← Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
How Avatar Medical Is Democratizing Medical Imaging
00:29:43
Avatar Medical turns CT scans into instant 3D models. See how Dell Pro Precision and NVIDIA GPUs are reshaping patient care.
Loading summary
Transcript35 lines
- [00:00]
A
Foreign. Welcome to Reshaping Workflows with Dell Pro Precision and Nvidia, where innovation meets real world impact in high performance computing.
- [00:21]
B
Welcome back to another exciting episode of Reshaping Workflows with Dell Pro Precision and Nvidia RTX Pro and GPUs. I'm your host, Logan Mahler. Have a very exciting episode today. We're mixing a little bit of healthcare and life sciences, a little AR VR and we're gonna get right into it. So we have Avatar Medical, we have two people actually on the podcast today. We have JJ Ross from Avatar Medical and we have my co worker Veronica, you've never seen on the podcast before. So Veronica, why don't you start a little introduction about you, what you do at Dell Technologies and then we'll move on to jj.
- [00:57]
A
Thanks for having me, Logan. This is my first podcast, so I appreciate you including me as I make my debut. So my role has recently changed. I'm super excited to be supporting Promax Workstations Pro Precision as we change names once again. So now I'm going to be focusing on healthcare, life sciences and energy industries, which I'm super excited to be learning about.
- [01:19]
B
Excited to have you out front and center right where you need to be. So jj, what about you tell us a little bit about your background, what you do at Avatar, and then we'll just jump right into it.
- [01:30]
C
I appreciate it. Thanks, Logan. At Avatar, I'm the clinical director in the U.S. so I come from a background of healthcare. My education says I'm an rn, but I've had a lot of, a lot of other experience in the healthcare field. I started out in orthopedic sales, so I sold joint replacements for 15 years. Hips, knees, shoulders, and parlayed me into the med tech world where I found a home at Avatar Medical. So I've been with the company since its inception. We're about coming on six years old now and I kind of wear many hats. I'm the clinical director, but I also work and am very involved in the technology front. So kind of what makes our software run and hence why I'm here today. Right. Was introduced to Veronica about eight months ago and obviously Dell's been supporting us for a number of years at this point. So excited to be here and share what we're working on.
- [02:19]
B
Fantastic. So up until, you know, Veronica introduced you to me, I had no idea what Avatar Medical did or was. So let's just start there, like, you know, 30 seconds to a minute, you know, what do y' all do? Like what, what do you provide Avatar Medical?
- [02:34]
C
We take raw imaging, DICOM or CT information and we bring that into our software solution and it creates an instant 3D reconstruction. We then use that reconstruction in a, in multiple ways. Primarily our use case is centered around the consultation space. So where patients see their physician and where physicians explain their imaging makes sense.
- [02:56]
B
So, I mean, I haven't been to hospital in a while. I think the last time I got like an MRI or a ct, I think I was getting my tonsils taken out. I think it's been. Or maybe my appendix out. It's been a while. So I think everyone's familiar with that kind of CT and MRI scans. So we take kind of the 2D version that you remember, like getting plastered up on that, like, light thing in the doctor's office to a 3D scan. Right. Question. When you say 3D, are we talking nerfs? Are we talking Gaussian splatting? I mean, you don't have to get any proprietary technology, but it is really cool to take something from a 2D right into a 3D. How does that actually work?
- [03:32]
C
So that's, that's kind of our secret sauce, right? I mean, so you, you bring up a valid point. 3D imaging isn't necessarily new. Right. It's not revolutionary. But, but we feel like what we're doing is evolutionary in the way we treat those images and the way we, we leverage those images. Right. So we're doing a reconstruction that's largely based on volume rendering. So we're voxel averaging and we're creating, you know, a 3D image through a technology called ray casting, which again, isn't necessarily a new technology, but it is new as it relates to, you know, giving physicians, you know, very fast, reliable, accurate 3D reconstructions when and where they need them.
- [04:10]
B
So that also kind of comes into, you know, doing a little bit of research before this episode into kind of VR. Right. And from your platform, how does VR kind of fit into the, the 3D scans that you do, you produce basically at, you know, I'm not going to say the bedside, but let's say at the doctor's office, from the 2D, like scene CT and MRI, how does VR fit into that workflow?
