Super Resolution Fetal MRIs and their benefits including: enhanced visualization of fetal anatomy, potential for 3D and 4D reconstructions, improved diagnostic accuracy in complex congenital anomalies and highlighted the ongoing advancements and clinical impact of these techniques.
All right, we'll be uh respectful of everybody's time. My name is Jessica. I'm the practice manager for the Center for Fetal and Neonatal Care. We appreciate you guys joining us this morning and hopefully didn't have too much trouble getting logged in this morning. Um, I know that we're trying a new method and just really trying to get the word out and allow people to join and, and get um more information through our Broadcast Med. Um, all these videos will be, or all these talks will be recorded and then you can access them, um, via the broadcast web. Um, website and so by getting and logging in this way, then it allows you to have access to those videos and stuff too as well. And then all the other offerings that we have at Phoenix Children's. Um, so it's with great pleasure that I introduced Dr. Luis can call this. We are so blessed to have him as part of our team, and today he's really going to give us an update on all the modalities and, um, you know, the new technology that we're offering here at Phoenix Children's. So I'll let Doctor can call this, um, take it from here. We do again offer one CME for this course. The code is 48,200. I'll go ahead and put it in the chat as well as include the number that you can text it to. So, thank you, doctor can call this and thank you everybody for joining us this morning. Well, thank you so much, uh, Jessica. Uh, let me just make sure that the correct screen is shared. Yeah Yeah, we're good. Thank you. So, I really, really thank you so much, you know, for the opportunity of presenting this, this talk. I'm pretty passionate about, you know, this, this area. I think it's like a very interesting area in, in, in fetal fetal imaging, particularly fetal MRI that we have been exploring here at Phoenix Children's. So, hopefully, uh, you know, you enjoyed this as much as I, I tend to enjoy. I really like the, you know, the imaging aspect of it and uh. And uh without further ado, let's go on with the with this uh talk. So we're gonna talk about super resolution 3 to MRI, uh, problem solving 2 and complex congenital anomalies. Uh, I do have some disclosure that I, I give a a doctor course for for Phillips. Uh, the objectives of this presentation are to understand the principles of super resolution imaging and slice the volume reconstruction in fetaRI. Don't get to, uh, too bogged down with the names. We're gonna see how it works pretty quickly. Discuss the benefits of this technology, uh, you know, including enhanced visualization of it anatomy. Potential for 3D and 4D reconstructions. Improved diagnostic accuracy and complex congenital anomalies. And also highlight the ongoing advancements and share real life case studies to demonstrate the evolving clinical impact of these uh techniques. So I'm gonna begin a little bit with what standard, you know, fetal MRI for us uh at at Phoenix Children's. So if you look at regular sequences that we use to to examine the central nervous system. And I'm gonna use an example of a fetus that we saw pretty recently here, and I'm gonna show excellent images of the brain using T2 weighted, T1 weighted, uh, gradient echo, echo planner imaging or EPI and diffusion weighted sequences and which what each one of these tell us about the same lesion. So on T2 weighted images, which is a fluid uh sensitive sequence. Uh, the, this area here is actually, uh, where, where the lesion is. It, it was uh referred to us as a possible tumor in the brain, but the lesion is actually a vascular lesion located in the level of the tortula that is bright on T1, so we know that there are blood products in into this area. There's a blooming artifact that we see on the grad and echo sequence also, you know, highlighting the presence of blood products. So this is a trombosis through sinus m formation. And the DWI sequence or diffusion we sequence adds information by telling us that there are potential areas of ischemia in the cerebellum and in the temporal parietal area. Um, Another way that we can image the brain is to use the sequence that we we we use very frequently actually, particularly on the Tesla, which is called a balanced turbo e echo or BTFE. It has other names in in depending on the vendor, so it may be called steady state preprocession sequences or through FISP, but, you know, for purposes we're gonna be using this acronym to describe the same sequence. Which uh the difference is, uh, compared to a titu weighted sequence is that instead of blood being dark, blood is bright. So here we have light bright blood in the umbilical cord, we have fluid in the ventricles, and we have this thrombosis dal sinus malformation in the tortula now seen in the coronal plane, and remaining portions of the brain actually looks normal. The other thing I want to highlight is this this in this 4th image here is that this sequence is also helpful to evaluate bones. So here we see the nasal bone and the maxillary plates of the palate. They are intact in a coronal sequence, so that helps us, you know, evaluating uh a structures as well. We have optional sequences that we use, actually fairly fairly frequently, although not successful all the time, but in this particular case, where the patient was referred because of suspected genesis of the corpus callosum, we use diffusion tensor imaging to map the uh white matter tracts in the brain. So here you see the diffusion, uh. In in the uh rostrum of the corpus callosum and its splenium up the corpus callosum here and if we post process the images to obtain ractography, you see the tracts, you know, of the corpus callosum uh in their entirety, uh, in the midline, so we didn't have agens of the corpus callosum. And there are other tracks that can be seen with this technology, for example, the corticalspinal tracts that we see uh on each, on each side in in this particular case that was examined at 28 weeks. The same sequences can are obviously used in in in fetal body imaging, and here's a comparison between the T2 and the BTFE sequence, one in which the blood pool is dark, and the second one where the blood pool is bright. You will see, for example, pulmonary vessels a little better using this one, you know, when they're contrasted with the lung parenchyma. And in this one, since the blood is bright, the contrast between the myocardium and the blood pool is much better, so you see, you know, uh, cardiac structures better using this type of of sequence. Here is this a little lower in the abdomen, just highlighting the differences. The images look fairly similar, but they are different because Here the umbilical vein is bright on BTFE and is dark on on on T2. Um, these are additional sequences that we use. These are three dimensional, uh, fast T1 weighted