Good afternoon, everyone. Thank you so much for inviting me to speak on my most favorite topic ever. So I am very excited to talk to you about the clinical applications for ultrasound contrast. So in this talk, I will give a brief introduction about ultrasound contrast, show you how contrast-enhanced ultrasound works, and then my favorite part is to kind of give you a tour through the body on how we can use contrast. So contrast-enhanced ultrasound is not new. As you know, it's been around for decades, primarily in Europe, Canada, and Asia. And in the United States, it was limited to cardiac indications until the FDA approved the use of Lumison for focal liver lesion evaluation in 2016. So when that happened, I was at Einstein Medical Center in Philadelphia, and I just latched on to the idea of starting a contrast-enhanced ultrasound program at my institution. So there are three major blood pool agents on the market, as listed here. And the basic premise is that a patient will undergo ultrasound during simultaneous intravenous injection of the contrast agent. And so here, this is a little depiction of a microbubble. So the contrast agent is this gas-filled microbubble that's less than the size of a red blood cell. And when it's under the ultrasound beam, it'll expand and contract like this. And then here, I have a transducer and a capillary bed. These microbubbles will stay within the intravascular system, so they will not diffuse out. And they will last for about five minutes. After five minutes, the microbubble will burst. It was stabilized by a protein or lipid shell. You breathe out the microbubble through the lungs, and the liver will metabolize the protein shell. And then here, this is just to show you the difference between the contrast agents that we use on ultrasound, which are the microbubbles, versus the contrast agents that are used on CT and MRI. So here, the microbubbles, you can see, stay within the capillary bed. And then below, the iodinated contrast or gadolinium are nanoparticles, and they diffuse out into the interstitium. And then they come back into the capillary bed, and then they're excreted by the kidneys and can be metabolized by the liver. Just so you know, my timer is not on. I'm not sure how to turn that on. Anyway. Oh, maybe it's... Sorry. It's okay. So how to perform a contrast-enhanced ultrasound. Oops. Let me go back one. So we need to have intravenous access. So as long as you have intravenous access, that is one major part. Another very important part is the area of interest needs to be visible on ultrasound, and it needs to be visible reliably. It can't be in a part of the body that's very deep that can only be seen if the patient is doing a handstand. A patient has to be comfortable for a sustained period of time. And then you have to have the ultrasound contrast software on your machine. And we use pulse inversion imaging in order to do the contrast-enhanced ultrasound where the signal from the soft tissue is canceled out, and only the micro-bubbles are visualized. And then here, this is just an image from a dual screen, a dual mode where you can see the contrast micro-bubbles on the left-hand side showing enhancement of a liver lesion. And then here is a low mechanical index ultrasound image that is kind of used for guidance. And then this is the setup that we have on a little platform next to where you'll be scanning. You need to have the saline flushes, your ultrasound contrast agent, alcohol swabs and IV, et cetera. And then here, we want to have the IV access very accessible, and you want to have a three-way stopcock hooked up to a saline flush. And then you want to have the contrast syringe hooked up parallel to the IV tubing. So when I was at Einstein, we started out by doing contrast-enhanced ultrasound of the liver lesions, and then we quickly expanded it to multiple different parts of the body. And this is a little table that we made that talks about liver, renal urinary tract. We ended up doing it on the testes, on the bowel. And then, basically, we said any organ or lesion visible on ultrasound with inconclusive grayscale ultrasound CT or MRI findings is a good candidate for contrast-enhanced ultrasound. There really is no limit with which you should be able to apply it to whatever you need to. So now, we're going to delve into some real-life examples. And the first topic that I will cover is the liver imaging. So this was a patient who had a MRI for workup of this rather ominous-looking mass that's in the liver. The patient was not able to tolerate the MRI, and so was then referred to our department for a contrast-enhanced ultrasound. So the images are showing an echogenic liver and then a rather mixed echogenicity mass that has this irregular hypoechoic margins. And so on the contrast-enhanced ultrasound, we can see that there is discontinuous peripheral nodular enhancement. And we followed this out for up to five minutes, and it had sustained enhancement. And this was a benign hemangioma. This is a patient who was undergoing HCC surveillance with MRI. So this patient had cirrhosis, was actually very claustrophobic, and was unable to lay still within the magnet. And the images were fraught with motion artifact. We saw that there was an enhancing area in the inferior tip of the right lobe of the liver, but we were not sure if it was definite for HCC. So we gave it a CT-MRI-Lyrads 4, which is probable HCC. So the multidisciplinary team suggested that the patient should undergo a contrast-enhanced ultrasound. So here, we were able to see that abnormal area exquisitely. You can see that the calipers are demarcating a mixed echogenicity mass in the inferior right lobe of the liver. So here's the contrast-enhanced ultrasound. And you can see this patient is breathing during the time of the examination, but because we have it in the sagittal plane, we're able to keep it within the field of view. And if you're watching this, I'm sure you're watching the enhancement pattern, you can see that there is a very hypo-enhancing mass, but within it, there are two nodules that show arterial phase hyper-enhancement. And this is that nodule-in-nodule phenomenon, and this is that natural progression into HCC. And we could see that the more superior nodule shows delayed washout, and so we were able to characterize and upgrade this mass to a contrast-enhanced ultrasound Lyrads V, definitive for HCC. This is a patient who has a history of gastroesophageal adenocarcinoma and was undergoing surveillance with non-contrast CT because the patient had chronic kidney disease, and he did not want to undergo an MRI. So on his CT scan, we saw a new hypo-dense mass along the posterior aspect of the right lobe of the liver. On the grayscale ultrasound, we were able to target and find