“Rapid quantitative imaging of high intensity ultrasonic pressure fields”
The Journal of the Acoustical Society of America 148, 660 (2020); https://doi.org/10.1121/10.0001689
Authors: Huiwen Luo, Jiro Kusunose, Gianmarco Pinton, Charles F. Caskey, and William A. Grissom.
In this episode, we speak with authors from Vanderbilt University Institute of Imaging Science, Huiwen Luo, Research Assistant, William Grissom, Associate Professor of Biomedical Imaging, and Charles Caskey, Associate Professor of Radiology & Radiological Sciences. We will discuss their research on rapid quantitative imaging of high intensity ultrasonic pressure fields, and the rapid projection imaging method they have created for mapping ultrasonic pressure fields. We will examine their conception of this method, and the future applications of this work.
Acoustical Society of America Publications Page
Music Credit: Min 2019 by minwbu from Pixabay. https://pixabay.com/?utm_source=link-attribution&utm_medium=referral&utm_campaign=music&utm_content=1022
Hello, and welcome to Across Acoustics, the official podcast of the Acoustical Society of America's publications office. This is the inaugural podcast where we will highlight author's research from our four publications. The Journal of Acoustical Society of America, also known as JASA. JASA Express Letters, Proceedings of Meetings on Acoustics, also known as Poma and Acoustics. Today, I am your host Malene' Walters publications business manager ASA. Today we will be speaking with co-authors of "Rapid Quantitative Imaging of High-Intensity Ultrasonic Pressure Fields, published in the August 2020 issue of the Journal of the Acoustical Society of America. Our three authors are all from Vanderbilt University Institute of Imaging Science and include Huiwen Luo Research Assistant William Grissom, Associate Professor of Biomedical Imaging, and Charles Caskey, Associate Professor of Radiology, and Radiological Science. Hello to everyone, and welcome to our first podcast.
How's everyone doing today?
Um, now can each of you tell us a little bit about your background?
Can we start with Huiwen?
Hello, everyone, I'm Huiwen Luo. I'm glad to talk with you here. Currently, I'm a fourth-year Ph.D. candidate in biomedical engineering, and I'm working with Dr. Will Grissom. So my research is mainly on visualizations of focused ultrasound by optical and MRI techniques.
Yes, my name is Will Grissom. I'm currently an associate professor in biomedical engineering at Vanderbilt here. I've been here about I think this is my 10th year. And before that, I worked at GE for a couple of years in their in their Global Research Division on MRI and and MRI-guided focused ultrasound. Before that I was a postdoc at Stanford within focused ultrasound. And before that, I was a grad student at the University of Michigan, working in functional MRI mostly. And before that, I was an undergrad at the University of Michigan in electrical engineering. So started in electrical and went to biomedical, Stanford, GE and then have been at Vanderbilt for many years now, and have worked with these three fine folks for quite these other fine folks here for quite a while now.
Hi, yeah, Charles Caskey. I'm an associate professor now at Vanderbilt University Institute of emerging science, in radiology and Radiological sciences. I have been there about going on eight years now, having having worked there at Vanderbilt, and came there by way of Dr. Catherine Ferraris lab, who was working at that time at UC Davis, which is where I did some of my training at both the postdoctoral and graduate student level and working in the field of ultrasound.
Great. Now let's dive into your research, rapid quantitative imaging of high-intensity ultrasonic pressure fields. Charles, could you please describe what high-intensity ultrasonic pressure fields are?
Sure. So this is the process of focusing sound that is in that is above the hearing range. So that's why it's ultra down. And it's the process of focusing it to a small point in the same way that a magnifying glass focuses light. You can create acoustic apertures that do this with sound. And so generally the process is you will be emitting source that's kind of set shaped like a spherical cap, and emits down to a small point that's about the size of a grain of rice. And so you have this small focal energy. And that's generally so what that's being explored for frequently is trying to use that energy in a medical scenario to try to either burn like ablate thermally specific areas that you may otherwise want to surgically remove. So that's, that's kind of the field of study there. And so there's a lot of activity in the medical world now try to use this to create incision-less surgeries and things.
Yeah, so you can you can so you can focus it to this point without affecting any of the intervening tissue, which is really what makes it non-invasive,
That you can focus to a point deep in the tissue without affecting intervening tissue.
Ah, I see. Okay. And you did say medical so, is there any other types of treatments or applications that utilize high-intensity focused ultrasound, or is it basically for these medical procedures.
