Ultrasound is the primary imaging modality used in children for the diagnosis of diffuse liver disease. It is a technique that can be performed in a short period of time and repeated over a short period of time. However, we should take into account that for a parametric assessment to obtain a reliable result, it is necessary to use standardized protocols. In contrast to ultrasound, it is increasingly used in children and it can be performed without anesthesia and in a short period of time in awake children. COS allows to visualize the vessels so it is a very important and nice technique to assess the vessel patency. In particular, it can confirm a clinical diagnosis of complications after a solid organ or a stem cell transplantation and provide additional information that are important for the management of the children and it can be regarded as a problem solving. This is a case of liver transplantation in a 12-year-old girl and with Doppler, it is doubtful if the portal vein is thrombocytic or not. And after contrasting ultrasound, it can be diagnosed as partial thrombosis. And this is another case with liver ischemia and abscess formation and other complications of a liver transplant. Shortly, the lastography is a technique now widely accepted as a substitute of a liver biopsy in several clinical scenarios and research has been conducted mainly in adults. There is a close correlation with the results of liver biopsy. A number of studies have been performed also in the pediatric population. When we use the term shear wave elastography, this term refers both to vibration control of the transient elastography which is performed with a standalone system, the fibroscan system and the RFI-based techniques that are the techniques performed using the ultrasound systems. For all the techniques, it is important to use a standardized protocol to obtain reliable measurement. And you can see here that there is a very long list. However, the most important is the quality of the beam-only image because we are using the strength of the ultrasound beam to generate the shear wave. So the quality of the image tells us if the measurement will be a reliable one or not. We should be aware of artifacts and most of them are below the liver capsule due to reverberation artifacts or around the vessels and this is due to vessel pulsation but also because the vessels may reflect back the ultrasound beam and therefore there is a false increase of liver stiffness. But in children, some of the items of the recommendation for a good acquisition are difficult to follow like the breath-hold or fasting intercostal approach. For the breath-hold, the SRU consensus has suggested to acquire a long sine loop to review the sine loop and to select the image with the most stable pattern. Studies have shown that a shallow breathing is acceptable. However, we have to keep in mind that free breathing can produce artifacts. For the fasting, liver stiffness measurement can be performed immediately before next meal, should not be mandatory in infants. However, it must be reported whether the child was fasted or not because after eating, there's an increase of liver stiffness that lasts for up to hours. For the intercostal approach, in case of liver transplantation, the epigastric approach may be used. However, the results are higher, the measurement is higher with respect to that obtained in the intercostal approach and this is likely due to cardiac motion. It's important to report that measurements were performed through an epigastric approach. And the transducer frequency is also important to report because the shear beam velocity depends on the RFI frequency. Different heart transplant systems provide a different liver stiffness measurement in the same subjects. Therefore, it's not possible to use a cutoff to mimic accurately the histologic classification. And due to the variability between vendors and between techniques, it's also important to use, if possible, the same equipment and the same transducer in follow-up studies. Respect to vibration-controlled transient elastography, the RFI-based techniques give lower results. So, the cutoffs and the values are not comparable. And shear beam elastography assesses stiffness that is related to liver fibrosis. It's just a number, a biomarker of liver fibrosis, but it should be evaluated together with clinical data and laboratory data. And this is strongly related to liver fibrosis. But there are several other conditions that may increase liver stiffness independently from liver fibrosis. And stiffness increases, leading to overestimation of liver fibrosis in case of acute hepatitis, infiltrative disease, obstructive cholestasis, congestive heart disease, or any other condition that increases the volume of blood in the liver. And in most chronic liver disease, there are inflammation and fibrosis. And it's difficult to separate the role of each of them. Therefore, the value should always be interpreted using also laboratory tests. And we should take into account that liver stiffness provides continuous numerical values, whereas any histologic scoring system is based on a categorical scale and separately from fibrosis and inflammation. Therefore, even in the best condition, there is an overlap between the consecutive stages of liver fibrosis. And in some etiologies of liver disease in children, like cystic fibrosis, biliary atresia, or in fountain patients, there is a complex and variable interplay of several factors that may lead to an increase of stiffness independently from liver fibrosis. These factors are mainly congestion, obstructive cholestasis, or inflammation. Therefore, we cannot recommend specific thresholds, also because studies were performed in children with mixed cohort of liver disease. And the rule of four that has been suggested for adults cannot be applied to the pediatric population. Based on the data in the literature, we may suggest two cutoffs. And one is up to 5 kPa. This value can exclude the liver fibrosis in children, whereas a value at 15 kPa or higher can be used to diagnose advanced disease, except in children with biliary atresia, cystic fibrosis, or fountain patients. And the shear wave elastography techniques are very helpful for the diagnosis and the follow-up of children with chronic liver disease. But cutoffs cannot be recommended because of the heterogeneity between published studies. This is a meta-analysis performed, including 12 studies, several