that hypo-echoic mass, and so we were asked to do a contrast-enhanced ultrasound. So here, this was the area that we targeted. And on the contrast-enhanced ultrasound, we could see that there is this brisk arterial phase hyper-enhancement. And you see that there's already starting to be washout pretty immediately, much less than one minute after the injection. So you can see at 30 seconds, we're already starting to get washout. At 3 minutes, 37 seconds, it was a black hole. This is the classic appearance for malignancy, particularly metastases in this particular patient. And then what we did was we did large field-of-view contrast-enhanced ultrasound, and we were able to scan the rest of the liver. And if you look carefully, you'll see that there are other smaller metastases throughout the liver that we were not able to catch on the CT scan or on the grayscale routine ultrasound. Moving on to renal bladder and scrotum. Here this was a patient who had an indeterminate renal mass on CT. So this was a CT renal mass protocol, and these are two coronal images of the right kidney. And the radiologist who read this saw that there was a mass in the interpolar right kidney. But you can see the Hounsfield units was approximately 20 when comparing the unenhanced to the enhanced. So it was not definitive for an enhancing renal mass. So he recommended a contrast-enhanced ultrasound to problem solve. So we were able to see the hypodense mass that showed up as a hypoechoic solid mass on the renal ultrasound, and we figured that the sagittal plane would be optimal for imaging this patient so that he could breathe during the examination. So here's the contrast-enhanced ultrasound showing you that there is diffuse hypoenhancement of that mass. And this pattern of enhancement is fairly classic for a papillary renal cell carcinoma, which is what we had suggested, and that is what it turned out to be on pathology. Okay, now this was a case I was working with a resident. We were reading a CT cystogram on this 71-year-old patient who had chronic kidney disease and a bladder diverticulum. So here is the CT cystogram. I know it's kind of an unusual-looking urinary bladder. So we were pulling the jacket, and we looked back, and we saw that the scan from 2011 showed a small diverticulum along the left side of the bladder. So on the pre-cystogram images, you can see that there is this kind of hyper-dense area within the expanded diverticulum. And you can see that the contrast in the urinary bladder did not go into this little neck or into this diverticulum. We did, so I was reading this case, and I thought, look at how superficial the urinary bladder is. This patient would be an ideal candidate for a contrast-enhanced ultrasound. And so we brought the patient to the department, and we saw the bladder very well. We saw the neck of the diverticulum. It was filled with internal echoes, as was this diverticulum. And I thought to myself, okay, I know what's going on. The diverticulum is going to be filled with enhancing tumor, and this is all going to be blood clot and proteinaceous material. So we did the contrast-enhanced ultrasound, and we saw large areas of frond-like enhancing tumor along the wall of the diverticulum. And then the neck did not show any enhancement whatsoever. The neck was just proteinaceous gunk blocking the diverticulum, preventing it from decompressing. And so this patient then underwent a cytology evaluation and was diagnosed with squamous cell carcinoma. So that was a very, very interesting case. So this was a patient who had a renal infarct and had a rise in creatinine. And the ICU team had called me and said, Dr. Rogers, Dr. Rogers, like, we don't know what to do. This patient's not improving, and we don't want to give iodinated contrast to see what the status of the kidney is. Because as you can see here on this coronal contrast-enhanced CT, we have thrombus within the right renal artery, and we can see that the majority of the right kidney has infarcted with very small areas of enhancement. And they did not want to transport the patient to MRI, and a routine ultrasound without contrast was not really able to give any information as to the status of this patient's kidney. So I said, of course, well, why don't you bring the patient down, and we will do a contrast-enhanced ultrasound to see how much of the kidney is viable. And then lo and behold, we saw that actually a large portion of that kidney had regained perfusion, and less than 50% of the kidney had infarcted. So then the medical team knew that they could just wait and provide conservative supportive therapy, and that hopefully his kidney function will improve. I also want to talk a little bit about the intracavitary use of ultrasound contrast. So with this, we can dilute the ultrasound contrast agent with saline, and depending on your ultrasound equipment and settings, you can figure out the appropriate dilution. So this is a case that was given to me by one of my former residents who was doing a rotation at a children's hospital. So this is a patient who had a 22-month-old female who had bilateral vesicoureteral reflux. And on fluoroscopy, she had a right grade 4 reflux, which is one of the most severe types of reflux. And so one of the other FDA-approved indications for using Lumison and contrast-enhanced ultrasound is intracavitary injection to assess for vesicoureteral reflux. So here we have a grayscale image of the right kidney. And on the contrast-enhanced ultrasound, so this is done similar to a cystogram. The patient's bladder is catheterized, and there is intravesical injection of the dilute contrast agent. And we image the kidney area to see if there is reflux of contrast into the collecting system. And you can see how nicely the right kidney grade 3 reflux. And what's really important and remarkable is that this can be done without ionizing radiation to our most sensitive patients, our pediatric patients. Okay, moving on to bowel. This is a patient who had a long history of Crohn's disease, and the patient's gastroenterologist wanted to differentiate active inflammatory versus fibrostenosing disease. So you can see these are axial contrast-enhanced CT scans on this patient. And this patient has classic terminal ileal disease in the right lower quadrant. You can see that there's wall thickening of the terminal ileum, some haziness of the fat around the ileum. And so the physician wanted to know if we could kind of follow this patient's disease over time. And I said, well, Dr. Wilson does a lot of contrast-enhanced ultrasound for Crohn's disease. I would love to try it at our own institution. And so we brought the patient in to do a contrast-enhanced ultrasound of the terminal ileum. And I will tell you, even before going to the contrast, you