I think it's most widely and it's most widely used in medical procedures there. I mean, there certainly are sound focusing applications outside of this bar and things like that. But yeah, the I do think that the main Yeah, the main use here would be in his medical and so the main applications is through the, that has really received the most FDA approval or in the skull, were really creating an incision in craniotomy is a huge invasive task. And so the process there is to focus the sound through the skull and into the, in this case, I think the main FDA approved application is ablation of the thalamus for essential tremor.
And so yeah, so it's sort of like a neurosurgeon being explored as a neurosurgery that doesn't require incision. And it's also approved for ablation of uterine fibroids as a treatment for fibroids. So it's a deployed technology and a lot of excitement surrounding it, now,
It sounds like it. Will, could you please explain the mapping of high-intensity focused ultrasound beams?
Yeah, so if we're going to be putting this energy into the body, at some point, where we can't necessarily see it immediately, we need to get know about, we need to know how much power we're putting into the body. And we also need to know where it's going. So there's so a very common task when you're developing or maintaining these therapeutic ultrasound systems is is mapping the beam to make sure that the amount of acoustic power you're putting into the body matches what you expect it to be. And also that the location and distribution of it spatially in the tissue matches what you want it to be. So regular beam mapping is is performed in the development and maintenance of therapeutic systems. When developing these new techniques and hardware both both on the imaging side, since for imaging, you need to calibrate where exactly your signal is coming from and that your transducer is performing as you expect it to be. And also in therapeutic ultrasound, where it actually tells you exactly it tells you know exactly where the tissue is and how much energy you're putting you're putting to it, to manipulate it. And instead, it's definitely desirable to be able to regularly map acoustic beams for regular quality assurance in the clinic to make sure that the work that there are no new aberrations in the beam or any failures of the devices that you might not otherwise see. To make sure overall that that that patient safety is is is maintained, right, and that proper functioning of the system is maintained.
Okay. And can you describe the old methods of mapping focused ultrasound pressure beams and why they are inadequate?
Yeah, so so hydrophones our devices, these are these are the really the gold standard method for mapping ultrasound beams. And what they are is essentially a single, usually piezo transducer, it'll be a piezoelectric element with a membrane on it attached to the end of a needle that is moved around a water tank, sometimes using a motorized or robotic kind of translation stage. There, you'll have your ultrasound transducer in the water tank, and then you'll have this needle also in the water tank that sort of facing the transducer, and you move the needle hydrophone around to map out the beam. And what the needle gives you is then time traces of the pressure that the needle is sensing at its endpoint. And you record all those measurements. And you can then say, Okay, I'll take the maximum value or the sort of average value of those pressures that I'm getting. And I'll make an image of that. And so that was often what we refer to as like the gold standard being mapping method. And it can be highly accurate and precise, but it's not very convenient or portable. So that is you have to have it in a water tank, you have to also have the transducer and water thing, which means removing the transducer from whatever clinical system it's set up into, which is not usually a trivial task. It's not very portable, you can't you know, just easily take a water tank and set it up precisely at whatever site and get it set and get it running very quickly.And the other way to think about it is that is that it's it's great at giving you lots of temporal information because if you set it at a point, you just start recording, you get a long time trace of what the ultra of what the pressure field is that you're receiving at the tip of that hydrophone. But then you have to mechanically move it to every other point. And so in for a lot of these quality assurance tasks and a lot of ultrasound experiments, we don't necessarily care so much about being able to efficiently get a lot of time information. We want more spatial distribution and a lot of these applications. And so so it's not very well suited for that task of rapidly getting spatial information because he had to mechanically steer it around. At the same time, they're also they're also sensitive to high pressures. It's easy to burn one out by cavitation or damage the end of, by cavitation, when you use actual therapeutic ultrasound pressure levels, so you often have to do the actual measurements at lower pressures and then assume that everything just scales up linearly, which isn't necessarily guaranteed that that will be the case when you go up to the, to the high pressures.
So we were interested in developing a method that kind of had the opposite strength that it would be kind of crummy giving you temporal information, but very good at giving you spatial information, and very rapid at that and that it should be very easy to set up. There are also other methods out there in the literature for characterizing ultrasound beams that are less used than Well, some of them are less used than than the hydrophone such as radiate radiation force balances are, are essentially these a mechanical devices that you turn on the ultrasound and you actually literally push on a lever and you measure like what displacement that gives you. And then that's like can be a measure of the of the strength of the ultrasound beam. It doesn't give you any spatial information, but it does give you quantitative information about how much power you're putting out from from the transducer. And then there and then there have been other art method falls into the category of optical beam mapping methods. And there I have is a long history of optical beam mapping methods based on you know, photographing beams or methods that are so called laser schlieren type methods, which are optics methods that are that are a bit more complicated than what then what we're doing here.
