Hello and welcome everyone from wherever you are tuning in from around the world to this webinar on Realizing Port and Counting CT's Full Potential. My name is Minmaan Singh. I am the Global Product Marketing Director for CT at GE Healthcare and I am going to be the session moderator today. Let me introduce you to today's speakers and panel. Professor Alan Luzziani from University Hospital Henri Mondeur, France, who will be talking on state-of-the-art today, where he'll cover spectral CT oncology and deep learning image reconstruction. Then we will have Assistant Professor Amir Pormonte-Zahra from Emory University School of Medicine and who will be speaking on PCCT as an emerging technology. And also we will be joined by Dr. Gianluca Fontone from Centro Cardiologica Monsino, University of Milan, Italy, who will be part of our panel discussion. We will be having a 15 minutes Q&A in the end of the session, so I would request the audience to put their questions in the Q&A section. So let's get started with our first session and welcome Professor Luzziani, followed by Professor Amir. Well hello and thank you for this very kind invitation to focus on spectral imaging and deep learning construction, seeing how already those two innovations actually empower CT. My name is Alain Luzziani, I'm Professor of Radiology at Henri Mondeur University Hospital in Créteil, France. These are my disclosures. So if we focus on the actual CT challenges that we are facing today, I think there are four main challenges that we have to handle today. The first one is of course x-ray dose reduction. The second one is to have a rational use of iodinated contrast agents. But despite those two reductions, we'll still have to focus on producing the best native contrast between pathologic and normal tissue. And last but not least, we have to integrate that in the workflow. So actually that's where spectral CT actually kicks in, because if we see how spectral CT today already provides innovation, we will see that spectral CT actually provides analysis based on the attenuation spectra observed at two energy levels on the opposite to polychromatic spectra, so that the actual material analysis that we are providing with spectral CT actually better reflects atomic numbers. And those high atomic numbers, we can be especially sensitive to them. And of course, that's the case of iodine, because we optimize the photoelectric effect with spectral CT. So based on that, spectral CT actually has promises where we could tackle some of the challenges that we mentioned in the first slide. If we look at some technology reminders, and we will focus on the rapid switching dual energy CT provided by General Electric, this is a single source, single detector technology. There's a fast KV switching between 80 KVP and 140 KVP, which means that the projection data are quite twice at each angulation with a high temporal resolution of 50 microseconds, which means that we can actually have a material decomposition in the projection domain and not in the image domain. And this has, of course, huge capabilities in terms of image reconstructions. Last but not least, we have both a large longitudinal coverage with a high CT of 80 millimeters in the high-end systems with a full field of view of 50 centimeters. Now, those are the capacities of the rapid switching dual energy CT technology. And the new paradigm today of those spectral CT technologies is that it actually provides what I call CT sequences, and that's actually quite close to what we do with MRI. With MRI, we are used to having different sequences. Well, that's basically the same with spectral CT. We have a lot of monochromatic matching, especially with low KV images, and we will see how that provides high contrast for tumor detection, for pathologies detection. This is a very important part of spectral CT. But we can also create new contrast with spectral CT, especially because of material decomposition. The best known material decomposition is the iodine over water decomposition, which means that those images actually reflect the quantity or the relative quantity of iodine that's present in each voxel. But we will see how spectral CT today already provides even more contrast than we actually can use in clinical practice. So fact-checking, let's see if we can tackle some of the challenges that I mentioned in the first slide. First challenging is, can we optimize contrast? Can we optimize tumor detection? Now, this is the image provided by your single acquisition arterial phase acquisition on the Revolution CT APEX platform with deep learning reconstructions. And this is a patient with cirrhosis. And what we're actually looking at is four different reconstructions, monochromatic energy level reconstructions of the same image. And as you can see, the lower we can get in terms of kAV, the higher contrast we have. And if you look at the 40 kAV image, we can see how we can better depict that hyper-enhancing hepatocellular carcinoma, which is located in the posterior part of the right lobe, which better strikes out than if we were looking at 70 kAV image. So we optimize