can take a look. These are images with a high-frequency linear transducer. Look at how exquisite the bowel wall stratification is. You can see the wall thickening and the different layers of the bowel wall, including the muscularis propria, submucosa, which is echogenic, and the hypoechoic muscularis propria layer. And so you can see that the area of significant wall thickening in the area of maximal disease. And this is the image from microflow imaging showing some increased vascularity of that bowel segment. So here what we did was we changed our ultrasound transducer to a low-frequency curved transducer. That is actually a better way of getting ideal contrast enhancement of the bladder. Linear transducer imaging is not as good with reference to getting optimal contrast enhancement. So we did it with the curved. And so here we are just fixed on that terminal ileum. And what we did was we gave the full injection of contrast, and we scanned for two minutes, and we measured the ROI over time. So you form these time intensity curves. You draw an ROI over the mesentery and then over the bowel wall. And you can see, so there are two different ROI curves, and you report the time to peak and also the peak intensity in the area under the curve. And what you can do is you follow the patient's values over time to see how they respond to treatment, if the biologics are working, or if they need to be changed. So this is kind of the first step in how, you know, if you want to start a bowel contrast enhanced ultrasound program. And now I'll say this is one of the most pivotal cases that I've had in my career. And so when I was working at Einstein and I had rotated to one of the other hospitals in my network, we had a patient who had come into the hospital, and I'm going to, I'll show you this Cine. So this is a patient who had this contrast enhanced CAT scan of the abdomen and pelvis for right lower quadrant pain, and the Cine is showing an abnormal appearing appendix. And the radiologist read this as acute appendicitis, but that there was also a hypodense area within the appendix and said that there's a possible appendiceal mass, but that the differential diagnosis included an abscess. And so this case ended up, you know, coming to me because I was known as being the contrast enhanced ultrasound champion now for like our whole hospital network. And they said, Dr. Rogers, is there anything that we could do perhaps to help differentiate mass versus abscess? Because clinically, if the patient is considered to have a mass, the surgery would be very different. It would be a right hemicolectomy. If it was just an abscess, it would be an appendectomy. And you could see this patient is not the most healthy patient. And so, and then this is just a blown up image of the appendix with the hypodense area. So I said, you know what, this patient is perhaps, you know, has a high BMI, but I definitely think it's worth giving it a try to see if we could see this on ultrasound. And if we can see it well, why not? Why not do a contrast-enhanced ultrasound? So here we have the contrast-enhanced, oh, I'm sorry, the grayscale ultrasound. Let me see if I can stop this video. So using the low-frequency curved transducer, look at how well you can see that appendix. That appendix is new to stay very superficial within the abdomen. And we're able to see exquisitely exactly where there's a break in the submucosa. You can see right over there. You can see how echogenic the fat is. And we can actually see the area in question on the CT. It's showing up as a hypoechoic area. So we said, yes, we're going to go for a contrast-enhanced ultrasound and see if we can determine the difference, you know, if this is a mass versus an abscess. So here we have the Sine. And so you can see that we have enhancement of the bowel, but there is absolutely no enhancement whatsoever in the site of perforation or in the hypoechoic area adjacent to that site of perforation. And so here are some static images. So the static images are showing you the intact submucosa on the left. Then you see the break in the submucosa, which is the site of perforation. There's no enhancement whatsoever in that area around that perforated site. And so I called the surgeon and I said, this is a abscess. There's no mass. You can go ahead and do that appendectomy. And the surgeon did that appendectomy. And I basically held my breath for three days until the path came back. And the path did come back as perforated appendicitis with intramural abscess. And so then everybody was like, well, contrast-enhanced ultrasound is amazing. So I think that if you have sites or if you have these problem cases and you can use the contrast-enhanced ultrasound to add value and change patient's management, I mean, I think that your practice will grow and expand and just people will want to do more, order more contrast-enhanced ultrasound studies. So that's kind of how we really got started. But we didn't limit it just to bowel. We started just continuing to apply it to many, many different things. Okay, so here I have a few examples in interventional radiology. So this was a patient who had a history, if we look here at this bottom image. This patient had a CT scan that showed a colon mass. So this patient had a colon mass and then a liver hypodense mass here shown by the arrow. And that was suspicious for an abscess. So the patient was referred to interventional radiology for biopsy. Now on a grayscale ultrasound, you can see a very subtle area as outlined by the arrows. But the interventional radiologist biopsied that area and it came back non-diagnostic. So he came to me and said, Dr. Rogers, can you please help guide us with your contrast to help see if we can get a more diagnostic specimen? I said, of course I will. And so here we showed that there was an area of arterial phase hyper enhancement. Now the enhancement was so quick that we barely got it. However, the washout was so pronounced that we actually used the washout phase to help guide the biopsy. And this time it was diagnostic and it unfortunately came back as metastasis. This is a different patient who had idiopathic thrombocytic purpurea, status post-splenectomy, and then underwent and actually had multiple splenules that had developed. And interventional radiology was asked to ablate the splenules. So they did an alcohol ablation of the three splenules and they asked if we could do contrast enhanced ultrasound to show if it was successful. So we were able to show macrovascularity in the largest vessel, I mean the largest splenules you can see with these, you can see the arterial flow on spectral Doppler. But then we did the contrast enhanced ultrasound and we were able to show that at least two of these splenules showed significant