images, uh, that we obtained to look, uh, particularly at the presence of and, and pattern of distribution of meconium in the colon on the coronal plane here, sagittal plane there and. This is uh another uh uh another uh utility of the WI sequence. So here is a baby that has, you know, kidneys on both sides, and we are interested in seeing, you know, is this, uh, uh, renal tissue or something else in this area, and when you use the diffusion weighted sequence, you see that this is actually a, a horseshoe kidney pretty uh confidently. Uh, finally, we, we also do cardiac imaging using a Doppler device to gauge the acquisitions. So the same BTFE sequence that you saw before when it is acquired using a gauging uh uh device like such as this gives you, you know, Cardiac motion. So you see the sepal effect here, the pulmonary artery and the, the, the ductalteriosis and the smaller aorta in this, in this image. Well, all of these are very good because, you know, they are fast sequences and uh the idea is that it will freeze motion, but motion keeps being a problem in in in in fetal imaging, particularly fetal MRI. So when we acquired these images, what's happening in the background is that we are acquiring a whole bunch of slices, you know, through the plane's corona axel or whatever plane we want to image. These are the voxels and the voxels are very interesting because you think of a voxel as, you know, uh, uh, similar dimensions in the X, Y, and Z axis, but in MRI the voxels are tend to be rectangular like this. So for example, if you, if you get a sequence of the brain that you acquire using 1 by 1 by 3 millimeter voxels. What's happening is this, uh, you get the implant resolution on the X and Y axis on the brain here and if you were to see the same image using multiplaer displays without any post processing you will see that the images, you know, reconstructed images are corrupted by motion artifacts, blurring, and so on. So you really cannot use this image as, you know, to do three dimensional imaging without further uh post processing. And this one is even, you know, highlights a little, a little more because the body tends to move more than the head so it's even harder to get, you know, good body imaging with motion. So in plane there's a little bit of blurriness and so on, but you still can see the anatomy but if you were to to to see this in in in the coronal and sagittal planes just from this acquisition, you will see that there's motion corrupting the sequence everywhere uh and uh and blurriness and and the images are really non-diagnostic, you know, when you look in using 3D. Uh, we have a research project going on with this particular sequence it's called uh KTSense, which is an accelerating, uh, acceleration method, uh, that really gets really fast images. For example, here you see the baby's heart and you see even a little bit of the maternal breathing motion when you acquired these images. These are done under uh research protocol and consent forms and so on and so forth. But if you were just to reconstruct this. Uh, without any emotion correction, you will get, uh, you know, blurry images as well. So this is one of the babies that we acquired using the KT sequence and uh Nick Robert, uh, who's our physicist and my partner in crime and all of these 3D reconstructions, really the brain behind all of this. Uh, he, he nicknamed this baby baby dolphin. And I was, you know, that prompted some curiosity on my part to say, OK, you know, if I had to scan a dolphin, uh, what we we have actually to do. So to scan this dolphin here, and this is a a a video that's available on the internet. Basically, the, you know, there's this arsenal of people. Taking the baby out of the water, immobilizing the, the, the dolphin, and then, you know, putting the dolphin into the scanner so it can be scanned without motion, and obviously we cannot do this with fetuses. So the alternative is to do post processing of the images and try to reconstruct them. So that comes to, you know, the, the, the, the, the technology behind the images that you're gonna see in the stock, which is called super resolution imaging. Which are basically algorithms to get 3 dimensional images from multiple stacks of 2 dimensional images. Uh, the super resolution algorithms typically have 3 pieces. The blurring motion correction uses a technique called slice to volume registration or SVR. So when we talk about SVR during the talk, that's what we're gonna be talking about, and then it rejects in the bed slices so it gets all the outliers and reject them to reconstruct the final volume. Uh, our pipeline that we use at Phoenix Children's, it comes from, uh, it's basically, uh, outlined here. You get the original stacks, sagittal axial and coronal. And then you see the corrupted planes in the original form. Then it goes to the super resolution the reconstruction uh uh pipeline. And then at the end you get isovoxil uh uh uh data data on the on the three orthogonal planes or any plane. So here's the axoplane, the chronoplane, and the sagittal plane after motion correction. And super resolution. So we basically derive everything that we use from this SVRTK kit that was developed by the uh biomedical informatics group from from King's College and this work was actually possible from uh Leadership Circle Foundation grant that we got in 2021 here at Phoenix Children's that really helped us, you know, propel this forward. And uh I cannot give the stock without giving full credit of this to uh Nick Robert, you know, Nick is really the brain behind you know all of this so we use uh we use uh uh we use uh a server that has uh uh very heavy computer graphics and everything that was granted to us by the the foundation, you know, to do this work. So this has generated a couple of papers and I'm not gonna delve too much into papers but uh so we we say OK, you know, if we're gonna reconstruct images, how many, how many slices we need to get until we get uh a good reconstruction of the brain for example. So we looked at a series of normal babies and babies with a series of congenital anomalies of the brain. And then we used 4 radiologists to score brain of areas of the brain in terms of anatomical clarity, you know, uh, included here the cerebral vermis, aqueduct, corpus callosum, septumussum sil and fissure. Uh, we use a liker scale to to go from 0 to 4, depending on the quality of the image that was obtained. And the bottom line of the paper is this that. Our ability to to to see good quality images increased as the number of stacks that we acquired increased and we use a minimum of 5 stacks to achieve, to achieve uh uh a good resolution. So let's say that we examine the brain, we tend to obtain at least 2 coronal, 2 axial, and 2 sagittal images, you know, to, to, to, to get reconstruction, sometimes more. So, real-time examples now in the fetal brain. So here you have a fetus that was examined at 27 weeks, and the fetus was referred for mild ventricular megaly. So