etiologies of liver disease. And as you can see, the accuracy of the technique in evaluating chronic liver disease is pretty high, with an NIUC of 0.91. For the follow-up of children, the SRU consensus has suggested to use the percentage of change in liver stiffness over time, having the value at baseline as the referral value. And this may help in the assessment of prognosis and response to treatment. Again, stiffness is a combinational factor, and fibrosis can be the main player, but it's not correct to translate this parameter directly into an histologic scoring system. Prefer the intercostal approach, holding the transducer to avoid any pressure on the abdominal wall, and be very careful about the quality of the measurements. The quality factors suggested by the vendors cannot always be met. However, in case of a value that is unexpected due to the patient's condition, it's better to repeat the measurement. And the US ultrasound system used to measure liver stiffness must be specified in the report. Do not compare the liver stiffness measurements that were obtained with different technique or with different transducers within the same technique. And in the absence of large data on chronic liver disease, do not use liver stiffness to exactly stage liver fibrosis. And for the follow-up, it's very important, liver stiffness measurement in detecting stability or worsening of the liver condition. Therefore, it's important to report the values from previous examination in your report. The value in normal children is up to 5 kPa. There are differences between liver stiffness values obtained with different systems. But when they are within the normal range, these differences are not clinically relevant. And there are conditions, some illness that may increase the level of the liver stiffness independently from liver fibrosis. Let's see just a few etiologies of liver disease for whom we have data. This is a meta-analysis in subjects with biliary atresia. And it showed that patients with biliary atresia had a very high level of atresia. A higher stiffness value. And the main difference was 2.3 kPa respect to patients without biliary atresia. And this is in cystic fibrosis, the recommendation by the experts. It can be used for screening and monitoring in the case of cystic fibrosis. Repeating the liver stiffness at least annually. And cut-offs were given for transient elastography and the two-dimensional short-wave elastography techniques. And as I said before, cut-offs are different between the two. And in case of fountain-associated liver disease, fibrosis is always present by adolescents. And fibrosis increases over time. However, in this case, there's congestion that is the main driver of liver stiffness. And the values of liver stiffness in these children may be in the range of liver cirrhosis or even higher, even without liver disease. And the position paper on rare disease has recommended to use the longitudinal assessment of liver stiffness in these patients. This recommendation is weak because it's based on very few studies, but it's very helpful. And liver stiffness can be used also in subjects with sinusoidal obstruction syndrome. In adults, it has been shown that liver stiffness increases even before clinical symptoms. And the current research is focusing on using liver stiffness to identify patients early for a timely treatment. This is a study performed in children showing that is very helpful also in the pediatric patients. And muscle D has become now the leading cause of chronic liver disease. And the prevalence of muscle D is 34% in children with obesity. And it is worrisome that studies have shown that approximately one third of children with baseline stethosis will develop definitive MASH in a two-year period. This is the largest study performed in patients with muscle D biopsy-approved MASH and cut offs using two-dimensional shear wave astrography for significant fibrosis and the technique fibrosis were given. And in this case, we may also quantify liver fat content using ultrasound. There are several algorithms to be used, the attenuation coefficient, the backscatter coefficient, or a combination of both. This study was performed using an algorithm that has a combination of both. So, in conclusion, COS is often a problem solving and stiffness value is just a marker of the disease must interpret together with other data. And it is important to follow a correct protocol to obtain reliable results to take into account that there are confounders when evaluating liver stiffness. And in some etiology of chronic liver disease, mainly in children, there's a complex and variable interplay of several factors that may increase liver stiffness. Quantitative ultrasound is of great interest, but also in this case, it is important to follow a standardized protocol for correct and consistent results. Thank you for the attention. Our next speaker is Dr. Jonathan Dillman from Cincinnati Children's Hospital Medical Center, and he's going to be speaking to us about MRI techniques for evaluation of diffuse hepatic disease from elastography to AI techniques. Dr. Dillman. Thank you, Dr. Schooler. Good morning. Perfect. So, we're going to talk about diffuse liver disease. Now we're going to, instead of ultrasound, move into the MRI realm. And specifically, we're going to talk about first non-AI-based approaches, and that's Graham will touch on AI-based approaches using MRI data to get at characterizing chronic liver disease in children. And again, we're going to focus mostly on fibrosis here. There's just not enough time to focus on iron, fat, fibrosis, et cetera. So, as was shown in the last talk, there are many causes of pediatric chronic liver disease. You can see the list here. But what they all have in common, for the most part, is that there's some degree of chronic or repetitive injury over time. Inflammation occurs. There's loss of hepatocytes. These hepatic stellate cells, which are essentially a type of myofibroblast, get activated, and then extracellular matrix, collagen, et cetera, gets laid down, and we end up with scarring or fibrosis. The reference standard for detecting fibrosis and measuring it historically has been to put a needle in the liver. Most of us don't like that. It's invasive. Children are going to require sedation or anesthesia. The bigger issues to me relate around sampling error. The liver is a very big organ, and when you take a tiny needle and stick it in the liver