Okay, Huiwen? How is your rapid projection imaging method different from what is currently being used?
So, as Will mentioned, our methods falls under the category of optical beam mapping methods, which has been used around for many years, different from the conventional instruments. What we're doing here is a stripped-down version of schlieren. That doesn't, does not require fancy optical setups, but still years rapid and valuable info information about ultrasound in so this method can be implemented in a small and portable package with no moving parts to rapidly rapidly map force beams. So there are no parts to experience. We're from the first beam. So our method is endurable finally. Yeah.
Will, how did you conceive this idea?
Yeah, so Charles and I were at a meeting of the Society for therapeutic ultrasound, I think like way back in maybe 2014. Even it was in Las Vegas. Yeah. And we were at the meeting, we were watching a talk by Gail Tahar about ultrasound dosimetry, or dosimetry for therapeutic ultrasound. And one of the slides that she flashed up on during her talk was a slide showing a technique that she talked about called xerography. Mean, referencing zebra stripes. And what she was showing was that you could take a ultrasound transducer and put it in a tank of water. And if you had, if that tank was clear, such that you could look through it, if you turned on the beam, because that beam would change the index of refraction in the water, because the index of refraction is, is proportional to pressure, because it would change the index of refraction in the water, you could take a piece of paper with black and white stripes on it, put it on the other side of the tank. And when you viewed it from the opposite side through the ultrasound beam, you would see that the stripes would blur in a pattern that looks like an ultrasound beam. And so she called the xerography because of the black and white stripes. And it's kind of she sort of referred to it as like a sort of a party trick and the ultrasound community for that people have known about for many years. And then later on this got me and Charles thinking as we're walking around Vegas, thinking, you know, is there a way that we could that we could build on that to to make it quantitative, because it seemed that nobody had made the method quantitative before really, really rigorously studied the the physics of it.
Yeah, and so
now that kind of conception, and then it sort of like, yeah, that it eventually led to us. Yeah, writing kind of fleshing it out more or writing a grant. And then it kind of came came to fruition that that way, just kind of really wanting to look at the signal that we knew was there. But then like thinking like, oh, how do you actually measure this and then and, and built it up around that? Yeah, we got a couple. We started by getting a couple of undergrad biomedical engineering undergrads working on it had a couple of kind of cycling through, just so we could get just enough little bit of like proof of principle data that to show that it could work. We also had like 3d maps, we had at one point, we had a student making 3d bea maps. They weren't quantitative, that he would just sort of back project take take photographs, as he was moving into the tank around the beam and taking photographs, or maybe we just assumed axial, I think we just assumed rotational symmetry actually. And then would take photographs and was able to reconstruct a kind of a cool-looking 3d beam map from that, and eventually, we kind of had enough momentum there where we decided we could write a grant on it. And so we wrote a grant for an R 21. This is sort of the high-risk, high reward track of grants at the NIH. And when we got that, I think it started maybe in 2016, maybe it's when we finally had the grant finally had some money to work on it. And then Huiwen joining, really, she really got it all together, once we, you know, really got it working properly.
And with that support,
yeah, well, Huiwen really took it home to like, figure out how to do it physically in the tank, and how to rigorously connect the theory together, you know, and how to do the experiments in a rigorous enough way that the theory that we could actually, you know, build our reconstructions based on the theory and then and then expect that our experimental data would work out with all of that.
Oh, very nice. Well, congratulations on the grant and then finding the right combination, people make all the difference.
Huiwen, what is the background-oriented schlieren imaging? And what is the CW version that you have come up with? I know you mentioned that as we were speaking, could you explain a little more?