you have the regional axial coronal and sagittal tituated sequences. Here is the reconstructed brain now you're using super resolution, same, uh same, same, same volum that the said seen on the axial plane and the coronal plane on the sagittal plane. So here you see details of the corpus callosum. You really see the aqueduct very well. You see the cerebellar vermis, you know, with lobulation, uh, the primary fissure is really well identified, you know, really, really good, uh, imaging with that. With this, we can, uh, we go from 3 millimeters thickness at the time of acquisition to a slight thickness of 0.7 millimeters, so it really uh significant increase in spatial resolution. And then we can segment the brain, uh, like, uh, showing here how it, how we actually do it and then you get reconstructed images. For example, here is the ventricular system. So here we have the lateral ventricles. This is the gave up softumilluidum. This is the 3rd ventricle here, and this is the aqueduct and going to the 4th ventricle over there. So you see in multiple projections. And then this is the, the ventricular system, with the, with the actual brain superimposed on that, the brain stem, cerebellum, and so on. So we can see that from multiple projections as well. And the next image is gonna show a surface rendered image of the brain and this is a 27 week fetus, and we know that at 27 weeks we should see the, the, the central sulcus and the precentral sulcus, and they're clearly seen on this image. This is the uh the insulin, the pseudum fissure, uh, and these are the images of the brain, uh, you know, reconstructed, uh. Using this technology. So this is a totally different brain. Obviously this baby doesn't have cleavage of the brain. It has a monoventricle, so this is clearly holous encephaly from the original images, no problem with that. There is a median, uh, median cleft, uh, over here. There's a, there's a cystic space in the back, but just for the sake of, of, of, of visualization, let's see how this brain looks, you know, using uh three dimensional reconstructions. So this is the coronal planes. The sagittal planes, this is the dorsal uh uh sac. And so after motion correction and super resolution, you get the dimensional volume, and after segmenting this brain, this is the appearance of the brain with holoroencephaly seen in the frontal, the left lateral projections, you see the very different civilian fissure over here, and there's some cleavage in the back, so this is semi lobar holorosencephaly. Now superimposing there you see the mono ventricle and there's a structure in green and the back on background that is actually that dorsal cyst, you know, that we see so it kind of gives us a pretty good idea of how this brain looks in the in the three dimensional space without having to do the reconstruction in our brains when we're looking at only cross sectional images. Now this is, I think, a, a, a more interesting case. So it's obviously a case with a very severe ventriculomegaly um seen on the axial plane here from the top of the head to the bottom. You see that the ventricul megaly is asymmetric. You see that there is a fox dividing the brain, so this is not holoroencephaly. There's a little bit of corpus callosum crossing the, the, the midline in the front. Then this is a third ventricle that is dilated and there's some, some, some mass effect from this cystic area into the cerebellum at the bottom of the brain. These are coronal images. The, the frontal lobe, and then as you go from front to back, you see that the hemispheres are being played apart by this, uh, this cyst, actually, that intermispheric cyst. And it becomes, you know, more pronounced as you go towards the back of the fetus. OK. And this is how this feet is seen on the sagittal view. Here you see the aqueduct, and there's a little bit of stenosis in the back of the aqueduct here. So there's an element of aqueductal stenosis. This is a dilated third ventricle. Uh, here you see the folks kind of outlined because of the difference in signal intensity compared to the to the 3rd ventricle and the, the, uh, that's that space over there. Now, these are images uh looked at after super resolution. So now we have clear triangulation of the planes, so sagittal, coronal, and axio and coronal through the cerebellum. A little, uh, a little close up in the stenosis of the aqueduct that you see here in the sagittal plane, in the coronal plane, and the axoplane. Uh, here showing the third ventricle in different planes. And here we are showing essentially that there's this genesis of the corpus callosum. Only this portion of the corpus callosum is formed. You see here crossing the midline in the axial plane, and this is how it looks in the coronal plane. Now, after rendition, this is how the brain looks like. So you see the the hemispheres do have gyrations, but they are really splayed apart by this very large inter hemispheric cyst. And then on this image here you see on the uh you know, seen from the top towards the bottom of the of of the brain you see a separation uh from the fox here you see the inter hemispheric cyst, you see the third ventricle in the background. So just showing how this image was actually reconstructed so the segmentation steps here were first to segment the brain stem, then we segmented the cerebellum, then. Segmented the hemispheres and see how the hemispheres relate to the to the brain stem and the cerebellum. Here is the superimposed third ventricle, then the lateral ventricles, and this is the large interhemispheric cyst. So this is a disorder known as asymmetric ventricular megaly with the interminiospheric cyst and the genesis of the corpus callosum or avid. So this is a a a disorder characterized by the smooth walled anechoic inter hemispheric cyst that displaces the adjacent brain, and the hallmark feature is asymmetric ventricular megaly which is often severe. So due to the type 1, A inhemispheric cyst in continuity with epsilateral ventricles, the affected ventricle with underlying cyst, usually at least 50% greater in diameter than the contralateral side. The cystic wall is often difficult to to visualize. There's usually absence of the cagon septum illustum and a callosal anomaly, which is either agenesis or disgenesis, leading to progressive hydrocephalus and macrocephaly. So these are comparative advantages from the literature, this was they were taken from uh Doctor Paula uh Woodward. So this is how it looks like in in in a case that that she uh published, and this is a diagram of the anomaly, and this is a specimen that I think correlates pretty well with what we saw with three dimensional reconstruction. Now moving into uh fetal body imaging. Um, there's a little less in fetal body imaging in terms of super resolution than you have actually for the brain and even for the heart. But this article here that's published by the group from King's College, uh, uh, in 2021. Uh, intended to summarize, you know, how were they