- [04:33]
C
Like I mentioned earlier, I've been with the company for just about six years, and when we first deployed, that's really what our focus was in. Right? So 3D imaging and then leveraging that information in virtual reality as a surgical planning tool. Right. So I don't think it's a surprise to anybody that VR has had kind of its ups and downs in the marketplace as it relates to healthcare applications, but really seeing something with depth is the value that VR brought to us. Right. So being able to manipulate an image to being able to, you know, understand depth capacity, you know, like this tumor sits superior or anterior to this blood vessel. That's all information that's harder to relay and to, you know, recognize in just a standard 2D image, let alone or a 3D image on a 2D screen. Right. Or in a non VR environment.
- [05:20]
A
So JJ, can we double click into that? Like, obviously it's great to see these Images in a 3D way because it's more visual. You know, as humans, we are super visual. And when it comes to explanations and understanding, can we double click more into, like, what are the clinical or workflow challenges that originally inspired the creation of the software? Like, what pain points are you really solving for that really pioneered this innovation.
- [05:45]
C
Yeah, that's a great question. And it's something that we've kind of differentiated ourselves with from inception. Right. So essentially, in medical imaging, there's basically two ways you get a 3D image. The traditional way, the most common way, is through a process called segmentation. So segmentation is, I always use a M and M as an example. Right. It's the candy coating on the, on the M and M. Right. So you're basically, you're applying a coating or a wrap or a mesh around a certain structure and then you're creating an object out of that. And traditional segmentation can happen in a couple different ways. One of them is through AI. Right. So AI is the big push, the big boom. And that's an automatic segmentation scenario. It's more challenging to get through regulatory channels like FDA approval or CE marking. And depending on the training of the, of the AI, it's not always accurate. Then there's semi automatic, which is you have a computer do it, but you're having another practitioner or tech confirm it. And then the third way is just somebody's manually, you know, drawing out the structure. So when we came to the marketplace, those were kind of the three options that we were up against. And when you talk about leveraging 3D imaging, there's, there's costs associated with that.
- [06:56]
B
Right.
- [06:56]
C
It's either time, which is probably the most common cost. Right. But then there's also extra staff, Right. Somebody that's making 3D reconstructions for you, or a radiologist that's outlining a structure to process this 3D, you know, image for you. And so really, the Background or the basic goal we had was to be able to do that same thing with the same accuracy, but have it available at the bedside for the physician. Right. So being able to instantly create this model and then tease out a region of interest in that model and then leverage it in a matter of seconds versus, you know, traditional methodologies can take anywhere between 25, 30 minutes for something basic or simple to several days. You know, the other aspect of that is not all 3D imaging is done in house. There are some companies that offer, you know, services where you send imaging outside of the hospital and then that, that reconstruction is done and then it's sent back into the hospital. And so as we are all aware, the data laws and phi regulations are strict. Right. So hospitals really enjoy it when you tell them you can provide them with a 3D image without sending it off site and then bringing it back into the healthcare environment. But that's kind of where our motivation lied, right? Is like we can, we feel like we can do this in a, in a more efficient way. We feel like we can do it just as accurate and we can get, you know, modeling in the hands of physicians when and where they want it. That was the, the primary focus and still is.
- [08:24]
B
So. Got a question, it's been from a couple years ago I was reading is that you received, and I'm not going to be an expert in the FDA here, but it was an FDA like 510k or L or some kind of clearance for your kind of VR surgical planning solution. So what I want to know is curious because it's a very highly regulated industry, right? You know, like you said, you know, you know, personal health information, you know, lots of disclosures. Like, I mean, this is more philosophical question, but I guess maybe not philosophical but more of a how did the fda, you know, that clearance that you received and how did it clearance and the rules around that influence where you wanted to take the development of Avatar. Did it kind of limit some of the things you could do? Was it not really a roadblock at all? I'm kind of curious how having that type of regulatory, I guess, oversight or approvals required affected how you developed your ultimate solution.