the detection of hypervascular contrast, hepatocellular carcinoma. So reducing kAV increases contrast. Okay, we can tackle optimizing contrast in tumor detection. Now, how does that stand out in clinical practice? Well, here's just an example to illustrate that. This is a 68-year-old male patient facing mucinous colon carcinoma. And this is the unexpected encounter of a cystic mass of the pancreas. If you look at the acquisition, this is a single phase acquisition, polyvenous phase acquisition, still in the Revolution Apex GSI platform with true fidelity high parameters, so deep learning reconstructions. And what you do see is that the cystic mass of the pancreas does have some areas with hyper-enhancement, which are better seen both at low kAV image reconstruction and on the iodine over water map. And if you look at those areas of hyper-enhancement, they actually pretty much very much match both high uptake on 18-FPG PET MRI as well as restricted diffusion. So this is not just a simple cystic mass, this is not a simple cyst, and this was actually confirmed to be a cystic metastasis of the mucinous colon carcinoma within the pancreas. A rare tumor, I acknowledge, but how spectral CT empowered the diagnosis of that cystic mass, which was actually confirmed to be malignant. Second challenge that we can maybe tackle is how spectral CT can actually allow us to reduce the dose of iodine that we have to inject for CT scans. The question of reducing a contrast agent is of course a challenge for all of our patients, but we have to focus on whether or not we're able to achieve the same contrast with a reduced iodine dose. And not only on the arterial phase, but also on pulmonary phase or on the late phase. And this is what we try to challenge using the Revolution CT APEC platform. We actually made a study where we compared two types of acquisition. The first one was a polychromatic CT acquisition with a standard iodine dose, with the standard dose which is recommended by French national authorities for optimal oncology examinations. That's 525 milligrams iodine per kilograms of patient weight. And we compare the acquisitions on arterial phase, pulmonary phase, and the late phase to a second protocol designed specifically for CT Revolution APEC platform, where we reduce those by one third. So in those patients, we injected only 350 milligrams of iodine per kilograms. And also the four phases, unenhanced, late arterial phase, pulmonary phase, and delayed phase. What did our study show? That we had an improved enhancement of the liver using 50 kV reconstructions with the reduced iodine contrast dose as compared to the polychromatic acquisition with standard dose. But not only had we enhanced arterial enhancement and pulmonary phase enhancement and delayed phase enhancement of the liver, but we also showed that we could better see arterial phase hyper-enhancement of hepatocellular carcinoma, washout in pulmonary phase, and even washout in the delayed phase. So actually, even or despite the reduction of contrast agent, because we were using spectral CT with deep learning reconstruction, we had a better visualization of this arterial phase hyper-enhancement and better visualization of non-peripheral washout. Second challenge is also met with spectral CT. Third challenge, can we create new contrasts? Now those, I think, new contrasts, and what I mentioned as the sequences, the CT sequences, they actually can truly help in clinical practice. This is an example, again, of a patient acquired in the Revolution Apex platform with deep learning reconstruction. This is a patient with colorectal carcinoma undergoing chemotherapy. And as you can see, that patient develops the induced by the chemotherapy. What you already see is that by reducing the KAV, you actually, and combining that with deep learning reconstruction, you reduce the noise, and you enhance the way you can see the tumors. But you can actually, with spectral CT today, create new contrast. And this is the GSI fat map. That's a map where each voxel is represented by the relative proportion of fats compared to the normal liver. And what you can see is that those liver metastases, you can very much better depict on this GSI map. Look at this small metastasis, centrally located in the liver. Of course, as posteriorly, we can see that metastasis, but it's just so easy to see it on this GSI fat map. So yes, we can create new contrast. And yes, that means that we can probably optimize the way we handle patients with spectral CT today. And this is an example of how we can improve our diagnosis. This is a 28-year-old female patient with sacrocellular disease and abdominal pain. So standard monophasic acquisition, revolution CT apex platform, deep learning reconstruction. This is a 50-KAV reconstruction. And there's a mass within the liver. It's not a just a cystic mass. And if we look at the way the tumor seems to behave on different mono-KAV level, that tumor enhances. And this is the confirmed by the iodine over water map. There is iodine within this mass. It's not just a cystic mass. But can we go further for the