using super resolution imaging and fetal body imaging with a couple of of examples. So For example, in 11 circumstance that we use supersolution all the time is congento diaphragmatic. And here you have the axial coronal and sagittal planes of a diaphragmatic hernia with the herniation of the stomach. And this is the 3D rendition showing the right uh lung, the left lung, the stomach, the top, the heart and the liver. So that segmentation kind of helps us estimate the lung, the volumes of the lungs and also to determine if the liver is herniated into the chest, how much of the liver is actually occupying the chest. This is another example of a baby with a teratoma, and then the rendition of the teratoma seeing green here. But the interesting thing is that in this rendition, you can also see how the teratoma displaces the tongue to the contralateral right side. And these are renditions of the of of of a fetus or two fetuses actually with a myelo meningoceal, so showing the The volume of the sack and and so on. So let's see, you know, how we have used this in some practical examples in, in, in, in, in our practice here to kind of like help us with cases. What we use in every case of diaphragmatic hernia, we use super resolution to get the lung volumes. So we basically get, uh, you know, isobloxyl volumes of the of the chest. Calculate the volumes for the right lung, the left lung, and calculate the total fetal lung volume. Based on the calculation of total fetal lung volume, we compare, uh, the, the, the volume of the lung to what's expected for gestational age, and there are two formulas that are used for that. The oldest formula is one, published by Francoise Ripens, uh, in, in, uh, uh, in, I don't know exact exactly when, but, uh, some time ago, and more recently by Marianna Meyers from Colorado Children's. They're kind of like very similar, uh, except that Mariano's uh graph has a little better distribution of, of fetuses in the early gestational ages compared to the one that was published by Doctor Reitens. And in general terms, And it's a little bit, you know, there's more granularity to that, but if the total fetal lung volumes are less than 25% is a is a is a a good predictor for mortality, uh, if it is less than 35% and the, the liver is herniated into the chest by 20%. Predicts mortality and or emo, and then less than 35%, uh, you know, but above these parameters like chronic lung disease. Uh, these are, uh, uh, this is work from Rodrigoano when he was at Texas Children's, uh, basically, uh, showing that the observed or expected to your lung volume has a good diagnostic performance compared to other methods to to predict mortality, and that is enhanced by adding the percentage of herniation of the liver into the chest. And a little less predictive for the need for ECMO compared to other imaging modalities, but uh also enhanced by by uh the percent of of of liver herniation to the chest. So, in this case here, which is a diaphragmatic hernia with the left lobal liver up, uh, the rendition of the right lung, left lung, liver down, and liver up in green. So This baby had a total fetal lung volume estimated as 13 to 15% of what's expected for age, and the liver was herniated 24%, therefore, like more than 20%, and the outcome of this, this, uh, this, uh, pregnancy was unfortunately, uh, neonatal fetal demise. This is another case where there is a little bit of liver herniated into the chest. And this is the rendition showing liver down, liver up. You notice that the left lung volume is much, you know, more generous in this case compared to the previous one. So it was estimated as 34.3%, so less than 35%, and the percent of the liver was 4% and the outcome of this fetus was intact survival with pulmonary hypertension. baby does have chronic lung disease but didn't need ACMO, you know, after delivery. So this is another totally different case seen by ultrasound as a cystic mass behind the left atrium, and these are the MRI images um on on axial T2, coral T2, and Saal T2, you'll see that there is a cyst in the middle mistinum. The relationships of the cyst with the other structures are not really, I mean, they're kind of clear, but not exactly clear where it's coming from. Uh, these are the super resolution images and here in the super resolution image you can see that the trachea and bronchi are better reconstructed. So you see the left mainstem bronchus, the right mainstem bronchus, you know that it's an arterial bronchus because the right upper lobe bronchus is coming from the, the, the right mainstem bronchus, and here we have this lobulate cyst that appears to be related to the, you know, to the bronchus. And this is the three dimensional reconstruction of the lungs with the trachea, the bronchi, and that bronchogenic cyst. Another case that we have probably a few cases of of of esophageiatricia that we have examined using this method. So this is a baby with obvious polyhydramus, you know, it has two cystic structures in the uh supremagestinum. Uh, one of them is the trachea, the one in the back is an esophageal pouch. So this is the esophagus seen uh distally as well. And really, really diminutive, uh, you know, stomach, uh, scene in this view. This is the esophageal pouch now seen in the coronal plane, the tracheum bronchi, and then you see here a distal, uh, distal esophagus. So, Most likely, uh, so you got 3 with a distal fistula. These are the sagittal images seen by MRI and also by ultrasound, where you could actually see the pouch, uh, fairly well. These are the reconstructed images, so the trachea and the bronchi, the upper esophageal pouch, this is the point of the esophageal atresia, and this is where the, the fistula is coming from. It's not coming from the trachea, but it's coming from the left main stem bronchus, and this was confirmed, you know, after delivery. Another similar case with polyhydramus, and you get here the trachea and esophageal pouch in the axial plane. The esophageal pouch in the coronal plane, a very kind of distended pharynx that has been reported in cases of esophageal regia. This is the trachea, there's the bronchi. Um, here, interesting, you see the nasopharynx, the oropharynx, the vocal cords here, the trachea, that's dilated pouch in the back. Seem better here and really you don't see anything towards the distal esophagus, you see this kind of like a little remnant of the esophagus seen uh here, not distended. And uh the two constructions were really interesting, so very different than the previous one. The, the portion of the esophagus uh that is the esophatria actually extends beyond the cria. The, the, the attritia is really below the crina here, and this is the continuity, you know, uh, with some fluid in the stomach. So this was uh esophage attriia without just a distal uh fistula that was also confirmed after birth. And then the last case of body imaging that I want to show is this case that was sent to us because of duodenorisia. Say, well, you know, why, why