- [09:21]
C
I mean, so from a regulatory standpoint, right, what the FDA really cares about is accuracy, right? Like is it accurate? Right. So we're taking a DICOM image, say it's a CT scan and we're going to do a 3D reconstruction on that. We have to, you know, prove that we're within an error margin of accuracy to the FDA as it relates to the original image. And the good news for us is that actually we're just reading the dicom, right? We're taking the information straight from the source and through, you know, algorithm and mathematics, we're, we're displaying that image then in a way that the user has full control over. Right. So I think it's important to note we are FDA approved for surgical planning, but we are not FDA approved for diagnostic use. Right. So meaning a radiologist can't build an avatar in our software and make a diagnostic decision based on the image that he or she may see in our software. But I think that, you know, our 510k clearance was. It's 510k just basically means a piggyback. Right. There's other similar technologies that have gone through the, the FDA approval process. So we're able to piggyback on some of that existing technology that's been proven. But then certainly we, myself and a colleague of mine from France, we personally did all of the, you know, measurement testing and image testing and all of that for, for our FDA approval, which that same information was then, you know, submitted for our CE marking in Europe. You know, it's a pretty grueling thing to go through. And then, you know, especially when you think of surgical planning, if a physician's going to leverage the information we're providing them and then that, that information is then going to, you know, translate to the patient, we better have it. Right. Right.
- [10:59]
A
It better be accurate from that perspective. I wanted to touch on that as well from like a clinical. Right. Whether it's the physician or whether it's the patient, what's the feedback been like in terms of working with that high resolution 3D medical image versus the traditional 2D image? What is the feedback that they're receiving and is it really moving the needle to having a better understanding of what's happening to the patients? Right. Like, what's the change there? Because I think we can all kind of empathize or sympathize with that, you know, experience of, hey, I'm looking at a CT scan and I'm not really sure what I'm looking at. So how is that making a difference for the patients and for the physicians as well?
- [11:37]
C
So I mentioned earlier, right. That 3D imaging in the medical space isn't new. Right. But what's, you know, the evolution of how we're handling it and where we're providing this image is new. Right. So, you know, two years ago we were able to, to raise money and actually receive VC backing on this patient experience, you know, business case. Right, and what does that mean? It just means that patients, when they come into their physician's office and the physician has a CT or an MRI of that patient, using our software to educate the patient on the imaging that they had previously done. Right. It seems so basic, but it's very impactful. So I've personally been in around 200 consultations where our software is being used in front of real patients. And we like to democratize this information. Right. Connecticut scans, MRIs. It's black and white, as Logan mentioned. The doctor comes in and shoves it on the light board. You know, that's, that, that's a little bit older than my time in the industry, but I do remember the lightboards or they pull up the black and white images on a computer screen and, and they're like, no, this is your. This is a brain tumor. Can't you see? And it's gray and it's white and it's 2D. And you're trying to understand and process this information as a patient, and it just doesn't work. Patients don't understand it. They, they truly don't. What we try to do and what we say is, like, we want you to see it like the physician sees it. Right. And we, and what that does is it creates clarity and it builds trust. And they're seeing it in a way that they've never seen it before. Right. So when you, when you combine this use case of know leveraging 3D in the consult room with the technology of being able to instantly render that image, what you get is a business model that's, that's scalable. Right. Because now, previously you couldn't build 3D reconstructions for, you know, say a busy surgeon has 30 patients in his or her clinic in a day. I'm of the belief that all patients deserve to be educated equally. Right. So why not every patient have a 3D reconstruction? Well, the reality of it is, in the previous technology and the way we used to do those things, it's not scalable. It's too cost inhibitive, it's too time consuming. There's no way to deploy that with our software. And I mean, quite honest with what Dell is helping us do is deploy those models in the consultation space and we do it instantly. You know, the average runtime between pulling the imaging from the PAC system to a 3D model is about 15 seconds. So, you know, the name of this podcast has the word workflow in it, right? And so workflow is critical in the health care Space. So when a physician comes in to see their patient, you know, that workflow of pulling imaging straight from the electronic health record straight into our platform, and within 15, 20 seconds, they have a 3D model that they can then leverage to help educate the patient is. That's really the revolutionary part, Right. Or the evolutionary part, rather, to what we're doing.
