(00:00:08) Ken Ono: Welcome to Hoos in STEM. I'm Ken Ono, your host, and the STEM Advisor to the Provost and the Marvin Rosenblum Professor of Mathematics at UVA. Our goal is to evoke flights of imagination and wonder by showcasing the cornucopia of all that is STEM at UVA: the marvelous world of UVA science, technology, engineering, and mathematics. Now, my first passion in life was cycling, bike riding, and I entered my very first bike race in the early 1980s, dreaming of one day racing in the iconic Tour de France, which, well, never happened. In my 40s, struggling with my midlife crisis, I rediscovered that passion in a new light when I embraced the sport of triathlon. Now, triathlon consists of three sports, and I didn't really know how to swim so the first thing I did was I hired a personal swim coach and I also dedicated many hours each week to well learning to swim, biking/cycling, and running. I especially liked the XTERRA off-road triathlon series where we biked and ran on single track trails in the woods. Now, to be completely honest, I constantly teetered on the edge of injury from pushing myself too hard. And to this day, I'm still paying the price. Just last week, Erika and I, my wife, we were on vacation in Roatán, which is a beautiful island off the coast of Honduras, and I could only swim 10 minutes as my right shoulder throbbed in pain from the overuse I sustained many, many years ago. Now, well, I wish I had access to the fantastic research that will be described by our guests today. Probably would have saved my shoulder. 

Today, we have the pleasure of talking with UVA's 2024 Edlich and Henderson Innovators of the year. Professors Silvia Blemker and Craig Meyer from UVA's Department of Biomedical Engineering have done groundbreaking research that helps athletes and people with aches and pains, which is all of us. Their award-winning research provides efficient analysis of MRI data to generate 3D visualizations of an individual's musculature, ultimately aimed at maximizing and maintaining recovery. That's the holy grail for both amateurs, knuckleheads, like me, and elite athletes. Craig, Silvia, congratulations for the award and welcome to Hoos in STEM. 

(00:02:44) Silvia Blemker: Thanks for having us. 

(00:02:45) Craig Meyer: Yeah, thank you very much. 

(00:02:46) Ono: So, Craig, Silvia, you are UVA's Innovators of the Year. This is a big deal. Our previous Innovators have done amazing work. And I don't want to list them all, but I'd like to remind our listeners of one such work, for those who are struggling with diabetes or have family and friends struggling with diabetes. Today, thanks to research done here at the University of Virginia done by Marc Breton and Boris Kovatchev, previous Innovators of the Year, there is a device called the Dexcom device, which is a glucose monitor combined with an insulin pump. There are half a million Americans that use these devices to manage their diabetes. That is work very much like your work, which will improve the lives of all of us. So, I'm delighted that we get to talk about your advances, but I do want to say that the Innovator of the Year, this is a very big award for the University of Virginia. So, as Innovators of the Year, let's begin with the basics. You started a company. Tell us about it. 

(00:03:52) Blemker: Yeah, we started a company called Springbok Analytics that originated from research that we started at the University of Virginia. So, all of it really stems from research done by my colleague here, Craig Meyer. We have another colleague, Joe Hart, who was a professor in Kinesiology in the School of Education at the time. I think that was circa 2009-ish, when we first had the idea for this company. 

(00:04:22) Ono: And what does the company do? What do you sell? What is your product? 

(00:04:25) Blemker: So, the product is a rapid assessment of muscle health all throughout the body at the individual muscle level. So, we can tell you how symmetric each of your muscles are, the composition of them, how healthy they are, and also how big they are relative to your characteristics, your height, your weight, your sex, your age. This provides people with this way to assess really their movement, mobility, and health. And we use this technology in a wide range of people, from elite athletes, we work with a number of NBA teams, all the way to people that have muscle disease. 

(00:05:07) Ono: Springbok, what does that mean? What's the origin of that terminology? 

(00:05:11) Blemker: We came up with the idea for... Springbok is the name because we were inspired by the athletes that we are working with and so we named it after springbok, the very fast-moving animal that jumps high, runs fast, and so, and it stuck. 