are we bothering about when you dorisia, but this case had had actually some peculiarities. So there's the severe polyhydrams, the bub double bubble sign, classic Adolinorisia, no problem here. Then when you look at this image, you start seeing that this is the bladder, but these structures that you see in the in the perineum here actually dilated bowel loops, and you don't know if these are like uh. Uh, large bowel, small bowel, what exactly is that? When you look at the perineum, you see a gluteal cleft, you see fat in the ischio anal fossa, but you really don't see a circular normal anal dimple here. So this baby has a perforated anus. Now, in the coronal plane, highlighting again the uh doinumattrigia. But There's a this dilated loop extended to the perineum over here that we see again. And then in this image, it kind of goes behind the bladder and beaks over here. So that's the point where this, this, uh, this loop of bowel is, is, uh, arthritic, OK. So what is this exactly? So for that, we use the uh 3D images. So the reconstruction shows here the duodenum attritia, the vertebral body is in the back. This is the bladder, and this is the loop of bowel, so it's a loop of large bowel that's actually ending here, and that's the arthritic segment. Maybe also had um a short uh uh. Sorry, lumbosacral spine with deficit of of sacral sacral elements also had here uh uh a spinal cord with a blunt terminal terminating really high, so the baby had associated um uh called the regression syndrome and a anorectal malformation. So, last, I'm gonna spend, you know, a few more minutes now talking about uh fetal cardiac imaging. Uh, and how she can help us with uh with uh with these cases. We decided to explore this a lot based on this paper that was published in The Lancet by David Lloyd and his colleagues also from the the group at King's College. In which they looked at three dimensional visualization of the fetal heart using prenatal MRI with motion corrected slice of volume registration. So this was a prospective single center study. And there are a couple of interesting things about this paper that I want to highlight. So they had 85 cases of congenital heart disease diagnosed by fetal echo. They were referred uh for fetal MR in cases where they thought 3G could actually complement the clinical information they had by by echocardiogram. All exams were performed at 1.5 Tesla and for hearts, the 1.5 Tesla works a little better because it's less prone to motion artifacts. Uh, notice that most of these fetuses were examined in the 3rd trimester, so 3rd trimester is actually better for fetal cardiac MR. 2nd trimester is really difficult to, to perform. And then All of these uh images were post-processed using slice to volume reconstruction. And then what I want to really uh emphasize about this paper is that they actually validated several aspects of the post-processing algorithm to see if they could be applied in clinical practice. So, they evaluated the reconstruction algorithm, how robust it was. They evaluated measurements, how they compared with fetal echo. They look for assessment of cardiovascular structures, so you know how many of the structures can you actually see and does the, the, does the, does the frequency of seeing uh structures increase with super resolution? And does it provide additional diagnostic information? So in order to look at the reconstruction pipeline. What they did is they get, they got uh images that were obtained actually from uh a child scanned under anesthesia after after birth, so the images were, were really not motion corrupted. They artificially corrupted images mimicking uh fetal movement uh uh from a uh a fetal case that they had at 3840 weeks, so you see here that these images got motion corrupted in all different planes and then they used this reconstruction pipeline to reproduce, to go back to the original volume. After, you know, all of them have been scrambled, and the output 3D volume was really uh uh very similar to the to the input volume data set. So these are comparative images from ultrasound and uh the fetal uh MR using super resolution uh reconstruction. So And then the measurements were compared and they were highly correlated, as you can see here, but there is a bias, a small bias actually minus 0.33 millimeters of smaller measurements done with MR compared to to ultrasound. In terms of visualization of structures like systemic veins, pulmonary arteries, pulmonary veins, aortic and ductal arch anatomy. The proportion of cases where these structures were seen. were higher when they looked at the super resolution reconstructed volumes compared to the original MRI images. Not only the visualization rate was higher, but the image quality was also higher when they looked with uh the reconstructed images. In terms of additional atomic findings, they found additional information in 10 of the 85 preuses compared to the original echo, and most of these, uh, additional information came from evaluation of the great vessels, uh, uh, and the arch and so on. So The right arch was found to have an aberrant left equation artery in 4 cases compared to echo, uh, retroortic illuminated vein in 3 cases compared to echo. The right arch was found to have a double aortic arch, not being a right arch only in two cases and bilateral SVCs in, in, in, in one case. These are original images showing here the double aortic arch. This is the reconstruction prenatally that shows the the entirety of the arch. This is the reconstruction from a cardiac CTA and in which case. Uh, the, uh, uh. This has been uh ligated. This is uh a case of a scimitar syndrome. Reconstructed, you know, using the technique compared to cardiac CTA postnatally. And this is a case where the pulmonary arteries are actually coming from the aorta, seen here and compared here with the uh the cardiac CTA postnatally. Finally, another case where you show a normal doctor's arteriosis in the hypoplastic transverse arch with coration of the of the aorta. So here are some examples from from from from our uh our institution. So the first one that I want to show is a patient that was referred at 26 weeks. She had a fetal MRI at 28, 28 weeks. Uh, the baby had dextrocardia with a non-obstructed supraccardiac TAPVR double that right ventricle, mtotrigia, malpo great arteries, hypoplastic pulmonary valve and artery, and the question of an absent ductal artery. So these are the uh the standard static BTFE by blood sequences without any gating. So here we have right and left in in every image. So here we have the thymus, we have the right SVC, we have the uh trachea. Uh, we have the esophagus in the back. Use the arch, the apex of the heart. And the uh interpeic IVC. After reconstruction, these are the multiplanar SSSCS FSC, sorry, 22 sequences. So Here we're using a minimum intensity projection, making the, the, the slices thicker, so we can see the entire vascular structures. And here what we're seeing is a confluence of the