- [14:33]
B
Super cool. I was going to ask. I mean, and I think I understand it, but I. Maybe I don't. Maybe I don't know. So, like, with the rapid surgical preparation, right. I'm like, kind of checking out your website and it's like, are you really, at the end of the day, taking a picture of the inside of my heart to prepare me for heart surgery and doing it in 3D and turning that picture around where a physician or a surgeon can put on a pair of VR glasses and. Bad analogy. But Ms. Frizzle My Heart and get inside my heart riding the magic school bus to check things out. Is that really what we're talking about here?
- [15:08]
C
Yeah, certainly. Right. It's a use case. Right. Like I mentioned, you know, our kind of primary focus right now is centered in the consult room. It's centered around patient education. But certainly our 3D models are. They're beautiful and there's a ton of information there. And so physicians that want to leverage that for surgical planning, they're allowed to. Right? That's. That's what our FDA clearance is approved for. And so, yeah, if you have, you know, complex surgical scenarios and you want to kind of rehearse, you know, what that anatomy looks like or what the pathology looks like, or kind of your, you know, think of it as like a blueprint or GPS of the body. Right. And kind of rehearsing that scenario. It's a very common use case for. Of our software. It can be done, you know, whether you're looking at it on a computer screen or you're looking at it through a pair of VR headset or an auto stereoscopic screen. Right. Like, those are all ways that you can leverage those models in our software.
- [16:02]
A
So, JJ, I know you mentioned that 3D imaging isn't new. Right. It's been around for a while, but y' all are essentially leading, right, from a PI. Like, y' all are basically being pioneers into how 3D can be used in a different way and really streamline efficiencies and stability. Right. What is the adoption rate like? I know you mentioned you've done a lot of consultations with, you know, clinics and with patients and, and learning, having learnings and feedback. Do you feel like there's still some thought, leadership and awareness to really push this forward or do you feel like this is an easy win for hospitals and physicians to start integrating into their day to day?
- [16:43]
C
So my personal feeling, right, is that, you know, everyone deserves to have the most impactful consultation or most rehearsed surgical strategy or. Right. I think every patient deserves to have a physician that, that truly cares about their performance as a physician as well as, you know, their communication quality and style to them as a patient. I think that's hugely important. The, you know, I think. But that's kind of my own take. Right. But we, we live in America and healthcare is a business, right. So in terms of adoption, we're working with a lot of large institutions throughout the United States. University of Miami Hospital, Hoag Medical center in Newport Beach, California, Mount Sinai Hospital, Northwell Health, the Billings Clinic. So, you know, we're really kind of in the early stages of commercialization in the United States as it relates to this, the patient experience use case. I think that it should be widely adopted, obviously. Right. I'm probably partial to what we're working on, but I think that patients deserve to understand and have trust and building, you know, understanding of their anatomy or their pathology. Right. It's, it's very clear. And you can, you know, if you just a simple Google search or a grok search in an AI platform, it's very clear to. That we know, and it's not made up that when surgeons have surgical plans, their surgeries go better. Right. We know that when patients understand their pathology, anatomy, potential side effects, the, that they're better patients. Right. They follow the rules more closely. They have a better grasp and understanding of what's being done to them or what their options are. They make more informed decisions. Right. So these are all really, you know, high impact scenarios. And you know, I think lastly, I'll, I'll say that with the patient engagement piece, right. This patient experience piece that, that we're leaning towards so heavily right now, it actually, you know, lines up with the business case of the hospital really well because we're helping to prevent something called out migration. So this is when, you know, Veronica, you, you tore your ACL, right? You get an MRI of that ACL and you go see Dr. Smith down the road and she tells you you need surgery. And you say, you know what, Dr. Smith, I didn't really care for your bedside manner. You didn't, you spent two seconds with me, told me I needed surgery. I'm not comfortable with that. I don't really understand what's going on. And you go get a second opinion, right? So that's called out migration, right? That physician spent time with you, told you you needed surgery, you weren't happy, so you went somewhere else. And there's a lot of different reasons. People travel for healthcare decisions, right? Or people travel for, for healthcare consults. But one of the main one is they just don't understand what the physician is saying or they feel like the physician wasn't communicating with them in a way they understood. And so by using imaging as kind of a meeting point between the physician's knowledge base and the patient's knowledge base, we find that more of those patients that come to your clinic stay. And when they stay, now that healthcare facility is seeing that, you know, that organic effect of capturing their business through, you know, the lifespan of that patient, like the kind of the holistic value of that patient increases for that facility because you chose to stay in that facility to have your procedure done.
