Applying AI in Ultrasound - Helping Clinicians Become More Effective
Ohad Arazi is President & CEO of Clarius, a company that provides a high-definition wireless ultrasound system that includes AI to help reduce the complexity of ultrasound. In this episode he shares about his time in the Israeli military and the surprising environment that encourages entrepreneurship, what you can learn while working in large and small companies, where AI is and where it’s going, how point-of-care ultrasound can benefit clinicians and patients, and how Clarius and their high definition, wireless ultrasound system is putting a powerful tool in the hands of all types of clinicians.
Links from this episode:
Mastering Medical Device:
Episode Transcript
This transcript was generated using an automated transcription service and is minimally edited. Please forgive the mistakes contained within it.
Patrick Kothe 00:31
Welcome! Hey, have you heard of that new thing called AI? It's seems like everywhere you turn you hear about AI. I wouldn't be surprised if I went into Starbucks someday and they told me about AI powered coffee bean. AI is such a buzzword. And it seems like some people are throwing it around, thinking they don't want to be left out of the party if they don't use it. But AI is actually can be a really powerful tool. And there there are some products already in use, and a lot of them in development that are truly using AI to solve big problems. Our guest today is Ohad Arazi, CEO of Clarius, a company that provides a high definition wireless ultrasound system that includes AI to help reduce the complexity of ultrasound, and therefore making it a better experience for the user. He's also got quite a bit of experience in applying AI in medical imaging, Ohad led Zebra Medical Vision, and also contributed in senior leadership positions at TELUS, Health Change Healthcare, and McKesson. In our conversation, we discuss his time in the Israeli military, and really surprised me about what the environment was like, and how it encourage entrepreneurship. What you can learn while working in large and small companies, where AI is and where it's going, how point of care, ultrasound can benefit clinicians and patients, and how clarius and their high definition wireless ultrasound system is putting a powerful tool in the hands of all different types of clinicians. Here's our conversation. Ohad, we've got some really interesting topics to get to today. But I'm always interested in how people find their way into our great industry. So what was your path?
Ohad Arazi 02:30
Yeah, Pat, thanks for that. I started my tech career in the military. Actually, I was a product manager in the army in Israel. And when I left the military, I joined a startup company that was founded by some guys that came from my unit, which we eventually sold in 2005. So subsequent to that, I moved to Vancouver, my wife's Vancouverite. She dragged me here kicking and screaming at first, but I've grown to love it over the years. And through complete serendipity I ended up at McKesson, McKesson has medical imaging headquarters was here in Vancouver, and I had come from telecom. And from the Intelligence Corps, I knew nothing about medical devices. And yet, I was completely hooked. And I had an amazing 11 year run at McKesson, I loved it. It was such a value driven company. And those were the years also when PACs, which is the software infrastructure for managing medical imaging, was really catching on. And it was just high growth. And we we had kind of all the benefit of being part of this amazing fortune five company. And yet we were in Vancouver, we were kind of removed from it. So we had all the benefits of being part of a big corporation, including access to CIOs and to providers into a channel. And we didn't really have any of the overhead kind of the big brother big corporate. And so that was really my first entry to the segment, let me tell you that I'm never going to leave healthcare, it's a one way door as far as I'm concerned.
Patrick Kothe 03:55
Got it. So let's go back a second in the military background. Tell me a little bit about the military background and what you learned there and are applying to your career.
Ohad Arazi 04:05
I see it's not an uncommon story for tech entrepreneurs or leaders in tech from Israel, in the military in Israel, first of all, is just such an innovative and disruptive framework to grow up and like most people anticipate or expect military to be extremely hierarchical and very structured. And I'd say that in my unit in the Intelligence Corps, it was really fostering innovation and disruptive thinking. And I was an officer there but as a even as a lieutenant, you have tremendous span of control. And you have, you know, a canvas to be able to be heard at many levels. And therefore I think that in many ways, kind of my innovation and disruption roots were laid down in the military, which is maybe not what most people expect, but I think is very indicative of the experience. Many Israelis are many entrepreneurs coming from Israel have with their military background in many ways, just like in Uh, in North America, often your alma mater is kind of your first network, as you start to move into the professional field. In Israel. It's the military, it's the connections you make with people in your unit, that many of which are then going into tech starting up companies. And so I came out of the military at the age of 23, with a very rich network of connections and a lot of practical experience in being a product manager, that basically accelerated my career tremendously.
Patrick Kothe 05:27
So military is mandatory in Israel, is that correct? It is, is it considered to be part of your education more or less?