pulmonary veins. This is the vertical vein that drains to the left brachocephalic vein, and then the left bracephalic vein is gonna drain into the SVC, so a supra cardiac total anomalous when everything is returned. Here's the image of the trachea and the bronchi, and this is a very different arrangement of the trachea and bronchi than we have in normal cases. So in this case, both bronchi are right sided. They are aerial bronchi because the right mainstem bronchis originate in front of the main stem bronchi. In each case they're very symmetric. So this is a case of right isomerism. This is the stomach of the fetus. And this is the T1 rated image that I have shown you before that we used to look for the distribution of of of uh meconium in the colon you notice that in this case. Most of the large bowel loops are seen in the right side of the abdomen. There's nothing in the left side of the abdomen. So this looks like a a a bowel mal rotation. This is how they reconstructed 3D images from these data sets, so these are the original data sets that got, you know, reconstructed and then rendered to regenerate this image. And this is how the image will look like from all different angles, uh, as we see here, and we're gonna go through and take a look at every component of that. So what we're seeing in this image here. Is the confluence of the pulmonary veins, the vertical vein, the left subclavian draining to the SVC. So the total anomalous pulmonary venous return. This is the IVC draining into the right atrium. This is the right atrium here. We see a bridging liver, so both both sides of the liver are right sided. This is the stomach, this is a midline, uh, gallbladder, and this is the small bowel, which is all on the left side compared to the right that we saw on the right, uh, on the other image before. This is seen from the back, where we can see the bilateral right-sided bronchi. And this is the rendition of the blood pool in the heart where you see the venous return, systemic venous return here. Here we see this right ventricle is dilated, and then the double outlet right ventricle with the pulmonary artery here, the aorta, and then the doctor's arteriosis in green that you see in in this image. And just to uh. To show the mole rotated bowel. So here's the image of the left bow on the left side, the right bow on the right side, and then if we superimposed on each other, this gives you a better idea of the of the mole rotation. So these are 4 static images uh obtained with T2 weighted uh sequences. Now I mentioned to you before and I showed you some images of this uh research sequence called uh called KTES which is an acceleration method to the uh our our PTFE sequences so very fast. So there's another uh paper that was published in 2019, sorry for the typo at the bottom. Uh, by the same group, now exploring, you know, reconstruction of cardiac volumes using the bright blood pool. So with the bright blood pool, you actually see the heart better than the vessels. And also with motion. So for example, in this static image that is presented in the paper, what they're doing here is they're. Doing a cross section through the left and lateral ventricles, which are seen in the axopplane here, showing the same ventricles in the coronal plane, so right and left, and then Doing almost like an M mode type of image where they can actually see the diastole and cystole for the left ventricularar wall, and for the right ventricular wall as well. And then multiple other planes highlighting other structures. And interesting enough, you can also extract, you know, uh uh the phase encoded uh velocity phase encoded. Images from from these data sets and actually get flow information. We have done this in one case, but it uh we we don't have the the capability of doing that on the flight for every case. OK. How do these sequences look originally when we acquired them? So we acquired them in in 3 different planes. We get 9 stacks of these images when patients sign up for research by doing that. This is the axial image, the sagittal and kind of coronal image. You notice that the sequence is very fast, so even like the maternal motion, uh, breathing motion. Is taking into account here uh pulsations in the maternal vessels and then the pulsations of the heart and everything. You can see clearly this is a case of a hypoplastic left ventricle in this, in this image. So, after we obtained these images. The heart is masked, the structures of interest are masked, and these structures, and only these structures are gonna be reconstructed using the reconstruction pipeline. So, it goes to a process of uh reorienting the, the frames according to the phase of the cardiac cycle in which they were acquired. And then the blurring and then slice the volume reconstruction frame frame several iterative processes until you get to a to a final, uh, volume data set. So this is an example of what you can achieve and it's not achievable in all cases unfortunately, but this is reasonable, uh, to achieve in a in a good case is this kind of like images of the heart. So this is the sagittal images showing the ventricular septum here, septum there and there and then. The outflow tracks, you know, coming off the heart on the left side and the right side crisscrossing over each other. So we can fully navigate the volume, uh, after, after the fact, you know, after this has been acquired. Um, I'm gonna, uh, show now a case, a real case that was scanned at 36 weeks. was a 42 year old uh pregnancy. She had a late prenatal care with first appointment at 28 weeks, and the ultrasound performed uh by by the MFM showed that diffusal hypoplastic left heart chambers, a possibly overriding aorta, a question of atherotaxy. The echo showed different information like an LSVC to coronary sinus, a double outlet right ventricle, a VSD model hypoplastic mitral valves, left ventricle, thick pulmonary valves, uh, hypoplastic transvers aorta, and possible indication for the MRI was possible heterotaxy. So in terms of electrotaxi, uh. We look here with the standard coronal and axial titu weighted images and the reconstruction of the of the bowel, we see that the stoms on the left side, the liver is right sided. Um, nothing really, uh, suggests the eotaxy, uh, uh, here, and then, uh, really normal distribution of the, of the, uh, micornia in, in the colon. Now, these are images of the heart. So this is the coronal plane, the exoplane, and the sagittal plane, seen simultaneously. So there is a disproportion between the left and right sides of the heart. There's a, there's a tiny VSD scene here, at least in this plane. But when you look at. The left also tract, you know, in a three dimensional controlled, you know, uh uh environment. You can see, you see that the septum actually continues with the interior wall of the aorta. So, no, no evidence of a double outlet right ventricle. And this is the pulmonary artery coming from the right ventricle. So this is the pulmonary artery, part of