don't look alike. Oh yeah, we are twin brothers. And he is also a Prof. At the radiology department, leading the AI labs.

Former Rutgers Academic Guy. And then also still doing something on Rutger with Rutgers, but also at UBC medical school. And the like the thing is, basically, first of all the reason why we started is like we like 10 years ago we lost our father because of late diagnosis.

And then one year ago our aunt, passed away again because of late diagnosis of breast cancer which could be avoided.

And it's like the thing is that it's not only personal, it's happening a lot recently.

And there are right now a lot of women like age 30, 35, having breast cancer, which is extremely young. And and what we, what my brother was focused on is basically okay, like ultrasound has this enormous potential because one,

It can be extremely mobile, it can be extremely small. But the biggest barrier was the low image quality, which was like, which was the biggest barrier and saying that okay, can we use ultrasound for preventive monitoring and then targeting early stage high risk group of people and population.

And then he started then developing these models and algorithms that can overcome these barriers. Like we're not doing like the trying to develop a new hardware, but okay, can we use existing hardware but improve the details on the imaging?

Our main focus is how we can basically bring ultrasound to the mass market with AI and then use that in technology for specifically for preventive monitoring and then reaching out to communities that are not able to go to the hospital, like minorities, low income families and then use this technology for early diagnosis and help radiologists not to replace them or not basically be better than radiologists, but more helping them to identify certain structures much more easier, much more faster.

Wow, what a journey. And I'm very unfortunate about your dad, so sorry to hear that. And I completely agree the two major issues that you have just decided pointed out, number one of course is the image quality, which again I'm going to decide I'm going to distinguish between image quality in terms of technical versus image quality because of, of tech, you know, acquisition issues.

And then the second one, of course you said how do we use ultrasound, in essentially real time preventative process.

We're going to come with first where the PONS comes in is especially that image quality. And more importantly how do you improve the acquisition signal as well. So talk more about it, what you do, what this technology does and then how do you see it moving forward in the future? Yeah, sure.

Soner:

There are a lot of companies doing image segmentation, automated measuring, pointing out or like doing the color coding everything.

Not only ultrasound and CT but also MRI. So. But when it comes to ultrasound there is this huge problem which makes it difficult as you said way in the beginning. Like it is, first of all, it is extremely user-dependent. It's not automated.

So not every device or every technician has is generating the same kind of image. And usually when you do image segmentation or like doing some AI on ultrasound, most of the companies are more focused on later stages which the lesions or like tumors are big enough to do a segmentation or to identify that part that is whether it's malignant or benign.

So what we do differently that is also one of the patented Parts like we do basically use the frequency level dynamics of ultrasound. So because at the end of, at the end of the day, ultrasound is RF frequency technology.

So what we do is like we said, okay, instead of trying to do something on the image, how about we do something on the frequency level, because that is kind of like, it's like a person seeing another person. So we as human, we see you is I see how we, I see you. Even if I am kind of, I don't see any color.

Not like, only like showing something there, but more revealing the futures because we can basically use that information and then reveal those features on very early stages that are not visible at all in ultrasound imaging.

So our our system like it is, first of all it is device agnostic. So it works on any device. It doesn't need like you buying any or upgrading your existing devices or ultrasound either is point of care or card based system.

So that is one of the interesting thing and the other thing that we have basically we have developed these, we call it like these digital contrast technology.

Which means that we don't need a, chemical agent injected to the patient to do a contrast imaging. We can do that without any chemical agent. We can do that directly on again on the frequency level that we can also reveal how the surrounding of those tissues are because that's also important to understand.

Last but not least, I think one of the interesting things that we have is because that is again working on using the RF data from ultrasound and then using the Frequency level dynamics.

We can basically scale the data set. It is not a generative AI feature, not like synthesizing imaging, but we can basically create real image from real image like by 50 times and then use that information to optimize the accuracy of the risk assessment systems as well.

And that is the biggest difference that we are making. And that is the only way we can say that, okay, we can reveal those features that are really early stage which basically may even like stage zero which you have, you are in the high risk group, you don't have any symptoms.

Usually the, the reports show nothing. But there is, there is something that is not visible. We are aiming to make that visible.

So now you have done some amazing jobs

So two questions. Number one, A, I understand we're not going to do generative AI and you want to make sure that we are actually. But more importantly number one, image augmentation through AI in ultrasound. And then in general, how do you see this technology moving forward in the future?

That's a good question. One of the things that we are seeing is the enhancement models that we have is basically focused on can we reveal like certain features?