Yeah, sure. So, as a traditional background or engineer schlieren is is a force of flow the virtualization qualitatively. So this technique, you use camera to image a background pattern through a non-unique non-uniform refractive index field. So original, background-oriented schlieren, imaging assumes non-uniform refractive index field is constant anesthetic. So, however, in BLS imaging have focused ultrasound, the refractive index field actually changes dynamically during the propagation of first waves. So we assumed the steady steady state ultrasonic waves and simplifies the hardware setup by allowing the beam to run continuously during the acquisition. So this is so CW actually means that continuous-wave, yes, yeah,
So so and it works out that that what we what we end up getting in these photographs that we take of the blurred beam is like a histogram of the history of where the little dots that were imaging in the background have been shifted to. So So, you know, in typical schlieren imaging, it's actually used a lot in, I think the most of the literature on this topic of background oriented schlieren is in is in like aerodynamics research and fluid dynamics research, where you have water or air moving maybe through a wind tunnel at with different distributions of pressure through that wind tunnel, which slightly changes the index of refraction as you're looking through it. And then if you take sort of a random dot pattern, or maybe it's a structure dot pattern, and put it on the other side of the tunnel, and look at it, you get the same effect, as in this the biography technique where, where everything blurs around. But but in those situations, you can usually assume that that the flow of the field is is not changing over time. Whereas here we have the ultrasound beam is constantly oscillating at this megahertz level frequency, but we're trying to take a picture of it with a conventional consumer-grade camera that can't nearly freeze time, at the at the fine resolution that you would need to to really capture an instantaneous point. So so that's why we when we end up calling it continuous wave is because we're actually getting many, many, many oscillations of the ultrasound within one photo. And so what we end up with is truly like a blurring pattern. So this schlieren, schlieren is German for streak in and that's why it's called this schlieren technique. And so that's why we that's why we call it CW is because we end up like looking watching the shift of the beam over time, and we can't actually resolve exactly where it's been at each, you know, microsecond level time point in the in the, in the ultrasound, you know, at the ultrasound timescale, but we can see this overall pattern and again, we get end up with a histogram, the photograph is actually a histogram of where the ultra of where the ultrasound beam has shifted the background pattern at each location of the background pattern,
Okay, and now, machine learning is all around us. And in this project, use the deep neural network, what did that do for you?
So,so when the ultrasound beam is fixed, so the back blurring patterns on the background pattern actually corresponding to the project, the pressure amplitude at each location. So we already have derived for the modal have to relate the project to the pressure pressure waves to the blurring patterns. But it's so hard to get the project the pressure from the blurring patterns. So due to the, so the using a deep neural network just come to my mind. And so in this work, we use the deep neural network to solve the difficult inverse problem of reconstructing the project the pressure amplitudes, from the blurring patterns in the photographs. That's why I try I decided to use a deep neural network. And finally, it actually worked.
Yeah, this is Huiwen's big innovation in the project, right? She This is completely her, something she brought to the project that really made it work in the end. And and and I think this is kind of a cool application of deep learning. Because in this application, we can easily if we have an ultrasound beam, and we have a background pattern, we can easily run the equations forward to calculate what the photos photograph is going to look like that we're going to capture with the camera. But going in the reverse direction is really hard. Mathematically, it's highly nonlinear. And and that's what Huiwen realized that that's kind of an ideal application for machine learning. And Huiwen realized that when she when she took this on, is that is that calculating those forward images that we can use, for example, for training, the neural network is easy. But going the reverse direction is really hard back to the information that we want to extract from the photographs. That's the hard part. And and we don't know of any other way to do it other than using machine learning. And that's where that's where she had that idea. And that's what really made it fundamentally possible to get these quantitative maps out of the blurring patterns that we photograph.
And Huiwen, how did you make sure that this particular method worked?
So as we mentioned earlier as hydrophones widely used, and are the gold standard in the field, so, so people believe the results of hydrophones. So that's what so we compare our results, always measurements of hydrophones to evaluate the feasibility and accuracy of our methods. Yeah.
Yeah, lots of careful hydrophones. We use optical hydrophones which can tolerate really high pressures. But they have a trade off and sensitivity. And they're also way more expensive than your phones.
And they still
I think we had to have the ones we used here serviced a couple times during the course of this project. Yeah. So you know, in one of the cool things about this optical beam mapping method is that there are no moving parts, right, there's really nothing to break unless you get the camera wet or something, I guess. It's, there's nothing to break, really,
There's not an object in the beam since the standpoint of things not breaking. And in terms of it being an invasive measurement, you know, having an object in the beam that you're trying to measure is not usually desirable. So that's, uh, you know, it can interfere. So it's nice with this method that it's fully optical like that. So
Very nice. And now, how can others reproduce this work? And what applications do you envision it for?
yeah, so So actually, how I described the details of the protocol of this work. And also, we shared our code for the control software, image acquisition reconstructions on GitHub. I believe the GitHub link is in our paper, so everyone can reproduce this work if they have the required equipment, and they can use our code to do the opcode matching. Yes, yeah.