the ductal arch seen here. Now this shows, you know, the full ductal arch. Here's the trachea that's also reconstructed as part of this because it's fluid filled. And this is the order that looked, you know, a little narrow, uh, right after the takeoff of the subclavian artery also increased distance between the takeoff of the left carotid artery and the and the left uh subclavian. Um, I think this shows really nicely the reconstruction of the vessels, so. What we're doing here is basically making a thicker slab. Uh, and going through that, you know, with the vessel superimposing to each other. So here you have the Venus return, the aorta and the, the, the ductal arch. So it does look like this aortic arch is somewhat narrow. And there's some distance between the takeoff of these two vessels and more narrow, uh, you know, just in the region of the Isthmus, kind of like supporting, uh, uh, the broad possibility of quatation of the aorta. This is seen from the 3 vessel and trachea view. And then it's shown here, but it's gonna show better in this one. we're gonna see the left SVC going to the coronary sinus coming over here, like there. There are many different ways that you can look at these images, and I probably spend more time on this than I should, but I try to see it in different multiple different ways. So this is uh the reconstruction of the blood pool and the trachea is showing here the the right SVC, the trachea, the aorta, you know, and that's the doctors coming on the background here. This is seen from the front, that uh pulmonary artery, you know, and the aorta over here. This is seen from the contralateral side. That's the last left SVC and the doctor's arteriosis, and this is seen from the top, so you can see the full extent of the arch with the takeoff of the left carotid and left subclavia and then narrowing of the of the Arctic arch and a little more narrowing of the doctors prior to joining the the the PDA and the scending thoracic aorta. So, in this case, I reconstructed again using like color coded uh uh images. So here you have the right SVC IVC, the right ventricle, the pulmonary artery, the ductal's arteriosis, this is the aorta, and this is the takeoff of these vessels. That's the region of the of the narrowing that we're suspecting prenatally and so on, but really no double outlet right ventrico, you know, in any of the images. So the findings were no eotaxy, a mild hypoplastic, uh, left ventricle, posterior malalignment TSD hypoplastic arch with possible partition, and a left SVC with no bridging vein. So these were correct outcomes. No ataxy, the VSDR was confirmed. The patient did have a mildly hypoplastic arch, but normalized after closure of the, of the PDA so she didn't need any, any further uh work after birth. Um, we're getting towards the end here. I just want to show this last case of the tray of a Punariatria. And uh we're just gonna go through the through the uh reconstructed images when we get to them. So here we have the uh The right ventricle, left ventricle, and the overriding aorta. And this is the reconstructed uh three dimensional image of that uh overriding VSD draining to the aorta here. In this case, the the fetus also had mappias that were uh seen actually well by ultrasound, one of them, but when we looked with them are, there were many more, you know, coming from the uh the aorta, you know, feeding, uh, the, the lungs. So The MRI has also helped us in, in, in, in these cases of pulmonary atresia to identify uh these major aortopulmonary collaterals that you see, you know, all over uh here. And this is the angiography uh after birth in the reconstruction, uh, reconstructed cardiac CTA. So, um, the outcome was, uh, theology follow atresia and the mapcus as shown before. And Just as take home points from, from, from, from this, uh. Basically what I try to show is that super resolution 3 MR using this technique called slice of volume reconstruction is capable of producing isovoxel 3D data sets with satisfactory spatial resolution and soft tissue contrast, not in all cases unfortunately, but you know, in, in a good proportion of cases we we we can get, you know, really good images using this this technology. Uh, image evaluation be performed using both multiplayer display navigation and 3D rendered images. Um, we do use it to obtain volumetric measurements of prognostic significance in congenital diaphragmatic hernia. And I hope that the cases presented here illustrate, you know, an adjunctive role of super resolution, fetal MR in complex anomalies involving the fetal brain, the chest, abdomen, spine, and cardiovascular system. So With that, you know, I would say thank you so much, you know, for the opportunity to present this work. I mean, it means a lot to me to be able to do that and uh I'll be happy to take any questions if if uh if you guys have questions. Thank you so much, Doctor Galves. We're excited to have you in this technology available to our families here at Phoenix. So, um, we do have one question in the chat. What is the fetal MRI role in the diagnosis of anal atresia? Oh, this is a good question. Um, In, in diagnosis of antrigia. So you can see the anal dimple both by ultrasound and MR and in fact, I don't think there's a single case of uh that we suspect uh on the ecttal malformation that we don't image with both ultrasound and MR. I like to see. Or confirm that I don't see the anus on on both, but what it adds is evaluation of the colon. So you can actually see where the colon uh terminates using the T1 weighted sequence. So you can actually see, uh, you know, the distal end of the anal atresia. So, and we're talking about here like the less complex cases, but if you, if you push the envelope further and you're trying to image, for example, coco anomalies, then the then the advantage is really, really high because Unfortunately, the, the rectum, the vagina, and so on, they are all uh pelvic organs. They're protected by bones. So, Every time you have a structural protected by bone, particularly when you're imaging the 3rd trimester, MRI tends to do, uh, you know, a fairly good job at outlining those structures a little better than we see with ultrasound. So I would say that I think there is a definitive role of of fetal MR in evaluation of, you know, any baby suspected of having an anorectal malformation in whatever form they are. Great. If you have questions, you can unmute, or you can put them in the chat. Uh, good, good morning. Good morning, Doctor Gunelves. This is Doctor Garbaik. Thank you so much for a nice presentation. I'm sitting here wondering if uh there is uh any type of series in looking at the development of the fetal brain in monody twins, where the uh where the donor twin passes and the uh recipient twin suffers a significant Anemic reaction because of the change in the circulation. And then subsequent MRI looking at the brain of the surviving baby, and how it correlates to whether or not there's going to be any neurologic uh