- [19:56]
B
Let's talk about a little bit about the hardware, right? I know, and Veronica will probably get more into this, but let's talk about, you know, what is required to kind of run this workflow. Let's pretend, you know, Logan Lawler, the budding heart surgeon, wants to run Avatar Medical. What type of, you know, cpu, gpu, compute is required to kind of run this? I'm not going to say at the death side, but let's say the surgical bedside, what is required to kind of run it? And you said you kind of stay away from cloud, people shy away from that. So I know that the process of kind of making Gaussian spotting nerfs and that reconstruction is, you know, fairly GPU intensive. So we'd love to hear kind of about, you know, the hardware requirements. And then, Veronica, anything that you kind of want to talk about related to the, the Pro Precision or Pro Max
- [20:39]
C
hardware that you've set, that's where Veronica and I kind of came together, right? This software meets hardware world. The bottom line is this, right? Without appropriate hardware, our software wouldn't shine. You know, I don't think it, it's so much that we're, you know, that we're anti cloud. I think that's where a lot of this stuff is going eventually. Right? But since we are handling Phi Healthcare facilities, you know, I don't think they, they mind hearing that, hey, your information is going to stay within the walls of your hospital. Right. I think that we don't get any pushback from that. Right. They enjoy that. But you know, the reality of it is, is our software does require compute. And that historically has kind of been one of the barriers to entry of, of expanding or scaling this type of technology is where do the, where do the GPUs live, right? Like, what does that compute look like? We're working with Dell in two ways, right? One of them is through the Rack system, right? Having a rack of GPUs that are behind firewall, that are running our software at, you know, workstations throughout a healthcare facility. You know, the secondary use case that we most recently have been involved with and Veronica has helped a ton with, is driving our software in combination with an auto stereoscopic screen. So I know that we're going to have a discussion with Barco down the road, but I need to say their name at least once because they're a wonderful, awesome, brilliant technology and great partner of ours in this project. But I came to Veronica about six months ago and said, hey, we want to scale this autosterescopic screen in combination with our software. And I don't want to bring a T2 max tower into every exam room. There's not room for it. What can we do to scale this hardware down to make it more accommodating in a consult room? And so that's when we really, I say we, but it wasn't me, it was Veronica put her head together with her team and came up with this new build of a Dell T2 micro with the help of some of Nvidia's new technology in the Blackwell series of GPUs they're coming out with. And it's shrunk the form factor down to a micro box. I mean it's tiny. It can be mounted on the back of the, of the auto stereoscopic display that, that, that we're commercializing. And it's made a huge impact. Impact in terms of like our scalability potential, right. In hospitals there's rules. You can't put a computer on the floor, right? You can't like you have to have, you know, locks on the computer, right? There's phi in it. So somebody can't run off the street, steal it and run out the front door with it, right? So there's, there's different accommodations that we make in healthcare facilities for hardware. And so that's where Dell's been hugely helpful is ident and helping us with, you know, more of a form factor, you know, tech stack that runs our software beautifully. But it's also really easy for healthcare institutions to say, yeah, that's, that's tiny. That works great. You know, we'll, we'll accept that. And of course we get to hang on the laurels of Dell as a name. You know, like we're working with Dell. This is what we sell. We're in the Dell catalog. Right. And Dell brings, it brings trust kind of instantly, you know, borrowed credibility for us as a newer company to say we're partnering with a company like Dell, that really kind of breaks down some of those barriers for us really early on, for sure.
- [23:54]
A
And I think you mentioned in healthcare, obviously in other industries too, reliability and trust are super critical. And obviously having the hardware stability and the software verified and certified on that device is crucial to them making a decision, a best informed decision of how to move forward and not to make this a commercial. But yeah, you know, Delpro Max workstations, Delpro Precision, we have such a wide portfolio that we really do have so many different options, whether it's the larger towers, whether it's the rack, whether it's the small form factor, given that they're all configured with Nvidia graphics, that's really going to help push that reliability, the consistency and ensure that, you know, the hardware platforms are validated for the use cases that are really driving changes within healthcare.