diagnosis? Well, look at the fat GSI map. Now that map shows that not only is there iodine within the tumor, but there's fat within the tumor. And this leads us to make a hypothesis that this could be a primary tumor containing fat. And one of the first diagnoses is, of course, steatotic hepatocellular adenoma, which was, of course, confirmed by a liver MRI. Let me just see, just show you on this slide the behavior of those different contents. Look at the lesion that is here on the line with the different components at different mono-KAV level. It doesn't behave as the normal tissue. And of course, because it contains fat. And that tumor was confirmed on the MRI to actually be a steatotic hepatocellular adenoma. Again, going further in the diagnosis of liver tumors using the capacity of spectral CT today. Last but not least, I mentioned all my acquisitions on our reconstructive deep learning. And I think really deep learning is a game changer, as we presented in the last RSNA 2021, where we actually use a phantom, a custom-made phantom, which looks like a liver with the right part of the liver showing pulmonary phase enhancement of the liver with hypo-enhancing lesions such as liver metastasis, while the left part of the liver actually mimics arterially enhanced liver with hyper-enhancing liver tumors. And what we looked at was how deep learning constructions could help improve the image quality. What we demonstrated that the best contrast nose ratio was obtained with deep learning high reconstructions, which were actually similar to 100% AZV reconstructions. But of course, with a totally different texture. And that actually can be understood by a higher detectably index, which is provided by deep learning reconstructions. So combining deep learning reconstruction with spectral CT improves the way we can detect tumors, improves contrast nose ratio. And those two combined together are actually a game changer. What about in clinical practice? Well, this is the kind of improvement that we can see. This is a standard acquisition of an oncology patient. And you can see how deep learning constructions actually changes the aspect of the images, the texture of the noise. So it's not just about just reducing the noise. It's about also changing the texture of the noise, which actually better matches the radiologist's expectations in terms of tumor detections. So to finish that up, some images of clinical practice, just mentioning how, like at my institution, just spectral imaging is always on and how we can improve our clinical practice. This is an example of a patient, 56-year-old male patient, renal cell carcinoma, which has been resected. Well, I think if you look at that acquisition, very low KV reconstruction obtained on the Revolution APEX platform with true fidelity high reconstructions. All of you detected those two enhancing nodules, which are actually peritoneal implants of the renal cell carcinoma, which are hyper-enhancing on this RTL phase. What's interesting is that when you look to a polychromatic CT, which was performed outside our institution prior to that CT scan, of course you can see those peritoneal implants, but you will acknowledge that they are less easily detected, especially the posteriorly located nodule on the peritoneal implant close to the liver surface than they are using spectral CT with deep learning reconstructions. Another example, this is a patient with cholestasis, 34-year-old male patient referred for CT, standard monophasic pulmonary venous phase enhanced acquisition. Look at the virtual enhanced images and the low KV images. What do you see? Well, you see that the content within the gallbladder is not homogeneous. And why is it not homogeneous? Because there's probably stones within this gallbladder, but it's not just calcified stones, of course, it's cholesterol stones, and those were confirmed on MRI. So reducing the KV also enables to better depict contrasts between standard liquid and cholesterol. So you can also improve the way you can detect some pathologies, especially in emergency situations. Last example I will show, this is a 74-year-old patient who was referred by the emergency department for abdominal pain. Again, standard monophasic pulmonary venous acquisition and revolution CT with true fidelity high. Look at the pulmonary venous phase, there's something within the stomach, all of you seen. What's interesting is that when we activate the virtual unenhanced acquisitions, well, that thing seems to disappear. So what is it? Well, this is iodine, and this is iodine within the stomach. And what actually the patient had was gastric ulcer with active hemorrhage, which was not mentioned by the emergency physician who referred the patient to the CT scan. But that was detected on the CT, and the patient was brought to gastroenterology who confirmed the gastric ulcer with active hemorrhage. So to wrap this up, spectral imaging protocol optimization is underway so we can still limit the x-ray dose provided that we use deep learning reconstruction. I hope I convinced that we can today already reduce contrast 18 dose. I also hope I convinced