damage anticipated. I, I, I, I think the answer is yes. I think there are, you know, papers that talk about uh imaging, in fact, It is used in in centers that do laser uh photo photocoagulation and so on to look at, you know, uh, potential damage to the brain like schemic changes or, or, or, or hemorrhage that can happen, uh in, in babies with between between transfusion now. I don't know if the question also, um. Has a component of a role of super resolution imaging into that like for example reconstructing the brains and following up uh you know, development of uh of volumes, for example, that will be the extra information that you would get with that. So, for that particular aspect, I don't know of, of anything, you know, uh. That I know of, but I cannot say that I did a literature search to to uh. To to answer this question with 100% yeah, the, the reason that I'm asking is, unfortunately we just had a situation where uh at 19 weeks, the uh the donor twin passed. Mom didn't want any type of radio frequency ablation done on that twin. And when that happened, baby B did suffer a pretty significant anemic result. And we're planning if baby, uh, if the, uh, recipient baby survives, we're planning on getting a uh MRI done about 3 weeks from the time this happened. Just put the baby at about 22 weeks. Is that, uh, is that a reasonable time to do it, or should we wait a little bit longer? No, I think, I think it's reasonable, like 3 weeks, I think you should be able to see. Um, I think, as a, as a rule of thumb. If you can image the fetus in the 3rd trimester, you will get a better MRI study. So Even, you know, for brain anomalies that you can examine after birth, today, you can get excellent imaging in the 3rd trimester, uh, and I would say after 32 weeks, for example, where this location is developed and so on. But in these circumstances where you need an answer earlier, I think it's fairly reasonable to examine earlier, and I think, you know, the interval 3 weeks that you're suggesting is fairly appropriate. OK, thank you. Hey Louise, this is Crystal Blade. Can you hear me? Hello? uh Yeah. OK, great. Hey, this is great job, Louis. This is Chris Lindbla really great presentation, really enjoyed seeing those last images. I hadn't seen those of the uh the MR uh and just the work that you've continued to push the envelope here at PCH so nice work. Um, I'm just, my question is, um, You know, there, there are things that we know from a fetal cardiology standpoint that MRIs and adjuncts such as assessing lung volume in the presence of um certain cardiac defects like Epstein's anomaly or phantasia. Um, you, you showed great examples of assessments of the great vessels. Can you, can you comment on future state of what you think is coming down the pipeline as far as um uh MRI adjunct for fetal cardiac imaging, um, and, and where, where things are going in your uh fetal uh cardiac MRI community. OK, so I think The points that you raise are are are just excellent. I think we are primed now for evaluation of, for example, pulary funjectation. I think we can do it very confidently. Um, I think that, you know, evaluation at a taxi, I think we can do very confidently, uh, no problem. Um, I think evaluation of the of of vessels can also be done fairly confidently if we do it in the 3rd trimester. Uh, in terms of where it is, I think is going, I think there are publications coming up looking at the morphology of the arch and the Doctor's arteriosus, you know, using center line assessments, a little more sophisticated, uh, statistical analysis to try to predict a little better which which fetuses that we suspect have, uh, you know, uh, an abnormal arch end up having pooration. So I think this is probably going to evolve and I think the. The limiting factor here is basically technology to to accelerate, you know, the transition between the images you acquire to the the final product of the reconstructed image and and and and analysis, right? I think that's the hardest part actually because everything that I showed here is kind of like artsy, right? We're kind of like. Working a lot on these images to actually get to the point that that they that they get. And I think A lot I think it's going to to to to happen in function probably and uh and uh and we talk about this a lot, you know, collaboration between uh imaging and cardiology uh of course cardiology uh has really you know the the is is is really the cardiologists are really the leaders in understanding cardiac physiology and function. So, if we can put our heads together and, you know, learn from those who are doing this, you know, for, for a long period of time, I think that's the frontier that we, we, we should be able to explore. Uh, but I'm not there yet. I hope I can get there or we can get there at some point in time. Yeah, definitely just seeing the progress you continue to make here as a center is, is awesome. So, uh, I can't wait to talk to you offline more about those other images that you've been doing. That's great. Congrats. All right. Thank you. Well, thank you guys again for joining us this morning to be respectful of everybody's time. I just wanna again let you know the CME code is 48,200. We are excited to um have our next conference on March 13th. Um, Doctor Elkwad from our neonatology team is gonna be joining us just to let us know the um updates and growth within um neonatology and care of the infants here at Phoenix Children's. Um, and we also have another great offering on Monday, um, in regards to, um, Our BNI team who will be offering a conference in regards to the heart of the matter, current understanding of neurological injury and congenital heart disease. We will send that out via our DG um distribution group and so if you're interested in joining, we do offer that. Um, this coming Monday. So, um, what time? That one's, yeah, that one's on, um, again, this Monday 7 o'clock and it is offered via Zoom, so we'll get that flyer sent out to everybody. So let me just say something. So I, I invited some friends from Brazil to join this conference. They are in aligned, so you know, I know that we offer this conference to our community here, but these conferences are so good, you know, maybe we should consider, you know, spending, you know, the how arises in inviting other people to, to, to join. I think, uh, you know, a conference like this on, on, on brain injury related to to consciental heart disease is so like, uh. You're not gonna find this everywhere and it will be a really a shame to limit to only, you know, our circle here. Yeah, just an idea, you know. Absolutely. We definitely have expanded, um, more of our distribution, but what I'll do is I'll take the list um of participants today and happy to forward that on to who all participated in the conference today, and then please um feel free to forward it on to people that you might think are interested. OK. Yeah, thank you. All right, everybody, have a great day. Thank you again for joining.