enhancement, whereas only one of them was successfully ablated. So this was helpful to guide therapy. Marching down to another topic now, so gynecology. There are various ways to apply to gynecology. I'm going to talk about one. This is actually an off-label use, but I will explain in just a moment. So this was a patient who had primary infertility and she had a comprehensive pelvic ultrasound to assess for infertility. So we're looking at the uterus, the ovaries, we want to look for any endometrial, malarian duct anomaly or endometrial pathology. Here's a coronal 3D and we were able to see an endometrial polyp. There were no signs of endometriosis. We did a saline infusion sonohistogram and we followed it by HI-COSI, which is hysterosalpingo contrast sonography. And these are static images from a HI-COSI, which is the hysterosalpingo contrast sonography. And basically what this is, this is an off-label use of the contrast enhanced ultrasound agent where we're injecting the dilute microbubble contrast within the endometrial cavity. And you can see it's in the cavity and then going out the fallopian tube. And it's going out the fallopian tube and then we have free spill around the ovary. And this is showing us that the fallopian tube is patent. So here I have this video showing real time what it looks like. So here we can see, this is the dual image where you can see this is the uterus, the endometrium, and we can see that there's the contrast within the interstitial, isthmic, and fundibular portion of the tube. And you can see the free spill around the ovary right over here. Really incredible. And then there's another, my last case is the gallbladder. This was a patient who had cirrhosis and had a finding within the gallbladder fundus. It looked like a hypoechoic mass within the gallbladder fundus. It looked like on microflow imaging that there might be vascularity within it. Same with the color doppler image. But this patient is not a good surgical candidate. So they asked if we could do a contrast enhanced ultrasound to differentiate tumor-effective sludge versus a gallbladder mass. And you see the gallbladder is highly conducive to contrast enhanced ultrasound because it's very superficial in location. Oops, sorry, let me go back one. And so, I see. I think that that slide somehow disappeared. Well, I will tell you what that slide showed. The slide showed that there was no contrast enhancement whatsoever within that fundal lesion. And it turned out to be tumor-effective sludge. And if you close your eyes, you can imagine that the patient had a follow-up MRI. And on the MRI, that tumor-effective sludge turned into a gallstone and it had fallen to the dependent portion of the gallbladder. So it totally moved out of its spot in the fundus. So I'm going to be closing up now. So I also worked with a few of the physicists at Jefferson. And Dr. John Eisenbrae has given me some future directions for contrast enhanced ultrasound. Clinical trials will harness the microbubble biofacts for augmenting therapy in interventional oncology. Microbubble fragmentation can increase vascular cell and biofilm membrane permeability. And there's also ultrasound-triggered microbubble destruction that augments treatment of HCC. And I also have a final closing remark. I think this comic sums it up really nicely. Contrast studies are indeterminate. But when you can be definitive, it is really important for patients. How would you feel if your radiology report had words like, appears to be possible, equivocal, cannot rule out, blah, blah, blah, blah, blah? Or try a contrast-enhanced ultrasound. So in conclusion, contrast is here, got to move on, and it can be successfully integrated into radiology practice. You want to have a champion to run with it. You want to have buy-in from colleagues and clinicians. And it adds considerable value in so many situations. Thank you. So this is a poorly titled session, I think. Because I think it's difficult to cover a tutorial on how to do parametric imaging. Because you probably need to take individual parameters and actually cover them individually to actually give you an understanding of the process by which to obtain the image that you're looking at. So I think it's more of a survey of the problem and the promise of the result. And some of this stuff has really come about, if you were here for the last session on a general tutorial for imaging, ultrasound had a lot of promises in the 70s. And I think the technology is actually fulfilling those promises a lot more as we move forward. Image contrast, I guess, for lack of another word I put, depends heavily on the modality you're looking at, right? So in histological imaging, using optical imaging as well as electron things, you have reflective light or you have the use of electron beam and the characteristics of a beam to magnify an object or an image. In x-ray imaging, anything that affects the attenuation of x-rays is going to affect the contrast in your image. That could be atomic number, physical density, thickness of an object, electron density in CT or something like that. So over the energy you're operating at, notice I threw a little parameter in there that actually a dependency parameter that affects naturally the attenuation of things based on the energy you're at. So what are some of the bases for ultrasound image contrast and how might they be exploited for quantitative parametric imaging? So that's probably, I think, one of the promises of parametric imaging is you're going to get something quantitative out of it. So I do cover a little bit later generalized elastography, for example, and that's something that Dr. McAlevey talked about in the first hour where you do have a new type of contrast based on a tissue characteristic, but you don't have quantitative values. But as my one slide in here that covers that subject, in elastography you're actually getting shear wave speeds and you're getting measures of the shear modulus and kilopascals and things. So you're actually quantitating things. So I think that's the promise of these characteristics that can also synonymously be called parameters. So the things in this talk that we'll try to cover some of these topics is to start out with, and I'm going to go back and forth on some of these topics, is the tissue absolute backscatter coefficient, the BSC. Also scatter size, which is tied up in the backscatter coefficient. The speed of sound, to some degree I'll cover tissue attenuation is being exploited, and as I referenced before, tissue stiffness. And then the scatter number density or scatter concentration is something also that's exploited as a parameter for characterizing tissues. In order to determine backscatter attenuation, it traditionally was very difficult to do without an