once or twice, you can get an area that's underestimating what's going on, overestimating what's going on, but not really telling you what's going on, on average. The other issue is when we stage liver fibrosis, or when our pathologist colleagues stage it, it's using, typically, a Metavir system, an E-Shack system, 1, 2, 3, 4. My 2 is your 3. Your 3 is my 2. And when we've recently looked at this at our institution, the kappas between world renowned hepatopathologists are still only in the 0.4 to 0.6 range. So that's a challenge. We need more objective measures of liver fibrosis. So we're going to talk about imaging now. And we're going to go from conventional anatomic assessment of images to elastography, to T1 mapping, and then the AI. So let's start off with conventional assessment. So when I open a liver MRI, and this is a patient, for example, this was a patient with unknown cystic fibrosis liver disease. This is a 12-year-old. It's very obvious this is a fibrotic liver. We can look, for example, for altered morphology. The liver is shrunken. You see surface nodularity. You may see areas of segmental atrophy or hypertrophy. The problem is, for me, if I see this, the fibrosis is already moderate to severe. And then for many of our diseases, many of our conditions, especially where there are therapies available, we want to detect it at a much earlier stage. And we'd like to be able to quantify it. And this is really hard to quantify. We can look for altered signal abnormalities. So I look for areas of sort of patchy or reticular increased T2-weighted signal. I look for areas of altered enhancement. You can detect fibrosis with either a conventional gadolinium-based agent looking for delayed enhancement, or it can appear hypo-intense on hepatobiliary phase imaging, which is sort of the reticular signal we're seeing here. But again, by the time we see these signal changes, the horse is out of the barn, so to speak. It's moderate to severe disease. It's also very hard to quantify. We can look for portal hypertension. So we can see a big spleen in this case. We can see some varices, a bit in an unusual location in the root of the mesentery. We can look for ascites, but again, a late marker of fibrosis. We want something we can detect earlier, and something that we can follow in a more objective way. So let's move on from our anatomic assessment to elastography. So elastography, as we heard in the last talk, is based on measuring the speed of shear waves or transverse waves as they propagate through a tissue, in this case, the liver. And it's no different than dropping a pebble in the pond, like you can see here. You drop the pebble, there are ripples. And it turns out in the liver, the stiffer the liver, the faster the ripples go and the wider the waves are. And it's as simple as that. The nice thing is we can go from shear wave speed to shear modulus to Young's modulus back and forth. There are different ways to present it, different ways to understand it. So the way it works, as I mentioned, we need to set up these shear waves or transverse waves. We're going to use a paddle, basically a diaphragm that we place over the costal margin over the liver. It's going to vibrate or buzz, and it's going to generate these shear waves in the liver that we're then going to use a modified phase contrast pulse sequence. And we're going to image these waves. The faster, the broader, the stiffer the liver. The waves that we image then get converted to a stiffness map. We don't actually image the stiffness map. The stiffness map is actually made from the wave images. And then ultimately, for MRI, we're not going to present shear wave speed. We're going to present the shear modulus, which is going to be in kilopascals, which is important to remember is different than Young's modulus, which is presented at ultrasound. So the ultrasound KPA value is always going to be higher than the MRI KPA value by about a factor of three. It's not perfect, but you've got to keep that in mind or people will get confused. The nice thing is this technique now is available based on all modern scanners. So here are some wave images. On the left is normal, nice, thin waves. On the right is a much stiffer liver. You can see that a red plus a blue line equals one wavelength. Think of it sort of as a sine wave. Maybe red's above the baseline, blue's below. But this is a stiff liver on the right. And this is a patient with PFIC. This is about as stiff as it gets. You can see that there's basically one wavelength, one and a half wavelengths in the entire liver. I mentioned you need a driver. So we have the diaphragm vibrating over the right upper quadrant. That's connected to plastic tubing. The plastic tubing goes out of the room and connects up to this device, which is essentially a subwoofer. And there are two settings that we have to keep in mind. One is how loud the subwoofer is. And that's our amplitude. And we can adjust that up and down. So if we're not getting good waves, we may need to turn it up. If our waves are too exuberant, we can turn that down. The other thing is frequency. So not just how loud the sound waves are, but how frequently they're vibrating. That actually changes our results. So if you look at the wave equation here, the velocity of a wave relates to the frequency of the wavelength. Amplitude or volume's not in here. But frequency is. If we go somewhere other than 60, if we go to 80 or 40, this actually changes our speed, which changes our results. So we have our frequency locked down at 60, as you can see here. We don't want to change that, except maybe in the research setting. So how well does shear wave elastography do from an MRI standpoint? Well, this is one of the larger studies. This was out of the Mayo Group, presented in Radiology in 2016. And it's good. There's no question, as histologic fibrosis goes up, stiffness goes up. But the thing I want to draw your attention to, just like with ultrasound, there is a lot of overlap between the different stages. So it gets very hard in adults and children to say, a given patient's an F2, a given patient's an F3. We can put you in a range. We can follow it over time. But be careful about trying to do staging. This is a pediatric study here. So this was done by Andrew Trout, published in Radiology in 2018. And this took all comers, as far as pediatric liver diseases go. And the area under the curve for discriminating 0.1 versus stages 2, 3, 4 was 