Ohad Arazi 05:37
I'd say it's part of your life journey, your education, most Israelis come back to do a degree like I went back to school, and I did a law degree. But I did it at a different place in life, right, I started school when I was 23. So I had already had some life experience. under my belt, I was a commander had, you know, lead people, and therefore my approach to completing my degree in university was quite different. But um, I'd say it doesn't necessarily facilitate a faster trajectory in your education, but it gives you I think, better skills to get your university done quicker with more focus, because you're just in a different place in life.
Patrick Kothe 06:14
It's really interesting, because because I would not have expected what you what you described to me, I would have expected to be very regimented and not having the ability to do freelance to do to do different different things in there. That is a really interesting thing. Is was that just in your regiment, or your your group? Or is that a crossed military, in your experience with with other people?
Ohad Arazi 06:41
I'd say there is a foundational ethos in the Israeli military of, you know, leveling the playing field and being relatively flat in terms of have, you know, hearing the voice of every individual in the service, but certainly, it will vary by units, right. So like frontline units will tend to be a little bit more regimented, because kind of, you know, command and control models are needed to be able to execute their versus working in more of an open ended intelligence environment. Give us more, I think latitude. So I would say it does vary, but there is kind of an overwhelming or an underlying, I'll say, approach, which I think is something that is generally very common in Israel, Israel is a very flat country, in terms of everyone feels like they can relate to the Prime Minister, right, it's like a very, kind of that culture is more informal, and is very direct kind of hierarchy. And, you know, doing the thing that's expected of you isn't always isn't always being put forward, but rather, you know, kind of doing the right thing is, is where the emphasis is being put.
Patrick Kothe 07:44
So you jumped out of the military directly into a startup now the startup community or in Israel is something that's been very well studied. And people understand that the startup culture in Israel is something that's embedded within within the community. So what what is it about that emphasis on startups? How did how did that start? And how did you catch that startup bug coming right out of the military,
Ohad Arazi 08:16
part of the story of Israel is kind of like success against all odds. And I think that's what you have to feel in a startup company, right? Like, everything is stacked against you, right? The big guys, Legacy incumbents, you've got no resources, no money, you typically don't know everything you need to know. And maybe that's actually an advantage. And I'd say that some of the disruptive approaches that I mentioned, coming out of the military, and just the general ethos of the country, like find a way to get it done, is very telling I think of the startup culture that evolved. And I think what's been interesting, you know, we've seen a lot of startup hubs. In other jurisdictions as well, actually, including even here in Vancouver, there's a very vibrant, smaller company kind of startup community. What's been interesting about Israel is that it was able to cross the chasm by having entrepreneurs that sold their first second company reinvest, and then foster back that ecosystem and give back to the ecosystem, put their capital back in and found their next company. And then over time, it became not just a country of startups, but of scale ups, and actually corporate headquarters are moving to Israel. And that's something that actually, if I jump forward to my time here in Vancouver, we've been trying to replicate of how do we create that next level of take the successful companies and have them reinvest? We're still not there yet. I'd say in Vancouver, we have a lot of smaller companies, eight to 10 person startups, including a lot in our space in the medical device field and digital health. But we haven't yet seen that kind of vibrant circle of kind of reciprocity of like second and third companies being found and building out a layer of executives that are then spawning the ecosystem back. And that's been I think, in many ways, Israel's real success story, even more so than the initial moniker of kind of calling Israel startup nation. I think actually It's almost more telling that it's become scale up nation, because that's actually a much harder chasm to cross.
Patrick Kothe 10:05
You come out of the military, you do the startup thing. And then you come to the US and you go to a large company and you stay for that large company at that at that large company, McKesson for 10 Plus yours. That's a little bit a little bit different pathway to because seems like when somebody gets gets a startup bug, they stay startup startup startup, you went to big company right after that, what were you thinking at that point in time?