That helps basically because one of the things that we don't get rid of the original image. So that is one of the things that we never do, basically, because. Because of the clinical efficacy and all this kind of stuff. So all regulations, like the radiologist, they always see the original image as well.

And that is also one of the advantages. They can choose whichever image they want to enhance and then get the results on that image as well. So because one of the things that we don't want to do that. Okay, we are basically deleting all the original images and creating a new image that is, based on.

On the, on the ultrasound scans. That is not what we do Because we don't want to change anything on their existing flow. We don't want to check. Create a new radiology flow. We don't want to change their existing, like habits.

But we are basically more. Okay, here is more detailed information. So, and that is one of the things like with data augmentation and especially when we talk about ultrasound, I think one of the biggest, things that is happening right now,\A lot of, A lot of health systems are right now looking ways for utilizing ultrasound for triaging. That means that, okay, like can we use ultrasound not in the hospital settings for regular things like either fast scans or others or just regular scans, but more can we use it outside your hospital settings and then reach out to those people that are again, like in a certain group, but the user will be a nurse.

For example, we cannot send every technician on the field and do the scan, but the user will be a nurse or a caregiver, basically a novice user.

In our system, what we are saying that, okay, you just need. It doesn't matter if you over tilt it. It doesn't matter if you move a little bit fast and not too fast.

That will create this enormous distributed and decentralized imaging network because then you will be able to reach all those people and with the help of, of the software then with the augmentation and enhancement we will basically show all this information that was never collected before.

And that is I think the biggest change that I am seeing the acceptance of that is happening really fast right now. Like, which means that okay, like like an ultrasound can be used anywhere, anytime, by a health professional, like by a nurse or carrier.

That will make a huge difference in how we call this preventive care or like

So from a clinical use perspective, regular radiology workflow, What if, how do you help clinicians adopt this technology? Soner is right. Radiologist behavior change is going to be even a bigger problem than FDA.

Thoughts on this Ed and Harvey?

Then I'll jump in real quick. So first, of all, thanks for coming, man. I appreciate it, what dropped me or drew me to the company is as a physician, I thought, oh my gosh, I've been told multiple times, hey, I need a line over here, can you help me?

Let's do ultrasound. And when I went through residency back in the day, we were taught how to do ultrasound. And when I saw the ability of, giving this tool to a nurse to someone that wasn't trained, but then looking at it from a AI healthcare preventative point of view, I thought, oh my goodness, if this could be at the local CVS, where I as a physician, like, hey, swing by the CVS, get this ultrasound done, then we all know that the future of medicine is going towards the home, meaning the hospital.

There's no need in the future to be going to the hospital. The more that can be done at my home, the better.

That Ed, I'd love to hear your thoughts. Yeah, Ed,

two things that came most quickly to mind.

One is always finding that clinical champion or business champion, you probably need both. So I'm coming from the point of view of health, system. You know, our hospital is, who's that champion?

So I can get it pretty excited about it, but the more champions you can get. So if you're, if you're a non clinical person, obviously you're going to want to find the clinical champion. If you're a clinician, you are going to have to find your counterpart, right. And come up with, a dyad situation.

And that's the only way that I found to be successful is if you come at it from either clinical only or business only, it's not going to happen. So find that dyad and then, and then you do some experimentation. You know, if you wait for the big picture, if you wait to do everything perfectly right, it never happens.

So you've got to do some skunk work. You gotta like allow stuff to happen on the side, right. Which most CIOs won't do and they'll be, oh my gosh. But, that's why we're stifled with innovation. So you gotta allow for some skunk works. Then you come back as the dyad and you're like, look what we've done.

Look, look at the capabilities. Not just theory, it actually works, works and we have these outcomes, then you can build the momentum. The second thing is, yes, you could lower the cost of medicine, but oh my gosh, you got to follow the money.

If you disrupt revenue cycle, they'll never happen. Right. And that's again another reason why we've retarded growth and innovation in healthcare is just because incentives financially so very important to work with, you know, the finance brothers and sisters and figure out how do we make this happen that doesn't hurt the hospital finances as being gun honest. Right.

Because they're going to be so protective of that. It's like, even though it's a good idea, yeah, we should move more of health care to the home. 75% of things that happen in a hospital today could be done at home. But the main reason we don't do it is I think financial is the number one reason we don't do it. So how do we work to figure out, you know, it's about whether it's a value based care scenario or something like that.

Can somebody share my screen right now? From the committee. There you go.

So, this is where the mammography is gonna come. And that's, I mean, truly excited about it as well. I have limited information, but we're going to. The whole point is that how we're going to rewire medical imaging and preventative care together. And this is exactly what ED works.