0.82. It's OK. It's not bad. It adds value, likely. But it's not perfect. The area under the curve is not 1 here. And if we used a cutoff value of 2.5 kilopascals, the sensitivity, you can see, was between 80% and 90%. And the specificity was 70% in this paper. We've recently looked at MR elastography in autoimmune liver diseases. And if we focus, that's PSC and autoimmune hepatitis. We see very similar trends. We can see Metavir here on the left, Eshach here on the right. But again, there's overlap. In this study, a cutoff value of 3.3 kilopascals had a sensitivity of 65% and a specificity of 90% for, again, separating 0.1 versus 2, 3, 4 fibrosis. And this was published, if you want to see it, in AJR in the last year. Stiffness has also been associated with the presence of portal hypertension. Below a certain stiffness, unlikely to have portal hypertension. Above a certain stiffness, much more likely to have portal hypertension. It's not perfect. But there is a very strong association there. If you look at clinical markers, such as the APRI or fibrosis score, which, for example, take a patient's ALT, may take their platelet count, may look at age, things like that. They go into an equation. Imaging is going to do better than those sort of formulas in general that are commonly used in the clinic. What about normal stiffness values by MRI? So this is a multicenter study that was done at Cincinnati, CHOP, MGH, and Wash U. It was published in Radiology in 2019. And the mean stiffness, you can see across the cohort here, was about 2.5 kilopascals. And these were all patients that had no known liver disease and were considered healthy. There are a few patients out here on the right that I suspect there's something going on. There's probably some sort of undiagnosed liver disease, because these seem a little high to me based on 10 years of doing this in the clinic. There was no difference in stiffness measurements between 1.5 and 3T, which is important. And this was, again, published in Radiology. MRI elastography failure in children. So we do a lot of these. This was published in 2017. And we had about a 4% failure rate. So what causes that? Sometimes it's a patient can't hold their breath. Sometimes they can't hold still. Sometimes it's a very large body habitus. We have patients that weigh 300, 400 pounds in our metabolic dysfunction-associated liver disease clinics. And then there's iron overload. If you have excess iron, it can null or kill the MRI signal. And you just can't track the waves. An adult study done right after our study using a very similar methodology had an almost identical failure rate. So children, in general, don't seem to have a much lower diagnostic or non-diagnosis rate. There are confounders. So we have looked at this pretty extensively. In any study where we look at patients without liver fat and with liver fat, we tend to see fat having a slight influence on liver stiffness. That is debatable. There are people out there that say that fat does not have an effect. In general, I think the literature is still very suboptimal around fat and MR elastography. That said, can we use MR elastography in the setting of a fatty liver? So here's, this is a child, an obese child. Loss of signal on the out-of-phase imaging. Clearly there's fat. The PDFF is 44%. The liver is markedly stiffened here. This is a patient that's going to have significant inflammation and or fibrosis related to MASH. At our place, this is gonna lead to a biopsy. Here's another patient with fatty liver disease, known MASH, on biopsy. We can see the original stiffness at eight years of age is 2.2 kilopascals. Five years later, look at the liver. Look how it's changed. It's 5.5, markedly stiff, markedly heterogeneous. So we can clearly use elastography in pediatric patients with fatty liver disease, even if fat does have a slight confounding effect. Congestion, is this stiffening due to fibrosis? Is it due to congestion? I don't know in this Fontan patient. I can't tell you, but I can do those followed over time. And if I see the liver getting stiffer, either their congestion's getting worse or their fibrosis is getting worse, and we're gonna wanna intervene upon that. And that inflammation clearly affects stiffness. This is a patient that had an original stiffness of 5.5. This is autoimmune hepatitis. They got steroids and their stiffness dropped to 2.1. And that was over a two year period of successful medical treatment. We'll move on, T1 mapping. I suspect not too many folks are doing it, but as I go around this meeting and talk to folks, especially about Fontan, there are lots of folks interested in T1 mapping in the liver. So when we do T1 imaging, T1 imaging is qualitative. But if we do mapping, we can actually get a quantitative number, an objective assessment of T1 or longitudinal relaxation. And in the human body, these T1 measurements are over a range of milliseconds from a few hundred to a few thousand. And every tissue has a unique T1 value or a range of T1 values. And these values change with pathology. T1, if you remember, is how fast longitudinal relaxation occurs. And the T1 value is specifically when 63% of relaxation has happened. And it's been shown in the liver to be associated with fibrosis and inflammation and probably congestion as well. Most frequently, we're going to use a modified Licklacquer approach or a MOLLE approach which we have borrowed from our cardiac imagers. And we've simply moved it down to the liver. And in its current form, it's a single slice per breath hold. We do four breath holds, four slices, and then we take an average value. So for each one of those breath holds, we actually get eight source images. Each one of these images was obtained at a different inversion recovery time. The curve is then plotted like I showed you, voxel by voxel, and then we get a parametric map on the bottom. And I can draw regions of interest in the liver. And you can see this is a very heterogeneous liver with areas of relatively normal T1 and areas of peripheral sort of markedly elevated T1 that are almost certainly fibrosis. And you can see there's associated capsular retraction here as well. So this is one of the very early studies. This is about 10 years old now that took all comers and adults and did T1 mapping of the liver and correlate it with pathology. And they actually did a correction for iron or T2 star and field strength. This is an FDA cleared product, but they showed that