(00:05:29) Ono: So, MRIs have been around for quite some time. So, the innovation, Craig, could you break down for us or describe the innovation that makes all of this work possible? 

(00:05:39) Meyer: Well, standard MRI scans provide usually 2D cross-sectional images of muscles so that a doctor can assess any internal problems with the muscles, using their experience with what healthy muscles look like. At the heart of our research was to take a group of these 2D MRI slices right next to each other and then process these images by separating out the individual muscles in the image through a process that's called image segmentation, and then we could assemble a 3D view of each of the muscles. 

(00:06:15) Ono: And so, Craig, you assemble through segmentation, you said, a 3D image?

(00:06:19) Meyer: Well, so you take the 2D image and you draw circles around the individual muscles. Now, we don't draw them around ourselves, and we'll talk more about that later. We use AI to do that, initially we did it manually, and we separate out each of these individual muscles. So, then you have a bunch of 2D images with circles around them and then you can compile an individual muscle based on combining all those different circles from the different slices. The central innovation is that we then can calculate the volume of each muscle very precisely, and we can then produce a 3D color-coded image that shows the volume of each muscle normalized by the size of the subject because larger people have larger muscles and then compare the size of each muscle to a database of similar subjects. For example, we could compare the muscle volumes of a triathlete to a database of elite triathletes, if we had one. 

(00:07:18) Ono: And what do you look for if I'm an injured triathlete and I come to you, and you have this database? What would we be looking for? 

(00:07:26) Blemker: Well, we would be looking probably first of all, at the symmetry of your muscles. So, let's say you injured your right shoulder. The most natural thing is to compare right and left because you're kind of your own. So, we can see which muscles are more different between, because the thing is you have many muscles. You have the rotator cuff in your shoulder, which is deep and then kind of on the inside, and then you also have several muscles on the outside. So, the question is where might there be an imbalance, given all of those muscles, and it's really hard to tell that just by knowing that it hurts. So, we could tell you which muscles are most different. We could also do things like the composition of the muscle. So, especially in shoulder muscles, it's common once you have an injury, actually your muscles become atrophied, and they also start getting replaced by fat. It's called fat infiltration, and so we can measure that as well. So, we can give you an assessment of that in your muscle and that could help your physical therapist figure out what exercises you should do, specifically targeted for your muscles. It might also inform a surgeon to decide is it worthwhile to do some kind of surgery, depending on what we find with the musculature.

(00:08:51) Ono: So, this is really precision information down to the individual muscle level that could inform say surgeons, and for elite athletics, the coaches that might be designing drills, precision drills for a particular athlete for their sport. 

(00:09:06) Blemker: Exactly. 

(00:09:07) Ono: So, tell us about some of the sports. This is fascinating. 

(00:09:09) Blemker: So probably closest to triathlon, we've done some water polo players. We've also done cross country runners and otherwise, basketball, soccer. We work with a number of Premier League teams but then also teams in the US, both male and female. We actually, those two, back to Craig's point about if you're an athlete, we could actually not compare you to like the general population, but we could compare you to other athletes. We have what we call databases, or those comparisons, for male and female basketball and male and female soccer. So that's something that we offer in those areas, and actually we have a beta one for Australian Rules Football, which is an interesting sport, because we have a lot of partners in Australia. 

(00:10:01) Ono: Really? Okay. 

(00:10:02) Blemker: Yeah. It's almost like, if you look at their muscles, it's almost like in between the soccer and the basketball players, is that's what, they're called or we call them, AFL. And actually, we're right now developing a pitcher database. Because you know, in in terms of the MLB, if they want to use this to manage their pitchers in terms of their health and performance, they absolutely need a reference database because those pitchers are very different from the general population.

(00:10:31) Ono: They must be completely asymmetrical.

(00:10:34) Blemker: They are very... it's actually really cool. We're just getting the results. We just completed several scans actually, including some UVA pitchers and pitchers from a number of other schools. And so it's been actually really exciting to see which muscles are most different and you see, because it's such an asymmetric sport, but not all one side. You know, your legs and your arms are all kind of moving asymmetrically. You see back and forth imbalance in really interesting ways. 