elaborate laboratory setting. So you've got water tanks, you've got very important, you have reference reflectors. And notice I used the word reference. So you have an elaborate experiment in a lab where you're making measurements on either tissue samples or phantom samples. That's not necessarily an image, but it's the basis by which we started. And groups develop good processes by which to determine backscatter coefficient accurately, and also the attenuation coefficients, and also speed of sounds in bulk materials as well as they could. But quantitative parameters are made more complex. One, we're moving away from this, but it always annoyed me that traditional B-mode acquisition, you have RF signals, but oftentimes those are totally removed. Even the absolute amplitudes are removed. So you take this RF signal, you just take the envelope, then you maybe low-pass filter it again, and then you just end up with these quasi-brightnesses of quasi-differences and acoustic impedances in the field, which are echoes. Big echo amplitude, little echo amplitude. Bright pixel, dark pixel. So that makes pictures kind of useless to you, other than on a qualitative basis. So you also have inherently complicated beam patterns based on the elements, based on the acquisition technique, which I list one, expanding apertures. User controls. I know transmit focus is going away. A lot of that 2D, a lot of that plane wave imaging is getting rid of that as a parameter. But if you've ever sat at still a standard of care system, you have these carrots over the side determining where you're focusing your sound on a 2D array. TGC, overall transmit gain, receive gains, all these knobs on the ultrasound system constitute user settings. And that will confound you in terms of comparing one set of data to another set of data, unless a concept where I'll come back to is you collect a reference. References are the key to, at this point still, in determining these parameter values even today. So also the tissue itself can confound you with, I mentioned attenuation, but you also have propagation properties, which are a key term for saying speed of sound differences in various layers of the tissue can confound your problem as well. So the reference phantom method that I keep alluding to is simple. You take an image on the left through your sample, and then you take an image to your right, which is a well-characterized reference. Characterized references, you go to these booths over here, they all have ultrasound phantoms. You put your little probe on there, you're going to get nice, well-developed speckle on the image and it's uniform. But if it's created well, it's going to have a known speed of sound, a known attenuation, and it's a known backscatter versus frequency. So with all those in point, a lot of times, it's just like in algebra, I tell my kids, right, if you've got two unknowns and two equations, you're good, and you just keep expanding unknowns and equations. If you know something, you can determine the other thing based on what you do know. And that's what a reference phantom is meant for you. And I will agree that I think that, for me, I like reference phantoms rather than reference reflectors. There's a lot of nuances to using a reflector, but we'll talk a little bit about some successful work there. But what you're doing with a reference phantom is you're taking ratios of those amplitude data filtered at specific frequencies, and then you're taking the ratio of the two and plotting them versus depth. And you end up with a descending line, it could be, right, or an ascending line. So you can determine the attenuation if you know the attenuation of the medium of the reference. So now you have a slope, something that looks like you can develop a linear fit to, and then once you know the attenuation of your sample, you can correct for the attenuation versus depth, and then you'll have a corrected amplitude versus the total depth corrected for that, and you can multiply then by your reference backscatter coefficient. So then you have your attenuation and your backscatter of your unknown sample to work with. And you have it filtered at specific frequencies, so now you have it over a frequency range, which the frequency dependence of scattering is going to be very important. This is admittedly, I think, a rather old reference, but when I saw it come out, I really enjoyed the article a lot, so I grabbed two slides from it. So frequency dependence of scattering is very tied up in, if you're interested in determining the scatter size of objects. And this picture by Sommer was really nice, where you can see that in, I think, oh yeah, there we go, sorry, you've got to look at this screen, not this screen. So this is small scatterers, these are very large scatterers, and the background is kind of in between the two. And then on the left, you have the low frequency, and on the right, you'll have a high frequency. So this says a lot to you immediately about what frequency dependence of scattering will result in terms of the backscatter coefficient. You'll get more, for large objects, will dominate at low frequencies, and you can see right here for the low frequency, you get lots of scatter here, and over here, high frequencies like small objects. So you'll get more scattering from small objects, small scatterers. And you can see that you really don't get a big shift in either way compared to these two outliers in the background material on the two sets of little sine waves on the periphery of this little object. So at high frequency, I would expect this to be hyper-echoic, and at low frequencies, I would expect this to be hypo-echoic. I expect it to be bright. Brighter is hyper-echoic versus the background, and hypo-echoic implies darker versus the background. I only left it, you can always go back and grab this reference yourself, but I only left it one example that I thought was kind of nice in this paper. And so what you're observing here is liver homangiomas in two different situations. So left is a case, and the right is a case over here. And they're done at two different frequencies, two and a half megahertz and five megahertz. Now there's no special parametric imaging in this, okay? And I'll tell you that this work was fortuitous. The only way he could do this, right, was because he was on an Accuson 128. People who are older in the room know the 128 was a great system, okay? But it was inherent. One thing about the Accuson 128, it had very narrow bandwidths. It did not have a broad bandwidth in its excitation, so if you ever look in RF signal, if you have a wide bandwidth, right, you're going to incorporate all of these scatterings across a wide range of effective scatter sizes, okay? And so that's why, unlike Kraft before, we