T1 went up with increasing histologic fibrosis. We have multiple papers now on T1. These are just some examples here. So we have a normal appearing liver. We have normal stiffness, but we have an elevated T1 in this patient. This gets a little bit probably at the confounded nature of T1. I mentioned T1 goes up with inflammation, T1 goes up with fibrosis. This liver has a normal stiffness, does not look at all fibrotic. There's probably minimal fibrosis here and this elevation in T1 is probably mostly inflammation. On the other hand, this is a very enlarged, very heterogeneous T2 bright liver, very stiff liver. This patient has a very markedly elevated T1. And in fact, under the microscope, this was both inflammation and fibrosis and both were severe. So still a bit confounded. Do we have normal T1 values in children? So the first study that looked at this in the liver in children was in a very small study. You can see here the ends are 11, 12 patients. It was published in 2019. The good news is a much larger study now containing more than 100 patients was published in abdominal radiology in the last year. And they actually looked at normal patients with liver disease and then patients with steatotic liver disease. And while the box plots absolutely overlap, you can see, for example, as we go from normal to steatotic liver disease, there's significant changes in the T1 values. We've recently looked at T1 mapping in autoimmune liver disease. And what we can see here, for example, is Metavir score versus T1. But this is interesting. This is not T1. This is actually the T1 heterogeneity or IQR. And in autoimmune liver disease, as the heterogeneity of the liver increases, the fibrosis scores are increasing. So it was kind of an interesting association. There was a less strong association between absolute T1 and fibrosis, but the T1 IQR or heterogeneity, there was a quite strong association there. And this is a table showing many of those associations. For example, T1 IQR correlates very nicely with fibrosis four score, APRI score, platelets, GGT, AST, ALT, all these things that you would expect it to correlate with. I mentioned Fontan. There are multiple Fontan papers now, at least four or five, showing that T1s are elevated in Fontan. We had one of the first papers in 2019. This is a more recent paper in 2021. And to date, the work has been relatively simplistic, showing that T1 values are higher compared to control populations. But we need more data showing how they change over time. Does T1 relate more to the congestion or fibrosis? We don't know those things. And work needs to be done and is being done regards to T1 mapping and Fontan. One thing that was interesting in this paper was that the T1 values didn't seem to correlate with the central venous pressure in the Fontan pathway, which gets at maybe it is indeed a marker of liver congestion. So in the last two minutes here, I'm gonna wrap up what's our future as far as assessing chronic liver disease in children and using AI. So in 2019, we took anatomic liver imaging. We segmented the images. We extracted radiomic features and then tried to predict the liver stiffness. And we had 105 radiomic features. We had 27 clinical features in the model as well. This just shows that many of the features are correlated. So you wanna get rid of the highly redundant correlated features. So we do that using a lasso method. And the gist is for predicting normal versus abnormal liver stiffness, when you look at radiomics, clinical, and then the combined, the area under the curve combined for clinical and radiomic data was about 0.84, which is pretty good. And then we went on to look at different cutoffs to see if we tried to predict greater or less than three kilopascals, greater or less than four kilopascals, for example. The best performance was actually right at three kilopascals, which is interesting because that's actually a very clinically relevant stiffness value. The features are what you had expected. They related to the size, the flatness of the liver, fat fraction made it into the model, and then some markers of grayscale signal intensity, how bright the liver is, and how heterogeneous the liver was. We then went on to deep learning, saw very similar results, almost identical results instead of using radiomics. And now where we're at now is actually taking, instead of just T2-weighted imaging, taking multimodal data, T1, T2 diffusion, doing radiomic feature extraction, deep feature extraction, and then doing our predictions, and not only doing categorical predictions of normal and abnormal, but also continuous predictions of what the exact stiffness is. Based on your T1, T2 diffusion, you're at 2.2 kilopascals, you're at 5.7 kilopascals. And that was actually presented in this meeting here. And this is the scatter plot that we presented here. There are still outliers, it's not perfect, but MR elastography or almost anything we do is not entirely perfect, so still work to be done here. So takeaways, MRI evaluation of liver fibrosis and chronic liver disease is evolving. It's evolving quickly, and it's increasingly objective, which I think is a good thing. All of the techniques that we see in adults, elastography, T1 mapping, et cetera, can be performed in children, and they can be performed in children with very low failure rates. MRE and T1 are certainly confounded. That doesn't mean they're not useful, though. We just need to figure out how to harness the information that's in there. And then we need more investigations, especially how these change over time, how they correlate with meaningful clinical outcomes. And then starting to combine these doing a multi-parametric approach where we have stiffness, we have T1, we have plus or minus AI, et cetera. I think that's where the best benefit and we'll have the most accuracy going forward. Thank you very much. Thank you, Dr. Dillman. Our last speaker is Dr. Erica Riedesel from the Children's Hospital of Philadelphia, who's gonna be sharing some challenging and emblematic cases and associated relevant information. Dr. Riedesel. Thanks, Dr. Schouler. Thank you for the invitation to speak. I'm thrilled to see this number of people in the room for a talk on advanced liver imaging. I'm excited to be here. I'm excited to be here to talk on advanced liver imaging, but I'm always curious and I did wanna start by asking by show of hands, how many people in the room are pediatric radiologists? Oh, look