Yeah. So we try it. Yeah, we tried to be real open with this with this work. I mean, it's in you know, and it's built from fairly, in terms of in terms of acoustic measurement, very affordable tools in the sense that it uses a, an iPad or a small screen, and then a camera. So it's all pretty, you know, I think it's pretty accessible in that respect. And so, you know, in terms of, you know, I think you also asked about, like how it can be maybe used and with that, you know, maybe they could find yoga find a role, certainly not as a full on replacement for the great acoustic measurement things that we have hydrophones and other things, but certainly to supplement that as a, maybe a routine quality assessment tool or something like that might be an area where this technology could find use you kind of have, because it's it's low cost and accessible. May, you know you could envision it being some that was performed maybe more routinely than some of the calibration methods that are currently available that require a bit more expensive hardware, maybe specialized skills, things like that. So
Yeah, one of the things that we're really interested in is, is mapping beams through the skull. And there, we're interested in more in sort of continuous-wave measurements as this technique gives you. And we also, you know, have a lot of different sculpt pieces we'd like to put in front of it. And and measure fairly rapidly. So it's kind of well suited to experiments like that, where you want to, you want a spatial map, but don't care about so much the temporal dimension, and you'd like to be able to rapidly change some experimental variable and repeat your measurements again and again, that this, this technique would allow you to do that, you know, changing like multiple one or multiple variables and an experiment, and quickly getting spatial beam information from that. I don't know of any other method that could do that.
In like high through put measurement scenarios. Yeah.
And is there anything else about your research that you would like our listeners and readers to to know or understand?
Well maybe Charles does a little bit about what what stuff we're working on now. Because Huiwen has kind of moved on from this project, and it has to do sort of with a big project that Charles is leading?
Yeah. So yeah, I think, you know, we're, we're super enthusiastic about the seeing what's happening with transcranial ultrasound, clinically current it currently, and we're really focused on trying to see how we can develop next generation of those tools, as you know, really, as a, as a preclinical, primarily research lab, we're, we're looking a lot it at neuromodulation with ultrasound, so the use of these, this focused acoustic acoustic beam at lower intensity, so not intensities, that will cause thermal ablation. But, you know, so, intensities that have very little thermal energy, but can affect neurons. So using this as a tool to either investigate the brain by by stimulating small focal regions, and seeing how they, how it interacts with, with the rest of the brain, or you know, if one node of the of a circuit in the brain is, is stimulated, and we do this within the MRI, which is also where a lot of the therapies are done. So in that case, you have the ability to really guide the beam as well as the MRI as a tool to see how the brain is functioning. So you sort of have this, this suddenly, now you have an ultrasound as a tool to probe brain activity, as well as image the outcomes of that probe probing and brain activity. And there's just been a lot of interesting hardware development, imaging, image science development associated with that. And so it's kind of a direction that we, we kind of are heading with this, a broader broader field and in strictly the measurement part, which of course, is the crucial component, because you want to know, you know, how well are you focusing through this goal and these types of things?
Yeah. Yeah. Last year, we got one of the awards. It was it last year, it started the the heel grant the heel mechanism, which aims to replace opioids with device treatment novel device treatments. So we have a big grant with Charles Lima, an investigator named Lehman Chen, and I have this grant, we're building up a human system that we're first evaluating preclinically to treat different pain regions in the brain with ultrasound. And at the same time as retreating with the ultrasound we're building the MRI capabilities to be able to do functional brain imaging to see what effects the the ultrasound is having on the networks of the brain. And so before Clayman joined my lab, she used to work in MRI and I was able to convince her when she joined to work on this crazy project. And now now, now she's back on working on sort of MRI technical developments. For this new project. Specifically, she's looking at looking using MRI pulse sequences to image the tissue displacement due to the ultrasound so we can see where in the brain the ultrasound is, is targeted and and and how strong that is, and whether it's distorted by the skull. These are all sort of critical measures that we need to know if we're going to be doing these these procedures.
I think our listeners have learned a lot about about your research. And I would like to thank you all for participating in our podcasts. It's been a pleasure chatting with you and has a layman myself learning about high-intensity ultrasonic pressure fields. This was definitely a learning experience for me. So I'd like to thank you very much for joining
Thank you for tuning in to Across Acoustics. If you would like to hear more interviews from our authors about their research, please subscribe and find us on your preferred podcast platform.
Transcribed by https://otter.ai