filtered the data to give you the individual frequencies. The ACCI sign kind of quasi-gave you just a little snippet here and there. And you can see that there's a negative contrast enhancement as you move from 2.5 to 5 MHz, and then from 2.5 to 5 MHz for these hemangiomas. And so this would imply, likely, that the hemangiomas are larger scatterers relative to the adjacent liver parenchyma. Because we saw in that earlier cartoon picture that larger scatterers had typically larger scattering versus smaller scatterers. And so this for me is a very good scatter size image that wasn't intended to be a scatter size image, but he calls it such. So I'm good with that. So one of the things I've mentioned there is frequency. So there is a nice phantom that was developed out of Wisconsin. This is what we term the effective frequency phantom. And it had two windows on the top and the bottom, so you could access these cylinders at multiple different depths. So you could take attenuation into account for the background medium. But what made these interesting is the background was made with small glass beads, but the actual cylinders used these highly filtered set of polystyrene beads. And so you end up with a cylinder that has many magnitudes of scattering versus the background material, which is pretty even across this frequency range. And this frequency range covers, say, 4 to 8 MHz. And so cylinder 1, or I'm sorry, this is cylinder 3, would trigger off very quickly at a low frequency, and then cylinder 2 at a slightly higher frequency. And then at cylinder, at about 6.5 MHz, you would see cylinder 1 kick off in terms of ecogenicity changing from hypo to hyper, significantly hyper, versus the background material. And so here's an example just on a clinical system at 7.5 MHz, 6.5 MHz, or 5.5 MHz. So at 5.5 MHz, I mentioned cylinder 3 kicks off, right? But the other two are still hyper-echoic. And then at 6.5 MHz, you would have chipped off cylinder 2. And then once you got very high, 7.5 MHz, you've got significant scattering from that incredible resonance curve for these polystyrene beads, orders of magnitude. So you get lots of scattering at high frequencies for this thing. Now here's kind of why I put this in here, and that's because frequency, I keep emphasizing, is somewhat important in the work that you're talking about in determining frequency dependence of scattering. And so here, the only thing that changed on this logic system was the transmit focus. But you see a stark change in contrast in these cylinders, so clearly the effective frequency of the pulse has changed significantly. So if you're going to do this type of work, you have to take into account the frequency and the bandwidth that you're operating with. Here's another example of coded excitation in a clinical system, where on the left, here's just a multi-transmit focus done in fundamental mode, and you have contrast. But then when you move to a coded excitation, that same contrast modestly disappears considerably. So you move probably to a lower effective frequency, or shift it to bandwidth, one of the two, in the coded excitation system. So you change the contrast significantly. So you have to keep track of contrast. And then they're getting rid of frequency altogether on clinical systems, right? A lot of systems don't even give you the frequency anymore, which is frustrating to me. Although I guess one of the things is I don't know if you're responsible for parametric imaging development. It's the manufacturer that's got to give it to you. But if you were trying to be like Summer and make observations of images based on changes in frequency, you should at least be given what the frequency you're operating at. Here you're given these famous terms of penetration versus general versus resolution versus high resolution, which you can imply. Penetration you want higher or lower frequency. Up to high res, you want a very high frequency. And you can see our contrast cylinders trip off at a higher effective frequency. So you go from negative contrast to positive contrast, as you would expect to tell. But you're not given this information, which is a little bit frustrating. And to return back, like I say, you're going to have to find the backscatter's coefficient versus frequency. And then you're going to, one way to do this is compare it to some autocorrelation function like a spherical or a Gaussian model or an exponential model. And get a best fit to autocorrelation to the relative size you think that little data set you're looking at. And then you're going to spread that data set across your entire image field and make size estimates. And here's a set of cylinders with correctly represented color imagery. So, I think actually it's missing, wait, hold on, just one second, okay, there we go. There's this, I forgot I put a thing in there. So you can actually tell the cylinder sizes are correctly estimated using scatter size imaging mode. We'll come back to scatter size imaging in just a minute. So I'm going to briefly cover, because I think he covered it in much more detail, another parameter and that's tissue stiffness. I mentioned elastography is representing a different contrast based on tissue stiffness. But elastography is often done with external palpation, right, which has a lot of problems in 3D and the representation of the compressing force is difficult, okay. But what you're looking at is you're trying to find the displacement in pre-compression versus post-compression, okay. And the bottom line is if it's compressible, it's going to go like this, and if it's incompressible, it's not going to change much at all, okay. So this is soft and this is hard, okay. So that's pretty straightforward. And so you represent that in data and you can represent that in slopes and you can actually then represent white is softer, black is harder in an image that represents the relative tissue stiffness of objects versus the background, okay. Now where is this important? Right here I depicted this a little bit, but I like this slide a lot, from Gerald. Where you actually see on a standard B-mode image, you see here's one lesion, but on the strain image, you actually see two objects. And more importantly, notice the size of this object to the right, the larger object, versus the size on the B-mode. And why is that important? Because these properties can be used diagnostically as well. I mean, scatter size makes a lot of sense too, right? So the stiffness, okay. But I did have a smart aleck guy one time tell me, oh, I can do that already because I can go, I can go with my thumb, I can say this is not cancer and this is cancer. And I said, well, by the time you're able to palpate that, I don't know, it may not be good. So you can see on here, cancers with the