Ohad Arazi 10:33
You know, I'd love to say, Pat, that my career was a series of well thought out decisions, but the reality is very few careers look that way. Right? It's kind of a, you know, a bunch of inflection points and pivots. I never thought to your point that I would be like a big corporate guy. And I joined like a fortune five company, 200,000 employees, you know, 200 billion in revenues, like imaginary numbers, in terms of the size and the scope of it. But as I mentioned earlier, you know, there was something about our medical imaging group, it was about a half a billion dollar business. So so not small, but certainly quite isolated. You know, McKesson is by and large, a drug distribution company, low margin, high volume. And at the time, it had a bunch of digital health businesses, and it really afforded us a lot of control over our destiny. And, you know, we were high growth, high margin business, which also gave us a lot of cachet. So I think it was just the right environment, to grow into and to walk into and to learn the ropes. I mean, I came back after having run a company at the age of 29. And being a CEO, to now coming in as a director of product for a much larger organization. So a step down. And yet I learned so much in those initial years, became VP, and then ultimately, SVP and GM running that business. You kind of have a perception of what big corporate is like when you're looking from the outside in. I have to say that at least at my time at McKesson, the company was all about customers like we would go to like the big Leadership Conference, the CEO of McKesson at the time was a gentleman by the name of John hammer grin. And you know, he would get up on stage and only talk about customers who wouldn't talk about shareholder value and optimizing EBIT da. And it was really very telling of like, how value driven and how strategic this company was. When I left McKesson, I think it was 186 year old company at the time. So like five year plans were short term plans for McKesson was so thoughtful and strategic about what it's doing, which was really, you know, to your earlier point, kind of counter to like this kind of quick thinking startup mentality exit at all costs. But I think that balance actually resonated well with me. And interestingly, when I decided to leave McKesson after transition to become change, healthcare, I went to another big corporate for for another almost three years. After that time, though, I had already been 15 years as a big corporate guy, which I never thought of myself as one. But I guess the numbers don't lie. And after that, I really felt okay, now, it's really time to take these skills and come back to my startup roots. And really what I miss most at that point, was touching the customer and touching the product. You know, when you're leading a really large business, I was at TELUS, which is a large, publicly traded company here in Vancouver, running a big part of their digital health business, as you know, 1400 employees, I felt there were eight layers between me and the product, eight layers between me and the customer. And I missed that. So I really wanted to roll up my sleeves, and get back into it. And then I came back actually, to my Israeli roots as well, and took the reins of an Israeli startup company by the name of zebra medical, which was an imaging AI company.
Patrick Kothe 13:38
Now you're leading clarius, which we're gonna get into quite a bit in the ultrasound space. But I want to focus the next part of our conversation into zebra medical, and what you were doing at Zebra medical, we hear so much about AI and and what it is and what it isn't, and how it can be applied to, to the, to the medical space. So can you give us a little bit about about zebra medical and how you think of AI within medicine?
Ohad Arazi 14:13
Yeah, absolutely. And, you know, to your point, so much has been said and written and thought about AI in the context of healthcare and when it first started, really, medical imaging was at the forefront because out of almost every aspect of digital health. Medical imaging is the one that lends itself best to automation because it's so data rich, right? Every image we take across any of the key modalities, you know, ultrasound X ray, CT, MRI, mammography, is by and large digital today, and therefore, there's a lot of information, a lot of pixels and zeros and ones, which can be interpreted and grouped and patterned in order to drive better outcomes. So so that's why medical imaging was really at the forefront of adopting AI and healthcare we've seen that evolve in a variety of other places like decision support, for example, or triage. But when it first came into medical imaging, there was a lot of talk about, you know, is this going to replace radiologists. And our approach at Zebra med had always been, we will not replace radiologists, radiologists that use AI will replace radiologists that don't. So we were likening this to, let's say, the autopilot in long international flight, like when I fly between Vancouver and Tel Aviv. You know, that's a, you know, almost 24 hour journey to very long hauls. And I'd like to know there's a human behind the controls. But I'd also like to know that human when they're navigating along six or eight hour shifts at the controls, they have a machine next to them that's helping them to do mundane tasks to make sure they're not missing information. And that's really the role to me that AI can play in medical imaging, it's not to replace the really complex tasks that radiologists are cardiologists does, but it's to help with the more mundane activities, the more one, the ones that are more time sensitive, the ones that have to integrate a large amount of data at a very rapid pace. And so we've seen that evolution, it's still really hasn't realized I'd say its full value proposition from a medical imaging perspective, at least. But I kind of see two main use cases or workflows emerging as we think about the application of AI in healthcare, or specifically in medical imaging. And the first word most of the industry started is post processing, so the image is acquired, and then we apply AI to better interpret it or better triage it. By having you know, the artificial intelligence look for patterns in the images kind of navigate all those myriad of shades of gray and pixels that reflect what medical imaging looks like to identify anatomy