at y'all. Okay. Any adult radiologists? All right. Any non-radiologists? Right, okay. That's very helpful because it allows me a little bit to tailor what I'm about to talk to you. So for those very last part of this session, I was tasked with talking about challenging and emblematic cases in liver disease imaging in children. My goals today are to provide some real life examples of pediatric liver disease from my day-to-day practice. The cases that I see the most coming across on my desk as an actively practicing pediatric radiologist to discuss some of the challenges that we see in those day-to-day practices. But I also want to have a bit of a focus especially since we have a predominantly pediatric radiology audience. Why this is important and why we need to continue to strive to do it well and even to do it better. So you could potentially rename my talk, liver disease imaging in children, why it matters. So we're gonna start with our first case. This is a six-year-old male with beta thalassemia who's been undergoing recurrent transfusions and the imaging question that we are tasked with today is to quantify liver iron content. So this is probably practicing in Atlanta, one of our most commonly seen quantitative liver imaging questions. So as you'll remember from training liver iron, overload can occur secondary to primary disease such as hemochromatosis, but also secondary to transfusion related iron overload such as in patients with thalassemia, sickle cell disease or myelodysplastic syndromes. Practicing in Atlanta, we have a very large beta thalassemia and sickle cell population. So that's why we see so many of these patients. And why is liver iron quantification so important? Well, we know that liver is one of the main iron storage organs and the first to show iron overload. And one of the key drivers in patients outcomes for patients with thalassemia is in fact monitoring and treatment of iron overload. It has made the biggest impact in these patients outcomes and their quality of life ever. And so our liver iron quantification measurements will change their therapy and we've seen significant downward trends as we've developed more and more quantification imaging in overload related medical complications for these patients. So when and how are children screened for iron overload? Well, historically recommendations have been to screen with serum ferritin levels every three months. Over time, we know that serum ferritin levels will correlate well with changes in total body iron storage. And we have threshold markers for those that are associated with increased risk of complications. The advantage, labs are really easy to get, right? They're very, very low cost. The cons to this approach, lab draws are ouchy. Most adults would not want to have their labs drawn every three months, let alone a child. And we know that serum ferritin lags significantly behind liver iron content. So assessment of liver iron concentration has become very much the standard of care for these patients. And the recommendations currently are for an assessment of the LIC after one to two years of regular transfusions. In very young children, in very young children, those who might really be challenged by MRI imaging, we can hold off on liver iron quantification if they have acceptable ferritin values. But nonetheless, MRI really has become the gold standard for non-invasive quantitative and qualitative assessment of LIC. So with MRI qualitatively, we can identify liver iron concentration because hepatic iron stores influence relaxation time of liver parenchymal signal. More iron equals darker liver is what I tell my residents. And so qualitatively, as you can see in our six-year-old little boy, you can see on this T2 weighted MRI with that saturation that the liver signal is significantly less than those of the other abdominal organs, but also importantly, significantly less than skeletal muscle. So we know that this child qualitatively has iron deposition. We can also do qualitative assessment with gradient echo imaging. This is using our in and out of phase images. And we know that the presence of iron decreases signal intensity on the in-phase images. Again, I always tell my residents, this is because the in-phase images are obtained at a longer TE. So of course they're going to have less signal because of the iron. But what about qualitative assessment? And this is where MRI really changes the game for these patients. So using R2 relaxometry and doing sequential acquisitions at multiple TEs, we can assess iron dependent decay of signal intensity over time. So here's our six-year-old male with beta thalassemia. And you can see with progressive TE, there's six milliseconds, nine milliseconds, 15, and then 18, that liver signal gets darker and darker and darker. In our practice, we use a specific commercialized technique called FAIRyScan, which is from Resonance Health that actually quantifies the amount of liver iron content that's present. This is what we get back in our six-year-old patient. All right, and you can see that we can actually get a quantitative 16.3 milligrams per gram of dry tissue liver iron concentration, which corresponds quite nicely. They give you this, where's my pointer? Very nice scale to compare to. And so this child, based on their liver iron concentration, is at very high risk of iron overload complications. And then we can do this over time. So this is this same patient, and you can see that we started liver iron quant at age two years, and we've done it yearly. And somewhere between age three and age four years, that quantification of iron went up quite significantly. They were started on chelation therapy. And now at their follow-up at six years, you can see that the iron quantification has come down appropriately. And this is really, it's this step where we are making the biggest impact on these patients, is making sure that they're getting to chelation therapy at the appropriate time. So what are the challenges with doing this in your day-to-day practice? Well, there's multiple MRI techniques available, and while liver iron concentration is definitely recommended, there are no consensus statements on standard of practice. We also know that there's variability in the way these different techniques reported out. So the biggest challenge for these patients is when they're at one institution doing it one way, and they move to another institution doing it another way, and how