(00:11:03) Ono: Well, that is fascinating. I was a huge fan of baseball and I like the book Moneyball, where the idea is to use data analytics to pick and choose your players and change strategy. So, you're doing the biological, biomechanical, kinesiological Moneyball. So, that's great. 

(00:11:25) Blemker: Well, one of the things we think is exciting about what we do in particular is for sure it could tell you something about an athlete's physique and what their potential is, but the information is provided in a way that they actually can do something about it because muscles are very adaptable. You just have to know which muscles need strengthening in particular. So, I think that's a cool way to have each person reach the athletic potential that they are after. 

(00:11:51) Ono: That is super. Craig, I think you were about to touch on this. Obviously, this product is revolutionary. All of these sports and Australian football. I'd like to do a little bit more of a deeper dive. You said that in your segmentation when you go from 2D image, MRI images, to assembling these 3D images. You mentioned something about now we use AI in this particular step. Can you talk about that? What kind of machine learning, what kind of AI, is in your current version of your Springbok?

(00:12:22) Meyer: How that started was when we started the company, Xue Feng, who had done his PhD in my lab and he had worked at a startup company in in Silicon Valley and he came back and he was, he is, employee number one and has been the Chief Technical Officer of Springbok all along but he's also a research assistant professor in BME, and so he still does some part-time work with UVA. At any rate, one of the things we did when we started the company... the first thing we did was apply for a National Science Foundation small business grant, an STTR grant, and we got that, and then Xue was employee number one. So, he hired a bunch of people to help, or you know, a number of people, to help with him as we got the grant, and what they did is they tried to do the segmentation. 

Segmentation of muscles is a really hard problem because the muscles are just right up against each other, and so you know, there's sometimes a little bit of fat there but basically, it's a very hard problem. So, he and his team tried all the all the state-of-the-art methods for doing this and none of them worked automatically. But this was right about when deep learning came on the scene, and especially when deep learning was starting to be used around 2015. And so then, he tried that, and Xue became a serious expert pretty quickly in deep learning and that worked. That worked and it's very fast and works pretty well, and we still check it. But so anyway, that, using deep learning, a form of what's now called AI, works very well if you know what you're doing in segmenting images. And so, we've used that in my lab for a bunch of different things, right. We've used it for cardiac segmentation and Xue is usually involved in that, and then we've used it for a variety of other things. We use deep learning to de-noise. One of our research topics is using low field MRI scanners, which are much cheaper, and therefore they're designed so that they can work in underresourced areas in the US and other countries. But we've used deep learning to remove that extra noise in low fields. That's one of the things we use it for. We use it for a lot of different things in MRI. 

(00:14:43) Ono: So as a mathematician, I'm aware of some of the advances in linear algebra which were thought to have been contradictory, the compressed sensing of Candès and Terry Tao. So, I assume that some of these ideas are implemented in your algorithm, is that right? 

(00:15:01) Meyer: Sure, yeah. In fact, one of our projects now is to do real time cardiac MRI, along with some collaborators. This is a Coulter-funded project and in particular, we're trying to reduce the amount of heart damage from treatment of breast cancer. The treatment, the medicines that are used to treat breast cancer, can also damage the heart and so we have a project working on trying to limit that as much as possible by imaging it. So the real time stuff, we use compressed sensing and combined with this spiral way of collecting the data, and we're trying to make that work in real time so that it can actually work in the UVA hospital. And so, we've got that working and that's a compressed sensing which takes some computation, but it happens in real time now. 

(00:15:53) Ono: Silvia, the work in your lab. So, are you also an MRI specialist or...? Tell us about your work. You came together somehow to form this company. 

(00:16:04) Blemker: We did, yeah. So I guess my lab, if you look at what we do, it's very interdisciplinary because we're interested in problems all related to muscle because that's kind of like centerpiece of what we care about. We're always drawing on new technologies or developing new methods to study muscle, to model it, to understand it, to manipulate it maybe. And so, I found myself crossing many fields. So I actually started using MRI as a tool to study muscle when I was a graduate student at Stanford University and actually used the scanner… we kind of crossed paths. Craig and I sort of crossed paths at that point because it's the way to study human muscle. Because if you really want to study humans, you've got to look at human muscles. Now that we have these really cool in vivo techniques, that's the way to do it. So, I also might call myself a muscle physiologist. I might call myself a mechanical engineer and actually my PhD is in mechanical engineering. So it's a wide range of things. 