or to come back with a better recommendation. And this is all today cleared by the FDA under a category called Computer Aided diagnostics. So think of it as a tool that's aiding the principal investigator, in this case, the radiologist or the cardiologist to make a better interpretation. And in that post processing workflow, I'd say there's two main parts. The first one is triage. So AI systems that read the images as soon as they're acquired, and basically before the radiologist has had a look. And then they're reprioritizing, the queue of the radiologists to promote the studies that have the highest potential value, for example, that might have significant findings. So the AI is looking for critical finding is moving it to the top of the list and then the the actual radiologists will read the study based on the priorities that the AI has, has driven into the into the workflow. The second part of that is AI tools that are focused on productivity are basically helping to automate some of those more mundane tasks that I mentioned earlier. Like in the example of the autopilot, where the AI is maybe drawing polygons or highlighting key anatomy or key findings that we want to make sure the radiologist doesn't miss, and is also helping to automate some of the measurements that are often time consuming for radiologists, I'd say that the jury's still out on the real value of those post processing workflows, especially on the triage side. Because that's, I think that we'll have a lot of value over time, if we can ever truly remove a radiologist for a specific type of procedure. You know, again, for the more mundane ones, like if we look at something like chest X rays, chest X rays are a very common procedure, they're taken in many emergency departments ICUs, almost any inpatient that has a chest or abdominal condition, we'll have a chest X ray done, when they're admitted, vast majority of chest X rays come back as normal, meaning no significant findings. And that, to me is one area where I'd love to see the industry as well as the regulator push forward to say for these types of procedures, we've now trained an AI to rule out maybe the 20 most prevalent and common conditions that are visible in chest X rays. And for these types of procedures, we won't require a radiologist to override the exam that will have a lot more value than just reprioritizing the cue because now we've truly taken a cost element out of the equation and improve the experience for the patient because they're being diagnosed sooner, and maybe being released sooner instead of waiting, sometimes even 1224 hours for that imaging exam to be read before they can be discharged. So on the post processing side, there's still a lot of work to be done. We're not there yet, but I see some breakthrough opportunities.
Patrick Kothe 19:32
So want to go back for a second and talk about what AI is and what it isn't. Because when you think of AI, what many people think of AI it's it's something that's continuing to learn and continuing to independently think about things and the regulations right now. To my knowledge, don't allow that to be out in the wild. It's like you can you can have a learning set And then you have to freeze that. And then you put that software out there. And now that is a frozen thing you can continue to learn. But then you have to update from a regulatory standpoint, this is the new the new machine learning or the new software, that you're going to implement it based on AI or based on machine learning. Is that correct? Understanding?
Ohad Arazi 20:25
It is Pat. Absolutely. So the way that the regulator, let's take the FDA, as you know, really the regulator that setting the bar globally on the approach of artificial intelligence in healthcare, they think of AI first of all, in the context of what I noted earlier, which is Computer Aided diagnostics. So it's a tool to aid the diagnosis. It's called CAD Computer Aided diagnostics, it's not a full rule out or rule in tool. So again, it's not cleared for its own independent thinking. And in that CAD category, there are several sub designations, which are, you know, allow the images either to be annotated or measured, or even for the AI to recommend a classification, like to look at the image and say, I think that this is what this condition is. But in all those cases, again, the final decision on the diagnosis or triage of that image is subject to a human in the loop, commenting on it, and the process to get a module cleared for each individual finding through the FDA entails three main steps, which is training. So first of all, training the algorithm to identify the pattern, show them a lot of pictures of a pneumothorax, which is a perforation in the lung and say this is a pneumothorax. And this is where the preparation is. And this is how you draw a polygon around where the pneumothorax is visible. The second is tuning. So after you have an initial model, you run it at scale to tune it and to show a different types of images, different quality, somewhat different vantage points, perhaps to be able to have that same outcome. And then the third step is validation. And the validation set is a set that is blinded to the AI meaning it's not the set that it was trained or tuned on. And that validation set basically has to prove that the AI is performing as good or better than the gold standard. And the gold standard could be a predicate device, another device that has been cleared to do the same thing or a similar thing, or a human reader. And so that is done on an indication by indication basis. And to your point, once a line is drawn in the sand to say this is what we are approaching, or attempting to clear with this module that is now frozen. And any further enhancements to that or changes to what the AI can diagnose or can identify will entail either an amendment to that existing clearance or a new clearance altogether.
Patrick Kothe 22:43
We have data that's coming in at that training set that's coming in. And as we're describing this, from an imaging standpoint, that training set are the images. But there are other things to help to describe a or diagnose something, and they may be blood work, it may be something else, physical or pain or whatever. There's other other pieces of information to help diagnose that. Where do we stand right now for using some of these cross platform things to be able to train an algorithm?