do we make those two numbers make sense? But the reason I'm presenting FAIRyScan here is by far, it's the most widely used, at least in the United States, and has demonstrated the most reliability over a wide range of LIC numbers, but it requires a service contract, so it's a little expensive. It's a send-out, it takes two to four days to get your report back. And there are some relatively long MRI acquisition times, which can be challenging in younger children. So why does all of this matter? Well, I've alluded to it a little bit, okay? But MRI evaluation of any sort of quantification requires accessibility, affordability, and specific medical expertise. And we know that that is not necessarily universal throughout the globe. And it's starting to be recognized in the beta thalassemia community as one of the major unmet needs and challenges in this patient population. So if you actually look at the worldwide distribution of inherited anemias, you can see red is where we see the highest number of this patient population, and these are classically in places where there's not a lot of access to some of these more advanced techniques. And while we often talk about this as a global healthcare problem, I think at least those of us in the United States need to start really thinking about it. When we look at this graphic, which is from the US Bureau of Labor Statistics, and demonstrates employment rates of MRI techs by area throughout the United States from 2023. So you can imagine that if you have, by chance, a beta thalassemia patient who's living in Montana, their access to some of these very important techniques is highly limited. And we definitely saw this in our state of Georgia. Georgia is down here, and you can see this bright green blob, which is where Atlanta is. But we would routinely have patients driving four and five hours for their visit to us as one of the only MRI centers that was able to offer quantitative liver iron assessment. So it's one of the reasons why we need to continue to work on this. And one of the future directions that we're definitely seeing, specifically in iron quant is with spectral imaging with dual energy and or photon counting CT. There has been some very preliminary work looking at iron quantification using this technique. In fact, there were multiple really phenomenal presentations in the scientific sessions just here at RSNA. So this is something that's definitely in the future for us. All right, moving on to our second case. This is very much in keeping with some of the things that we've learned from our first two speakers. So this is probably the second most common thing that I see in my practice, which is a 12-year-old male with obesity who is being initially evaluated in our Pediatric Adolescent and Obesity Clinic. And their question to me is specifically, can you quantify how much fat is in this child's liver? So all of you have seen images like this, right? We've all seen the ultrasound images of hepatic steatosis, and we know that the liver is brighter and more echogenic. And so qualitatively, all of us in this room would agree that this child does have liver fat present, okay? There are some older quantitative evaluations that we can do with ultrasound. One is called the hepatorenal index, which is a direct ratio of the echogenicity between the liver and the renal cortex. When you're doing hepatorenal index, you're going to draw regions of interest in the liver and in the renal parenchyma. You want them specifically to be at a similar depth from the skin so that you're not having to deal with any technical issues with depth of penetration, et cetera, and it's truly just a direct comparison of pixel brightness. This has only moderate diagnostic performance for detection of liver fat. It's really not very good at quantifying it. It's actually much better for us to just be able to say definitively, yes, this is fatty. And we have some threshold values of 1.5 to 1.75 where we can say with much more reliability, yes, this is truly hepatic steatosis, and these children should go on to further evaluation. Where MRI comes in, again, qualitative evaluation using chemical shift imaging. Here, you're going to have signal loss on opposed phase images, as we see in our little 12-year-old with obesity. But then quantitative evaluation in MRI for liver fat fraction has really made a big difference in taking care of these patients for our clinical colleagues. So spectroscopy is considered the most accurate non-invasive technique for detecting liver fat quantities as low as 0.5%. It's not affected by iron deposition, fibrosis, or any of the other coexisting liver pathologies. The other technique that is in use is proton density fat fraction, or PDFF. So if you haven't made the connection yet, there's PDFF, there's USFF, and there's soon to be CTFF. They're all just fat fraction techniques. I'm not going to go into the physics of all of this, but the advantage of proton density fat fraction is it allows mapping of the whole liver, so you get to see a holistic fat quantification. It's been well-validated against liver biopsy and MR spectroscopy as reference standards. It has excellent reproducibility at different magnet strengths. You can do it at 1.5, you can do it at three. And at least here in the United States, is FDA approved for this exact indication. And because of that, has become available for all major MRI vendors for the specific purpose of tissue fat quantification. So at Children's Healthcare of Atlanta, we were a Siemens site. So what I'm going to show you are cases from the Siemens with their liver lab protocol, right? This provides you with both PDFF and spectroscopy. So how I do this when I'm sitting at the workstation, using the liver lab technique from Siemens, there's always the first quick look that I want you to be doing, okay? So you're going to get your in and your out of phases and here you can qualitatively say, absolutely there's fat deposition here. And then one of the other quick looks that I do is I actually look at the Dixon water sequences and here you can see this is the segmented liver volume that the computer is going to use for all of its quantitation. So I wanna make sure that it's in fact including the liver, that it hasn't left some of the liver out or included some of the other organs. And then as well as for the spectroscopy, I'm always going to double check the ROI position because you want your ROI, it's a little bit like