margins are typically larger on strain images than benign lesions oftentimes are one-to-one in terms of size. And so you would expect to see a lot of spreading out of, that's why I guess you get a good surgeon and you get good margins, because you want to get everything rather than what you see. Okay. Maybe other types of optical imaging like fluorescent imaging will enhance surgical margins a lot, because then you can tag things and it'll be quantitative imaging in the OR where you actually see, cut, let's light it up, that's good, okay. Here's, get back to another strain image, this is thyroid, where here's a strain image on the left versus the mean bone image on the right, and you see under freehand compression, you see that the nodules are a little harder than the background parenchyma in the thyroid. And then you combine those two things we just talked about, stiffness with freehand elastography with the scatter size imaging, and another thyroid nodule, and you can actually see the nodule on the B-mode image, I'm sorry, I'm sitting here, I see it on mine, but you don't see why I'm doing it on yours, excuse me, I'm getting on the tail end of whatever my son gave me, which precluded me from having turkey in person. So that was kind of unfortunate. So here you have the B-mode image and you have a hypoechoic thing versus the background in the liver, in the thyroid, excuse me, and then you actually have the scatter size image on the right, where you actually have the objects are somewhat smaller than what are on the tissue parenchyma that's adjacent to it, and the overlay compressed combined image with a scatter size reference line here, given right here, and it actually shows like 50 microns for the nodule versus about 90 microns, which is consistent with 80 to 90 micron in the normal parenchyma for the liver that's observed. I should mention that, you know, I mentioned frequency dependence for scattering, right? You can have the tissue itself, the homogeneous tissue itself, can confound you, right? You could have a situation where you've got a dominant effective scatter at one frequency and at a higher frequency, another one dominates. Just think of large text and then the fine text you're supposed to read that you're going to get sued about, okay? So when you magnify the image, you don't see the big text anymore, but you do see the little text. That's what I'm trying to say is scatter size imaging can be something akin to that. This was something Dr. McAleary talked about earlier, and that's with shear wave elastography, right? Is using this, and I should mention, because they didn't put this traditional slide. I have no financial interest or disclosures whatsoever, okay? They simply said, hey, we want this topic, or you want to give the topic, I'll give the talk. And so I just pulled from vendors willy-nilly, okay? And these last few slides, some of them are from historical sites as well as vendor sites. And so here's where you're actually using, like he mentioned, the large amplitude pulse, the push, okay, the tissue, and you create a shear wave, goes across, and you use traditional imaging techniques to measure, or the detection pulses to measure the movement of that pulse across the tissue, okay? And that shear wave speed is then developed across the region of interest, right? And then you can have either two images, you can have a shear wave speed image, or you could have it converted into the shear modulus in units of, like, kilopascals or something. So you now have a quantitative parameter to represent stiffness, and this is there now, okay? So that's what I'm saying. Technology has reached up to this. Now, this is a very old slide, but I couldn't give this talk without giving deference to, it's like 19, well, the original, oh, that's in the 80s, it should, I think it was an 86 article, I apologize. The reference, but there's some newer stuff in his work. So Glitzy and Fulepa developed some parameters known as the intercept, slope, and mid-band fit, okay? And so they'll take one set of data from their patient, and much of this work started in ophthalmology where they were wonderful at classifying benign versus cancerous processes in the eye, which was, the eye was awesome because you didn't have attenuation effects, okay? And they then extrapolated this work out into other things, liver and other things. But this is where they started and found a lot of success with these parameters, okay? And what turns out is the slope and the intercept are both scatter size dependent, but the intercept also not only has a scatter size dependency, but something on the concentration and the impedance of the scatters. So they were able to develop images based on these parameters that have quantitative estimates. This is the reference reflector as the reference, I noticed to say the reference word again. And here are some representative from this article of the slope, intercept, and mid-band fit images versus the B-mode image on the bottom right. And once again, you can't even see what I was saying. The slope, intercept, and mid-band fit images given right here, and this is the original B-mode over on the upper left, okay? I'm not going to talk a lot about this either, but I will say that he talked about tomography, right? There's been a system, it started out, I'm not sure if DERC was the first one, or I forget the other ones, but there was a large group of people that developed the CURE, which was a computerized ultrasound risk evaluator, I forget what the acronym stands for, for the original system. It's been bought up a couple of times, and if they're here on the floor, I don't know, but they're SOFU, is you lay prone as a woman, and then your breast is held in this little water bath, and you have a little ring set of transducers. One sets off, all listen, including the transmitter, and then you run around the ring and collect But what makes this a beautiful image, right, is you actually have parametric imaging going on right here for tomography. You have exceptional resolution that's diagnostically worthy, but you also have sound speed images, reflection images, and stiffness images, okay? So we're beating mammography a lot in one parameter, right? All these states are saying we've got to report density? Well, they aren't reporting density, okay? They're taking these 2D projections and saying quasi-image obscurities, and even the BI-RADS estimate isn't classifying density anymore, it's classifying how many image obscurities do you have in there, okay? This system, right, Z is equal to row C, so you've got speed of sound directly related to density. So you actually have a way to determine breast density with ultrasound, quantitatively, okay? Unlike x-ray mammography, which I think is great for what it does, but