And they said from going from the hospital to the home and essentially basically going from a reactive care to a proactive care where, you know, you can share an ultrasound, and then you can do this. And then the system's under pressure because people think, the money is the problem, the money is not the problem in the sense that hospitals would be okay doing it.

what do you see the ultrasound as a player in preventative, proactive care.

So I think one of the things that we say that we, because we are focused on right now breast cancer and liver diseas. But one of the things with mammograms is also done what we see is, which is also affecting ultrasound as well. So one of the things that we, that we have done during that study is basically, okay, what happens if we use our system combined, with ultrasound and mammogram. But what happens if we implement enhancement results on dense breast tissues and show the details in a much more detailed and easier way to understand and like to make diagnosis. And we also, during that study we also had access to additional metadata which is basically race, ethnicity, ethnicity, race and age, and location and also like the the background of those people as well.

Because usually like that is where AI comes becomes biased because it's not trained on different type of data. That is also something that we are, that we want to overcome because with our technology we can basically close this gap and create more diverse data sets to optimize the risk assessment, of the AI features as well.

With our tech, with the, with the study that we had, we achieved 64% improvement on the AI accuracy for early stage diagnosis. It is like a huge improvement. Even like 10%, 15% are like really good. But 64% was focused on what happens if a radiologist use only existing images, like only existing ultrasound, only existing mammograms, and then use that model to identify early diagnosis versus what happens if they use our.

With our image enhancement results. Like original image plus image enhancement results plus scale data. And it was like a big difference. So that was one of the biggest interesting outcomes on the study that these 64 improvement. 64 improvement.

So the Bell curve, the problem statement was to improve image quality for improved detection and diagnosis.. And the idea behind it is that the way that he has achieved that is basically that understanding number one, decreasing three.

Three important things. One is image enhancement, which is already done, but image enhancement at the level of, or basically control for race, ethnicity, breast density. And then last thing that he has actually controlled for is the operator itself. Because then even if you have, less experience, operator image enhancement can improve the overall outcomes.

So Ed and Harvey, do you think that these kind of technologies are going to be only cash pay or you're going to see that insurance companies actually, quote, unquote, incentivizing everyone by demonetizing these kind of preventative cares in the hospital and saying, get it there because it costs me less.

So it's a very different dynamic we are in. So Harvey, and Ed, thoughts on that

The way that we finance healthcare makes innovation and is the killer of people and creates less than ideal experiences.

So it's really frustrating. I was hoping that value based care would but it's been slow to adopt and there's good and bad with it and then usually it's such a carve out that it's really not covering entire populations for everything

But until we align incentives and truly put the patient at the center, I don't know how this is going to work. So man, I don't know that I have a solution other than I would love to see value based care more holistically adopted making those hard choices.

All I was going to add is, I agree with you, Ed. It's a tough one.

But you know what? This is why we're doing this podcast. We need a lead. Right. We need. But lead by example. One idea would be, I know this is kind of moonshot, would be to go to a, major insurance carrier and be like, here's what we got. And if they are the main insurance one and they see the writing on the wall, they're like, you know what?

Yes, let's do this. And then I really feel the rest will follow.

So the same same, thought we go to the government, that's a moonshot again. But say, hey, this is what needs to happen and if that happens, I can see the rest of the industry following.

Beautiful.

So number one, does your current technology, does longitudinal. Number two, talk in general, we talked about breasts, but where do you see the technology going forward?

And some insane new use cases.

Like okay, dense press is like like a big problem. Like that is still. It's not only ultrasound problem as I said it's a mammogram problem. And what we can basically reveal that is like much more without needing like these rescans

So with the enhancement results, it's basically like really easy to identify what's going on without doing those repeated scans. So that's one thing. But on the other hand, the same technology is also like indication agnostic as well. So it can be directly adjusted without needing like months of training or others or like hundreds of thousand data.

And coming to your second question, I think what is becoming really interesting with, within our field is the potential of creating these extremely valuable biomarkers.

for example in certain cases that we, that we do a study on, on Alzheimer's or dementia or others really rare disease.

So that is how we see the future will look like. So, and one of the things that like two years ago when we first were speaking with Harvey, we were saying that okay, like ultrasound will become this consumer product. But at that time everyone's saying it is impossible, it will not happen.

But right now, two weeks ago, we had a meeting. One of the five big medical device companies approach us and they said, okay, we have these ultrasound devices, but we are aiming to, to develop a consumer version of that.

Real quick. What do you think? Big picture. What, what's the one thing you think your company is going to do for healthcare in the future?