Goldilocks, it needs to be in the exact right spot, not too close to the liver dome, not at the edge of the liver, it can't be over any of the portal vessels, it really needs to be nicely seated in the middle of the liver parenchyma. And then this is what we get. So here you can see the color gradient that comes up with the Siemens Liver Lab and you can see where your patient falls on the segmentation and where that ROI fat fraction falls so that you can see, oh, well, it's definitely not in this zero percent, it's definitely not in this super red percent, so we're somewhere in between. And then we also can report out these percentages of fat signal fraction, which can then be compared to the semi-quantitative scale from zero to three for steatosis, right? So our patient is sitting somewhere around 2.7% for their ROI, but 7.2% for the entire liver. And so then we can say, this patient is probably gonna be somewhere in here in terms of the degree of steatosis, probably about a stage one. And where this is really important for our clinicians is that this 5% or greater fat fraction is important for them in making the diagnosis of NAFLD or what is now called MASLD. So what are the challenges with doing liver fat quantification this way? Well, there's a lot of tech QI, QA that's involved. There's actually a lot of training that your techs need to do. They have to check those liver segments for you. They have to put the ROI in place. They have to know how to troubleshoot these cases. We also have to have very accurate ordering and protocoling practices by our clinical providers and radiologists. This is where really knowing what they're looking for is helpful. We can't just have them say elevated AST, ALT. They really need to ask us for fat quantification in order for us to do it. And also patient tolerance, right? This is a relatively long sequence that the patients have to sit through. So why is this important? Why does this matter? Well, as we've heard, there's a significant increase in obesity occurring in our pediatric population that's been called by experts the most rapid population scale shift in human phenotype ever to occur. So what that means is we are definitely seeing a rise in pediatric obesity. And for those of you who are practicing in the United States and you haven't seen this, there's a new AAP clinical practice guideline that was published in 2023 that has some new standards for our practicing pediatricians in terms of what counts as overweight, obesity or severe obesity, as well as some new screening guidelines. And I'm pointing them out here. So the AAP is now recommending that all children who meet these criteria at age their 10 year well child visit undergo standardized screening for dyslipidemia with a serum lipid panel, screening for NAFLD, MASLD and screening for type two diabetes with a serum hemoglobin A1C. Now this is a very long and a very lengthy document. So I went through it for you and I found the appendix where they actually talk about liver fat quantification. And they're listening to us. They're 100% listening to us. So liver biopsy is the gold standard, but then they say, oh, we don't really wanna do that because it's ouchy and kids don't wanna get poked. So, but ultrasound alone isn't going to do this. So they are actually recommending that any child who is considered overweight at the age of 10, undergo ultrasound with transient elastography, that's the fibroscreen and or MRI imaging with spectroscopy. So as we are continuing to practice our pediatric radiology here in the United States and probably also in Europe and other centers, we are going to start seeing more and more of these patients coming through our doors, which means we need to be prepared for it. So the Affordable Care Act in the United States covers general healthcare and prevention services that includes obesity services, okay? But we need to start thinking about not only access to these imaging techniques, but also cost effectiveness. As more and more of these patients, especially those from poverty areas and low income areas are going to be coming through our doors, how are we going to be able to take care of them and give them the information that they need? That's a high quality, but also cost effective. So my time is very rapidly coming to an end, which means I'm not going to get a chance to go through much of Fontan, other to say, and I actually don't really need to because we heard a lot about Fontan-associated liver disease. But the last thing I'm going to do is to show you my slide on why I think understanding elastography, especially in the setting of Fontan-associated liver disease is growing in importance. And that's because there's a growing number of patients, again, with Fontan-associated liver disease. The estimates right now in the United States is that we have around 70,000 patients that have undergone the Fontan procedure and are alive today. Their average age is somewhere around 23 years. But everywhere here in the United States, around 900 Fontan operations are performed. And these patients now have a 97% survival rate, which means that this patient population is about to boom. And these are going to be patients who are coming to us with their liver disease. We don't have any comprehensive studies to clearly define best practice for how to manage and monitor these patients. Radiologists are often at the center of the comprehensive care for these patients. And maybe the first to suggest their Fontan-associated liver disease and that they might actually benefit from liver imaging. And the transition to adult care, as our patients become older, we need to work in collaboration with our adult colleagues because what we know is that 20 to 75% of patients experience a break in their regular follow-up after transfer to adult care and are at risk of developing severe disease simply because they get lost in the system. So hopefully, at least those two that I was able to get through provide some good real life examples and representative cases of the pediatric liver disease that I've been seeing in my practice. We've discussed the challenges and also discussed developing need for continued advances in liver imaging and why we need to continue to strive to do better. Thank you very much. Thank you.

pediatric liver diseases

imaging techniques

ultrasound

MRI

shear wave elastography

MR elastography

liver fibrosis

artificial intelligence

radiomics

standardized protocols