(00:16:55) Ono: So, this company is largely about working with real athletes or people with musculature problems. So presumably in your early training, you had this dream that you want to develop therapeutics. 

(00:17:28) Blemker: Absolutely, yeah. I mean, you know, it's actually why I wanted to do biomedical engineering to begin with. I'm fascinated by biology. I wanted to do something that helped people in the medical field but I love problem solving, which is engineering. So, to begin with, I got interested in muscle biomechanics but then learned about some really important problems that are involved where muscle impairment is the problem. In particular, one of the early applications I was interested in was cerebral palsy and the movement disorders that are associated with cerebral palsy. 

(00:18:07) Ono: I would have thought that cerebral palsy was primarily a neurological disorder, so how does that translate to the musculature? 

(00:18:16) Blemker: Well, it's fascinating. So cerebral palsy is an injury that happens to the brain around, during or around, birth and it's sort of, it's almost like a stroke that's happens in the brain but in an infant. So that means as the infant is developing, there's abnormal patterns, coordination patterns, that are sent from the brain to your body which means that the signals muscles are getting are different than they would traditionally mean, which means they grow differently than they would have. Bones actually also grow differently as well, because they adapt to the forces that are applied to them, to the loads. Which means that as children are developing, are starting to walk and move, they are hindered by the abnormal growth and adaptation of their muscles and their bones. 

And so, in terms of how these things are treated, helping them in particular not have their muscles restrain or constrain their movement, it's typical to do a number of different surgeries on their muscles to release them so that they're no longer restricting. And so that's where the interest in the technology actually came from, is that because movement is so complicated, it's really hard from the outside to figure out which muscles within the body of a say, a child of with cerebral palsy, is really the one that's causing the difficulty in their movement. The challenge is if you choose the wrong muscle, then you might be weakening a muscle that's already weak and you're not addressing the muscle that's actually the culprit. So that was the actually what drove the whole thing is a conversation with a surgeon who... we were talking about different ideas about muscle and developing models and using imaging and he said, well you know one of the biggest things for me is understanding which muscle to operate on. That work was with Mark Abel, who was a pediatric orthopedic surgeon here at UVA, practicing at the time.

(00:20:14) Ono: Despite the fact that you're both in BME, it's not every day that colleagues get together to form a company. I'm really interested so maybe we'll begin with you, Craig. What was the spark? How did the two of you together get together with Hart to recognize that you had the idea for what became Springbok? 

(00:20:32) Meyer: Well around 2009, Silvia came to me and said that we should apply for a Coulter program grant. Coulter program is a program at UVA BME funded by a foundation and basically, their goal is to take your technology and move it into clinical use, move it into practical use. And so, we said okay, so we'll apply for that grant. We fortunately got the grant and then the Coulter program continued to encourage us. You know, we were making very good progress, they were funding our research and trying this out. The idea worked. All those things were going well. We were starting to try it out on a variety of different people. And so, then we said, well, how are we going to translate this? And obviously, you can license your technology to a company but that's not easily done. So we were encouraged by the Coulter board to start our own company, and we talked, and we got advice on how to structure the company and those sorts of things. And so around 2014, we were in a coffee shop, the three of us counting Joe Hart, and we decided, well, let's go ahead and try to get a grant, a small business grant, that will help us get this company started. 

(00:21:50) Ono: This is the NSF grant. 

(00:21:52) Meyer: Yeah. And so, David Chen of Coulter had introduced us to a program officer at the National Science Foundation, the NSF, and he sort of encouraged us to go ahead and apply for a small business grant. So, we needed to start a company. You need to have a company in order to get a small business loan. So we sat around and we decided, okay, so we named the company, because you need a name too, and we decided in 2014 to start the company. So, we started the company, we got a lawyer, and we'd already at this point had some inventions and we started the company, did all the paperwork, which there's a lot of it, and then we got the grant. 