Ohad Arazi 23:19
Yeah, what you're referring to Pat is what I'd consider longitudinal data or longitudinal records. So today, I'd say most AI is trained on images on the pixel data. And basically, pattern recognition is applied to say this pattern is this or that finding. Now of course, we can optimize that process, if we had more access to longitudinal information, which in the context of medical imaging could mean the reason the exam was ordered to rule out a specific condition or a little bit of background material. It could be the actual annotations, and the report on that image. And it could also be additional clinical context from the medical record, like lab work or prior history. I haven't really seen AI models yet that are integrating the clinical context with the imaging context to make a decision. I think that also the FDA right now is only clearing AI modules in medical imaging specific to the image data itself. So it's not yet allowing for additional context to help modify the disease state classification of that indication, for example, but the longitudinal record does have a lot of impact in refining your training data. Because basically now instead of having your AI trained on a very large data set that may or may not have the finding that you're looking for, you can train it on data that is already has kind of soft labels, because if we already know from that this report was produced on that image, that image has a specific finding, for example, or we know what the end pathology was. And so I think that that's very important, especially when you're doing deeper classification. Like if you're trying to train the algorithm to say, well, this two millimeter nodule in your lung is malignant, in order to properly train that I would actually need a dataset that had path findings. Because it wasn't wasn't just a suspected finding, but I want to train it on nodules that actually were validated through pathology, to say that they were malignant. And so that will certainly improve the fidelity, the specificity of my of my algorithm, but I haven't yet seen and I think it'd be an interesting breakthrough to examine, being able to cross reference information from multiple sources and having the AI integrate those two, to make a better decision at the point of care,
Patrick Kothe 25:37
isn't that we're where we should be going. I mean, taking as much information as you possibly can to make a firm diagnosis. I understand there's regulations that are going to need need to be put in place, but it seems like this longitudinal data adds more actionable information to help drive those decisions. So it seems like that that would be a next step in some of these AI platforms.
Ohad Arazi 26:03
No, I agree. Because today, the synthesis is being done by the human, the human might have AI that is helping to recommend a suspected finding from an imaging perspective may be an AI, that's helping to parse through the textual data and recommend a finding there, and then the synthesis between the two is being done by a human. I think to be fair, though, the degree of complexity even in a single data source is so high, which is why the FDA is not yet clearing for AI to fully rule out a condition. And it becomes so much more complicated when you integrate an additional source, right? If you were to bring Path and Lab findings, in addition to medical imaging, many complex elements that have to come together. And I think it's a reminder for all of us that, you know, generally healthcare is somewhere on the spectrum between science and art. And believing that it's fully science, meaning that you could just create an algorithm that takes all the inputs, and then comes up with the best outcome is, is probably still a little bit of a reach relative to where we are today. And the capabilities of our AI systems. I did want to just follow up on a point. I mean, earlier, I started to talk about post processing in AI, and mentioning that it has value. But I think that the ROI model there has not really been that well established. Another domain in AI and medical imaging that actually has a lot of value is more real time AI, meaning not AI that is there to replace the ultimate decision by a radiologist or by a cardiologist, but rather AI that will help us to position or to acquire a better image in real time, because it's helping us to optimize the view. While we're taking images that principally exists in ultrasound, which is a domain that I'm focused on today. In ultrasound, what we're seeing is that we can apply AI to detect anatomy while we're acquiring the image. If you've ever had an ultrasound image taken, you know that it requires a very high degree of proficiency, right? How do you position the probe the transducer, because it's leveraging sound waves, not a static view of of you know of ionizing radiation going through your body, and therefore the ability to position the probe is non trivial. And certainly as we think about the proliferation of ultrasound, putting it in the hands of more users, it requires software and AI to lower that hurdle rate for novice users to be able to acquire images. Today, most ultrasound images, certainly in a traditional radiology or cardiology or obstetric setting is being done by a highly trained professional called us demographer. That's a two year degree of someone who specializes in image acquisition for ultrasound. And what's really been exciting about seeing AI and ultrasound is real time AI, that is now enabling less proficient users, and ultimately, actually the patient to be able to acquire good enough images for meaningful point of care decisions. And so that AI is about still about the same concept of pattern recognition. But it's not doing it offline to come up with a diagnosis or a triage, it's doing it real time. To help arm the human will say, Hey, move a little to the left, move to the right, or I'll accentuate the contrast, or the depth of field or the game to give you a better image for you to make a better decision. So it's more kind of making the human a cyborg now they're kind of armed with AI in real time, as they're working through their, their healthcare process.
Patrick Kothe 29:21
Training a cinematographer the old school way was quite complex and took quite a bit of time to get to to all of those those nuances. And even now when I when I go to training courses, and part of it is was was snog, Rafi and emergency department and Ed docs, that's the business that I'm in right now. It's amazing how far people have to go, how low their, their level of learning is, and how much they have to learn to be able to implement point of care ultrasound, AI is going to help to lessen that or shorten that learning curve.