- [24:43]
C
Six years ago, when I first started, you know, we were driving our software on a Dell G5, you know, laptop. I have one sitting over here in the corner of my office collecting dust. And it seems like it's 25 years old, right? It's, it's five or six years old. It's not that old, but it just speaks to how quickly the evolution of, of the technology is changing. Right. So I do think that, you know, a lot of our success in getting our software into the hands of our users is directly correlates to the technology getting better. Right. The form factor, the size getting smaller, more efficient runs, you know, runs cooler, less power draw, like all these things that we're, you know, Dell's helping us shape as it relates to hardware is just making our jobs easier to get software installed in more places. Right. It's kind of, kind of simply said, that's fantastic.
- [25:34]
B
So we've been going for about 30 minutes. So I mean, it's. Time flies when you're having fun, as my father likes to say. So, Veronica, any last questions you want to go to before we wrap it up?
- [25:44]
A
Since this is still a fairly new solution that's being implemented. Right. You're in the thought leadership awareness stage of adoption as healthcare leaders are evaluating this tool. Right. What Advice would you give them when it comes to aligning that software, that hardware piece, identifying their pain points, what would you recommend to them as they go through those, those motions?
- [26:09]
C
As the clinical director? Right. I have a really patient centered kind of approach to this. Right. I mean, I kind of tell all of our potential customers, like, listen to your patients. You know, listen to them. Like, give us the opportunity to put this in front of your patient or patients and just truly listen to what they have to say about it. And it's, it's very impactful. Right. We've proven that over and over again. And I think in terms of expectation, you know, as it relates to feedback from a hardware standpoint. Right. That some surgeons get into the weeds, like what makes it run, how does it drive? You know, like, they're, they're interested in that. Some don't. Right. They just want it to work when they need it to work. But I think that, you know, I think that's kind of the. My biggest takeaway is like, let your patients guide you to this decision and truly listen to them and you'll come to the same conclusion that I have, which is this is a really powerful technology.
- [27:02]
B
I mean, you kind of answered my question, but maybe recap it here is that, you know, let's say someone just tunes into. Is just tuning into the podcast. You know, you have 30 seconds to make your elevator pitch on. Why Avatar Medical, what would that elevator pitch be?
- [27:17]
C
I mean, I'd say Avatar Medical democratizes access to medical imaging. Right. We take something that's super, super complex and we quickly boil it down into something that's digestible for patients or for healthcare providers. Right. You know, surgeons, most of them will admit that they're not as good as a radiologist at reading medical imaging. Right. But so, so in, in a sense, we're democratizing that information for the surgical environment as well. Right. Certainly they're, you know, light years better at that than. Than the patients are. Right. But I would say that's kind of at the heart of what we're doing, is we're taking something really complex and we're trying to make it simple in a way that more people can understand.
- [27:56]
B
I love that. So if someone wants to, you know, connect, learn more about Avatar Medical, connect with you. Obviously, go to AvatarMedical AI is the best way to learn more, check out your platform is to click the free trial button. Or is there a better way for them to, like, reach you and start learning more in that process?
- [28:17]
C
Yeah, I mean, AvatarMedical AI is a great place to start. Yeah, if you click that free trial button, it'll have you fill out a questionnaire that goes straight to our Chief marketing officer. You can also send me a direct email. Jjavatarmedical AI happy to engage if. If there's any interest.
- [28:32]
B
So, jj, Veronica, appreciate the time. Always love hearing about cool use cases that Delpro Precision. Delpro Max Enable. It was a great episode. I mean, I don't know what I think. Maybe. I mean, I'm sure I'll have heart surgery one day and maybe it'll all be done in AR VR. Maybe. I don't know. My pitting heart attack. I'm losing my hair. It's. It's all gone. I'm a rusty truck. With that, we'll go ahead and wrap it up. So thank you always for listening. Check out the episodes below the links to Avatar. JJ's information will be in the show notes below as well as Veronica's. And with that, reshaping workflows with Dell Pro Precision and Nvidia RTX GPUs signing out until the next one. Thank you all. See you later. Later. This podcast was produced in partnership with Amaze Media Labs.