(00:22:29) Blemker: So the key thing that we had to do at our company was transform the technology into... something that was essentially a research tool into something that could be commercializable. And the biggest limiting step was how long it took to produce one of these, kind of effectively, reports of somebody's musculature. It took to process one scan... it took like 50 hours of time of user time, expert user time, and obviously that was not going to work. So that was the first big task, and that was the focus of our first grant to the National Science Foundation, was figuring out how to do that. The benefit we had was not only were we licensing the technology from LVG, we also licensed all the data that we generated at UVA, which formed the basis for training the AI that we ultimately developed through the leadership of Xue Feng. So that was really the first several years of Springbok, was figuring out the problem of the segmentation, and AI was the solution. So it was around 2019, when we filed the patent for that technology and published a paper that describes the AI for muscle segmentation, which by the way has actually now been implemented by many of my colleagues across the world. Now it is the way to do muscle segmentation. It’s all based on that paper. 

(00:23:57) Ono: What is it like starting a company that's a real for-profit company? Because as a scientist, it's probably something you never thought about. 

(00:24:06) Blemker: You know, honestly, it's not that different from being a PI. I mean, as a PI running a lab, you're constantly selling your research, right? You have to bring in money to fund your lab. It's selling but I mean it's very different, and the point of it is very different, but there's a lot of fundamental things that I think do cross over quite well. 

(00:24:26) Ono: Craig, same question. What is it like to go from being a lab scientist to partner in a for-profit company? 

(00:24:33) Meyer: Yeah, well, it was interesting. So, I was legally president of the company and so I was also the sort of the grant office of the of the company and so had to do a lot of things about you know, signing leases, and getting insurance, and all these things, and we moved around to various different different places. So I was a little bit used to that because I had had been at Stanford and we had looked into starting companies out there and so I had some exposure to the startup world but still, it's different when you have to pay everybody and you have to make sure that the you know... and the grants were great but sometimes I had to go around and talk to the founders and we all would loan the company some money just to make it through to the next bunch of money, right? And we would go to meetings, and we would talk to people, and try to find out who might be customers, and so all of that. 

(00:25:24) Ono: So, time is beginning to run short here. And with all of our guests, I'd like to hear what was your personal path, perhaps beginning in high school or elementary school or college, that ultimately led you to the University of Virginia as distinguished faculty in BME. 

(00:25:40) Meyer: Well, when I was in high school, I had a friend a year older than me who went to Stanford. And I decided, well you know, I don't want to spend the money to do it now, but I'll go to grad school at Stanford. Okay, so it wasn't a very direct path but I went to the university in Missouri. And then after I graduated, I considered various options, but I ended up going to be an engineer in San Diego out of college. And so, then I did go eventually to do what my plan was in in high school. So that was a great place to be and a great time, because it was just when commercial MRI was coming about. It was a great lab to work in. We did a lot of sort of fundamental MRI research there. And so, I did a lot because I was working on fast imaging, fast MRI. We did a lot of cardiac imaging using MRI. But I saw an ad for this type of job, biomedical engineering professorship, to do cardiac MRI. So then I decided to go visit and see. So I contacted them, and they invited me to come visit. And it turns out that it was a purposeful thing that three different departments had decided to do, and they wanted to hire people. And so, Chris Kramer is a world known cardiologist who does a lot of MRI research. And so he was one of the people had been hired before me, Fred Epstein, who is an excellent cardiac MR physicist. And so that, and the fact that I took my wife with me on that first visit to UVA. I wouldn't, you wouldn't normally do that on the first visit and that was because we were about to have our first kid, and I knew that she wouldn't be able to go to see this town across the country. But Fred Epstein's wife showed my wife around town and my wife liked the town. It's been great to be here, you know, really good people to work with, an excellent department in biomedical engineering and colleagues that you can start companies with, and all kinds of things. 

(00:27:45) Ono: Thank you. Thank you for sharing, Craig. Silvia, your story. 