Ohad Arazi 29:59
Absolutely, you know, some of the traditional barriers of making ultrasound more accessible, have been addressed, we've seen the form factor shrink from these legacy cards that were, you know, kind of anchors in their department, we've seen them become portable, we've seen them become wireless. But I'd say the next big hurdle that we have to cross in terms of really making point of care ultrasound available in many more care settings, is the skill required to acquire the images is, is the usability. So we started with price when we talked about portability. And I'd say that the last frontier is usability. And ultrasound system requires, as we've discussed a high degree of proficiency. And it also has many inputs, right? There's all these knobs and buttons and external controls. And the reality is that often for point of care decision, AI can get you to a better outcome of seeing the image, you need to make that point of care decision, whether it's interventional, or diagnostic, quite seamlessly. And I think it's not too dissimilar to how we've seen digital photography emerge. I mean, if you think about photography, a lot of the AI and pattern recognition and photography is driven, not at post processing, some of it is, but a lot of it is driven at you know, allowing you as a lay user to take a very good image, right where in the past you you needed your your high end Canon D 200, which had a tremendous amount of controls. The reality is that your iPhone can often take a very comparable image because it's loaded with AI, that is adjusting the brightness. The contrast is finding the faces, the eyes, the smiles, all those things where you manually had to do. And I think that our approach and handheld ultrasound or point of care ultrasound, is very similar, where we want to lower that hurdle rate and make it much more accessible.
Patrick Kothe 31:42
So let's talk about Claire's for a bit. You mentioned that ultrasound has got a lot of different uses a lot of different form factors, a lot of different clinical areas that that it can go in, how do you segment the market? And what segment are you going after with clarius.
Ohad Arazi 32:02
So Clarus is a digital health company. We're based here in Vancouver, Canada, and we're on a mission to make accurate, easy to use and affordable ultrasound tools available to all medical professionals across multiple specialties. And that can apply equally to nurses in the developing world or EMTs. In an ambulance, family doctors providing rural medicine, or surgeons performing safer and more accurate procedures. So we focus on a wide variety of use cases and care settings. But the specialty aspect is at the heart of all of them. So you know different from some other players that are more trying to create like a generalist tool, maybe like more of the equivalent of like a kitchen butter knife, were like a block of chef's knives, right where we have 10 different probes, seven for human medicine and three for veterinary care, that are each focus on specific specialty use cases. Because in order to bring the power of you know, high performance ultrasound imaging, with in such a small form factor, we really had to optimize it for specific use cases. And I think that that is a very important barrier for entry historically is that handheld ultrasound was perceived to be more of a toy kind of a novelty thing, as opposed to a real tool that can give you access to imaging that's on par with what you get from traditional card or complex systems.
Patrick Kothe 33:25
So ultrasound probes are designed differently for different things, whether you're looking at blood flow, or you're looking at an anatomical issues, where it is in the body, what type of structures are in there. So there's a lot of different types of probes. So what you're doing is you're designing probes for specific applications.
Ohad Arazi 33:44
Exactly. And we do that again, with a very unique form factor, I'm holding our ultrasound here in my hand, it's the size of a smartphone, it costs under $3,500. And yet, it's a full fledged ultrasound machine with a quality output that's comparable to cart based systems that we have all seen at the hospitals and you can appreciate how different this is then the big old cards that get wheeled around with all their cables and knobs and buttons. And the idea that is not only fully wireless, but it's powered by artificial intelligence, which makes it very easy to use, even by clinicians that haven't been trained as an ographers, for example, and that to me is the big advantage of handheld ultrasound is that you can make very complex diagnostic decisions, or guide sophisticated interventions at the point of care. You don't need to send the patient out for a test and wait for the result. You're right there at the bedside, making the decision on the spot and getting an immediate outcome. And in order to do that, we really have to lower the hurdle rate so that you as a specialist that are focused on a specific procedure or type of anatomy are not an expert in ultrasound, we're now giving you another tool to superpower you to make a better to get a better outcome right at the point of care.
Patrick Kothe 34:57
So the device that you showed me doesn't have a screen on it. And obviously, with ultrasound, you need a screen. So tell me about that.
Ohad Arazi 35:04
Yeah, the way that it works is we activated Bluetooth module to uncover tethered devices that are running the clarius app. And so that could be iPhone, iOS, or most recently now Microsoft devices, so tablets or phones. And once it finds one, it establishes a connection using Wi Fi Direct, which means it's not reliant on the hospital network or the network and an ambulance or other care settings that might not even have a Wi Fi network. But rather, the Wi Fi is driven from the device itself. It's right in kind of this middle part of the of the probe. And that creates a connection back to let's say, to a tablet, by way of example, and driven by AI much of the workflow on the tablet is substantially simplified. Essentially, if you know how to swipe and you know how to pinch, you can use our ultrasound because as you pinch, for example, you can change the depth of field, and the AI will automatically adjust the game will automatically adjust the contrast, instead of you needing to adjust those on a manual basis. And that's what allows you to maintain very simple controls, using tools you're already familiar with, like an iPhone or a tablet, in order to drive the workflow in comparison to kind of this fully integrated system that has its own screen, its own probes, and a bunch of external controls or inputs that need to be manually adjusted by the end user.