it doesn't have the same value right there, okay? You have the other parameter that you're going to see on the floor, if you go over there, I'm sure. I only pulled two of them, you'll see both of them right here. GE has the UGAP, the ultrasound guided attenuation parameter. I didn't really know much about it until I put this talk together, but I'll say this, that right here, you'll see this amplitude variation versus depth, and you see, oh, I thought ultrasound dropped off with depth, right? Well, yes, unless you take into account those system dependencies, right? You've got TGC, you've got overall gain, you've got focusing, you've got all those things that make this thing crazy, okay? But here is a corrected curve, corrected for those system parameters, okay? And somewhere up here, and I talked to you, I never really read the slides too, I apologize. I don't, because I don't like to read my slides. But what they do is they take, before they give you the system, or before you're going to do this, you take a whole bunch of references, or they took a bunch of references, at specific frequency settings with system settings. My assumption is, what they do is, they then, if you're taking a picture of my, we'll call it fatty liver, okay? So you take a picture of the fatty liver, and you'll end up, it'll take one, or it'll take a number of data sets at those system settings, okay? And then take the reference phantom that was collected prior, and then correct for the system dependency, and come up with the slope, and give you an attenuation value for the region of interest, or the regions of interest, to create an attenuation image, with good imaging. And there's another company out there doing similar work, and that's Canon. I didn't see a lot of, I couldn't find any others right off, I'm not saying I was exhausted about it. But they also have, if you notice here, curves implying that they have system dependencies, okay? And you can kind of tell one of them is a focusing dependency, you could have something with TGC or something like that. And then they combine all that to give you, over on the right, the corrected curve, which then you get this slope we saw with the UGAP, and they call theirs the ATI. And then they come up with, for these fatty livers, you come up with 0.67, 0.82, and 0.94. You're going up from mild to moderate to severe fatty liver. But I will say this, you might say, well, I can just look at the pictures. Well, one, you don't get a number, okay? These are giving you real parametric numbers that are based on something. Secondly, don't think fatty liver is just fatty liver, because that crossover to cirrhosis, right? You're going to change the attenuation significantly, right? And you'll change the scattering significantly. So there's a complex problem going on that you'll need both parameters tracked if you really want to look at cirrhosis versus fatty liver. So I only have like a couple more slides. I've got to go quickly here. I'm in my last minute. I'll say this, that there are other manufacturers that produce these research systems, like the Visual Sonic Systems. What they do is they'll let you collect, they'll store the system settings for you so you can go back the next day and take your reference then if you didn't have time. If it's getting late, you had to get home, you had to get the soccer game, all that stuff. So you've got, and the power spectrum are given there on the green, and the overlapping curve are acquired day one, day two, day three. So you can see it kind of works from the power spectrum that they do store the system dependencies well so you can re-image later on. So those tools are available. This is the last two slides I have here. Statistical parameters, I clearly can't cover the subject, but I'll mention three off the bat. But they're keenly associated with determining the goodness of your data to fitting either a Rayleigh or a Gaussian set of data. So am I Rayleigh-like or am I non-Rayleigh, am I Gaussian or am I non-Gaussian? So you have the kurtosis, which is just the ratio of the moments, and then you have the three over beta term, which was developed at Wisconsin, and then you have the Nakagami parameter. I didn't see that until I saw some literature on this one. But one of the problems is, if you have fully developed speckle, I'm implying you have lots of scatterers, this ain't going to be the parameter for you. What you're looking for is low scatterer concentration, and you'll deviate, you won't get a fully developed number. You can see in the picture here, for a large volume, three over beta approaches zero. Because I get lots of scatterers in there, and I get true Gaussian statistics. So the three over beta parameter approaches zero. Kurtosis approaches three. So if you're lower than those values, or higher than those values, in the three over beta case, then we're talking that you can do something, and you can even use a reference phantom with known concentrations of scatterers to come up with an effective scatterer concentration estimate for your data. And this is a very old paper by Cook, right, this is, once again, 1986, I kind of like old papers, I don't know why. But you can see for a phantom, but then in the normal liver, you were just below three for the mean number, and for, kind of, the other value that falls below that was one abnormal case that he reported on, three standard deviations out of the mean for the other one, so the statistical parameter worked for him that long ago. So I'm saying there's some work in that area that people are doing as well, to elicit scatterer number density as a parameter of interest. Whether it's a bulk parameter, or you can measure over regions, and actually give scatterer concentration images is another parametric imaging that's coming. I did not highlight it over here, I highlighted two things that Kiba, someone told me Kiba is remaking itself, but C, contrast enhanced ultrasound, but then PECUS, and shear wave speed, and then the volume blood flow stuff, I didn't cover the top one or the bottom, I covered the little things related to the middle two, but I'm only related to PECUS. PECUS is a different approach, and the literature has comparisons between PECUS and, for example, the Litzy-Fulepa parameters. So I highly recommend you go out there, look at Kiba for all of the modalities, especially for ultrasound, and you can see the work that's being done to develop more parametric imaging. Sorry, it was just an overview, I wish I could have covered it in more detail, but thank you very much.

ultrasound contrast

contrast-enhanced ultrasound

CEUS

ultrasound microbubbles

liver lesion evaluation

diagnostic imaging

vascular structures

pulse inversion imaging

microbubble technology

interventional radiology