(00:27:58) Blemker: Sure. I am the child of professors. My dad was a mathematician so science and math was like all part of it, but I always liked biology and anatomy. My brother would do model cars, you know, where you put together the car and you paint it and all the like you can get those. Well, my mom somehow found the anatomical version. So there was a heart, there was an ear, there was an eye, and then that was right when biomedical engineering was kind of emerging as a major in engineering schools. And so I learned about it, you know, randomly from one of those letters you get in the mail from universities. Now we get them all the time for my son. And so then, I started exploring that field and realized that would be a really cool fit for me, combining math and physics with biology. And so I went to get my undergrad in biomedical engineering from Northwestern University. They had one of the earlier programs. UVA actually had one of the early graduate programs too, in BME. I was always interested in human movement. I was a ballerina as a child, so I always thought human movement was really cool. So I found a passion in in biomechanics and muscle biomechanics early on. And then I got my master's degree in BME at Northwestern too but at the time, my advisor got a job at Stanford in mechanical engineering and he said, well would you like to come with me there to do your PhD, and I was like sure. So that was fantastic. 

It was really great to go to Stanford, and it really kind of opened my horizons and allowed me to really expand my skill set and understanding. So one of them was I took MRI classes from Craig's MRI advisor, who is like, you know, one of the world renowned experts in MRI. I had him in class and learned about MRI from him, which was phenomenal. I actually collaborated with the lab that Craig was in to do MRI imaging of muscle, which was really cool. It was just a really great place to learn all the interdisciplinary things that I did, computer science too. So yeah, actually I remember when, so this is around the same time, because Craig, when he had decided to go to UVA, I think you were doing some sort of last lecture or like a final presentation to the imaging community or to the MRSRL or something, must have been something like that because I attended, as I was a graduate student at the time. My friend Deanna was like, "Oh, there's this guy, Craig Meyer, he's giving this presentation about, you know, where he's going for his faculty job. I was like, "Oh, that sounds..." I was starting to think about it." So Craig showed us, showed pictures of the MRI machines that they were going to build here and talked about how what a cool place it was. So that kind of put it on the map for me. UVA's just such a great environment for doing the kind of work that we're doing. I think it's the scale. There's expertise in everything but it's not too big. Everybody gets to know each other and it's a very collegial place. So I really value that. 

(00:31:00) Ono: Well, great. For undergraduates or prospective graduate students who want to get involved, what message, what advice can you offer them? Craig? 

(00:31:08) Meyer: Well, you want to do your research about what the lab is doing, what my lab is doing, and you want to see if that interests you. The other thing is that, you know, they should talk to grad students whenever they can. So the first thing they should do is see if they could, you know, shadow a grad student in the labs that they're interested in. 

(00:31:28) Ono: Silvia, same question. 

(00:31:30) Yeah, I echo those things, definitely getting to know the grad students. I honestly think, for my lab, it works really well if a student just comes by the lab and introduces themselves to the students who are there. I also encourage undergraduate students to get to know their teaching assistants because they're all PhD students for example, in biomedical engineering, and they all work in labs and they're phenomenal people. They are good at being TAs but they're also really smart, and get advice from them. They've just been through it, you know, they were just start deciding to go to graduate school even.

(00:32:05) Ono: Big brother, big sister. 

(00:32:06) Blemker: Yeah, exactly, and they love to mentor. We have a really fantastic environment of mentorship between graduate students and undergraduates. So I think they're a huge, almost untapped resource, I think, for undergrads especially and even if it's not to get involved in research honestly, it's like thinking about next steps or how to succeed in college, all those types of things. 

(00:32:31) Ono: Well, Craig, Silvia, thank you very much. Congratulations again, Innovators of the Year. This is a big deal. It's been a pleasure having you on the show. You're going above and beyond what Jim Ryan, our president, expects for the university. The research is great, translating it into therapeutics, improving elite athletes to everyday people with aches and pains. That's wonderful. I love the benevolence. Thank you for being great and good. I'm Ken Ono, STEM Advisor to the Provost and the Marvin Rosenblum Professor of Mathematics, and you've been listening to Hoos in STEM. 

• Innovators of the Year’s Technology ‘Springs’ Athletes to Better Health (UVA Engineering News, February 10, 2025)
• Springbok Analytics
• Wallace H. Coulter Center for Translational Research