Patrick Kothe 36:25
So you would need several of these probes depending on what type of patient population or what type of issues that you're, that you're looking at. So if you're if you're a primary care physician, you may may need a few of these probes to deal with the issues that you're dealing with.
Ohad Arazi 36:43
I'd say Pat, most use cases focused on a single probe, because it will be reflective of the type of procedure or the type of point of care decision they're making. Like, for example, orthopedic surgeons will usually use a linear probe with relatively high frequency, because they need to be able to visualize relatively superficial structures. And they need to see them in a very high resolution because they're often using ultrasound to guide a procedure. Like for example, they're holding the ultrasound with one hand and needle with the other. And they're guiding the needle real time to perform an injection, let's say in the rotator cuff, which is a great alternative for pain management, in comparison to let's say, taking narcotics today, which is you know what, what used to be the gold standard for this, there are a few care settings where more than one probe is useful, I'd say emergency rooms are probably the main one, where you'll see a balance between two types of probes one that is more of a generalist kind of chest and abdomen probe for lung, and abdominal conditions. And the second is a cardiac focus probe. Usually, to focus on cardiac imaging, you need a dedicated probe, because the form factor of the transducer of the modality has to be quite small has to fit in between the ribs of the patient. In order to acquire that good, you know, apical or for chamber view of the heart, compared to what you would be using for more general abdomen and chest you'd be using usually a curved array because you want a bigger field of view, you need to have quite quite deep penetration, you're dealing with organs that could be quite deep in the body, especially depending on the patient's body mass index. But that's probably the main care setting where we see dual probe use. Interestingly, in most of our other users, orthopedic surgery, plastic surgery, medical esthetics veterinary, we're seeing more of a single practitioner single probe for the vast majority of the time.
Patrick Kothe 38:33
So when we think of a physician, many times we have a mental image of them with a stethoscope around around their neck. But they also have pockets in their, in their white coats as well. Should they be carrying a probe along it and use it much like a stethoscope.
Ohad Arazi 38:51
Absolutely. And I love the visual picture that you painted Pat, because I think that's a bit of a view of where healthcare is going. I mean that the stethoscope is a 200 year old miniaturized megaphone, right that, you know, I think even you know, younger doctors today are much less apt to have a stethoscope. And I think the alternative to it is being able to see inside the patient's body right instead of just to listen. And you're right. So we we often see now, handheld ultrasound really becoming a personal device that is like the stethoscope to stethoscope wasn't a departmental asset. It wasn't something that you had to book and share. It was a reflection of you and your practice. It was right around your neck, kind of part of an extension of you. And that's that the role of handheld ultrasound versus legacy ultrasound, because the legacy cards were much more of an asset that was shared, you had to book it, you had to move the patient. It wasn't an extension of you was an extension of your department or your ward or your office versus handheld is really it's part of you. It's in your hand, it's in your pocket. And I think it's also really personalizing the experience for many of our practitioners. And so I absolutely believe that that's where we're headed. it. And I think that you know, the barrier for putting this in the hands of patients also isn't that far off. Because if you think about other diagnostic devices, if you think about pulse oximeters, or you think about glucose monitors, even EKG EKG is a full fledged diagnostic system. And we've seen some vendors, take an EKG, miniaturize it and create more of a closed loop system to deal with a single condition like atrial fibrillation, where we see monitors that are, you know, for example, the Apple Watch is cleared as a monitor to monitor atrial fibrillation that uses ECG technology. And I think that ultrasound is going to break through in the not too distant future, where for monitoring specific conditions, we will have it in the hands of the patient, of course, empowered by AI to the point we made earlier, where you're eliminating that operator variability and the degree of proficiency needed to acquire the images. And you're instead of making an open ended diagnostic platform, which isn't really a fit for something you'd put in the hands of a patient, we will have to convince the regulator that as a closed loop system may be monitored remotely by a licensed practitioner that is overseeing the exam in real time, you can actually monitor a condition. And so I think that that breakthrough is probably not more than two to three years away.
Patrick Kothe 41:17
So where do you sit? We clarius? What? Where are you approved for sale? What are your plans?
Ohad Arazi 41:24
Yeah, so we are now on our third gen product, the company was founded in 2014. And so we've got over 18,000 systems out in the field, in across North America and Western Europe. That's where we sell direct, as well as we're active in 70 countries through a network of distributors. So it's a class to medical device cleared by FDA, eu MDR, Health Canada and registered in 70 countries, so many, many other clearances that had been attached to it over the years. It's really exciting, I think, to start to disrupt the legacy or conventional approach to how medical devices were implemented and sold even it was usually like a b2b sales process that was long. But of course, for $3,500 device, we could never have it as a six to nine month sales process. So we've really made much more of a B to C model, around the whole experience from how you buy it to how it ships in under a week, and is in your hands in this beautiful white box, kind of the whole unboxing experience just evokes that personal device, consumer electronic that I think is creating that stronger connection, that stronger bond between the practitioner and the device, more akin to the example you gave of that very strong connection that physicians had with their stethoscope. What's been really interesting, just to add at a point to that is that it's not just doctors that are using it. So we are intended use is cleared for any licensed practitioner. And so given now that we've lowered that hurdle rate using AI, we're really seeing substantial adoption by physical therapists, by chiropractors that are performing medical injections, by nurse injectors, especially in facial aesthetics, Ma's pas, so a variety of roles across the healthcare ecosystem. And I think that's really key to alleviating some of the traditional bottlenecks we've seen in this kind of physician provider system, which now especially in this post COVID environment, where we're dealing with such substantial healthcare shortages almost everywhere, we really have to arm more practitioners at variety of kind of license levels, to have better tools at their disposal. And that's a big part, I think of what this AI enablement is driving is putting it in the hands of not only MDs, but many other practitioners that are licensed, but can deliver care and are accessible to their patients in a variety of care settings.
Patrick Kothe 43:47
Ultrasound can provide such great information, and putting it in a form factors. And using AI to be able to help people across this learning curve, think is really going to help with the practice of medicine, providing more and better information. So I wish you success. And your mission to bring this out further and further and more and more people. So congratulations on what you've done, and good. Good luck with the product in the future. I've got one one last question for you. And it's kind of where we started. You've gone from startups to large companies, back to startups, you've had a really interesting career. So if you're going to give advice to somebody who's kind of at the beginning of their career, how would you provide some information that could be helpful to them as they start to navigate how they want to build their career?
Ohad Arazi 44:45
Maybe one thing that I've learned is that what I've enjoyed and what I think propelled my career is the ability to vacillate between different environments. So we talked about big corporate versus startups and going back and forth on Actually, is what made the journey so much more enjoyable. And what allowed me to learn in different settings, right? I wouldn't take anything away from the years that I was in a big corporation had a lot of great people to learn from culture, you know, process, kind of organized thinking, starting with the military, through McKesson, and TELUS. But I also wouldn't take anything away from the startup experience that I had even early in my career, when you're walking into a four person company, and you have to learn everything, and you're making stuff up as you go along. And so the balance of those two actually is what is, is what has made it fun, right? It's kind of getting a chance to be Roy Disney, who was like, you know, if you're familiar with the Disney brothers, Roy was this hard nosed operator, super organized, execution oriented, delivered results versus Walt Disney, who was this kind of open ended dreamer, blue sky thinker. And so I always think about the relationship between like being Walt and being Roy, and really not having to choose who you are, maybe you inherently know where you'll end up. But in your career path, the ability to vacillate between the two vacillate between a startup and a big company, being Walt being ROI, is I think, what makes the journey fun, and what at least in my experience, really accelerated my learning curve.
Patrick Kothe 46:18
Such an interesting perspective on managing your career and and also on the impact AI is having, and will continue to have on the practice of medicine, a few of my takeaways, first AI in medicine, or how God knows what they do, and what they don't do. And he said, they don't replace the physician, they help the physician become more effective and efficient. And it's really important to understand and communicate this to the market, otherwise, adoptions not going to be possible. The second is, is learning wherever you are. And Ohio described how he went back and forth between big companies and startups. And that going back and forth, made him smarter. And while he was at the small one, he was thinking back on things that he learned in the big ones and vice versa. So he really grew where he was planted. The final thing is be ready for education. Many of us go directly from high school into college, but he had that stop and a military and that time where he was able to become more mature, have more experiences, and then go and learn at the at the university level. Many of us are already done with that with our education's from that standpoint, but the concept of being prepared to learn. So what are you prepared to learn today? How are you going to continue your education? Knowing what your past experiences are? What are you ready to learn today? Thank you for listening. Make sure you get episodes downloaded to device automatically by liking or subscribing to the mastering medical device podcast wherever you get your podcasts. Also, please spread the word and tell a friend or two to listen to the mastering medical device podcast as interviews like today's can help you become a more effective medical device leader. Work hard. Be kind