you that with spectral CT today we can already create new contrast. What I mentioned is sequences, CT sequences. With the concept of sequences we can improve in some situations how we can tackle some tight diagnosis. And of course with the workflow my answer is always systematic use of spectral imaging so that we can always benefit from the advances of spectral CT data. And with that I would like to wish you all a good clinical practice using spectral CT with deep learning reconstructions. Thank you very much for your attention. Hi everyone. Today I'd like to talk about the state-of-the-art and future of photon counting CT. My name is Amir Pomerteza. I'm from Emory Georgia Tech. I'll start with the basics and you've probably heard this now that photon counting technology has become more mainstream and people are talking about it. I'd like to just bring up the idea of energy integrating detectors which are the conventional detectors used in almost all of the commercial CT scanners right now. And it's very simple. They operate by this indirect transformation of x-ray photons first into light through a series of crystals and then to electric pulse. What happens because of this indirect transformation is that the effects of energy and number of photons are both combines into one number which we call intensity. And if you look at this bottom figure you can see you could have like five low energy photons or three high energy photons and they could give you the exact same intensity value. So this message of energy and number of photons is garbled. And on top of that because we're doing this electric pulse detection and we're measuring the height of these pulses there's always electronic noise. And something else that happens is that because we have this indirect transformation to light first each detector pixel has to be optically insulated and this optical insulation has a certain width. So we could make the detectors very small but the optical septa will always have a certain portion of this small detector pixels so geometric efficiency goes down by pixel size. On the other hand in photon counting detectors and I won't go through the details of each detector there are multiple there are many detector types. One thing they have in common is that they transform x-ray photons into electric pulses directly. And what happens there is that we can measure the number of photons what we call count hence photon counting and their energy separately. In this case I'm showing you an example with four energy bits. So what happens if you look at this bottom figure is that you'd be able to detect single photons so each photon creates one electric pulse. So we can count the number of photons we can have three or four pulses we're never going to have three and a half pulses. So something big happens there is no electronic noise anymore in the count signal of photon counting. That is very important and we'll talk about it more. And because there is no optical insulation because we don't have any this intermediate light transformation exam step we can make the pixels very small. I won't go through the details of that but take my word for it geometric efficiency will not be effective at least for the count signal. And so two things happen we can measure these counts of course our counters have to be very fast to be able to detect time each individual incident photon and that's why it's taken some time for photon counting to hit the market because people have been working on these fast counters. Anyways this is the basic comparison between the two detector technologies and these differences bring about some very fascinating advantages for photon counting. I divide them usually into these three groups what I call spectral information or spectral advantages. It gives us color and contrast it's because we can see we can measure the height of those pulses that I just showed you and the height of the pulse means the energy of the photon so we can detect color or energy of these photons which gives us more contrast. And I told you we can make the detectors very small so the spatial resolution of the systems can be very high. We have examples and lots of papers at 150 micron 200 micron about a quarter to a fifth of the resolution of current CT scanners. And of course we talked about also noise properties we talked about electronic noise because electronic noise is not as problematic in photon counting it also brings about a lot of noise and radiation dose advantages. All right so spectral imaging with photon counting CT can be divided into multiple applications. I'll give you some examples here and there but it's a new field a lot of what we want to talk about is part of the future of photon counting CT not the state of the art. The first one the most exciting one is k-edge and multi-contrast imaging because we have more than two energy bins we can do material decomposition and for example we can separate iodine based contrast agents from gadolinium based contrast agents. These are the two FDA approved contrast agents that can be injected. There's also a vast line of research that talks about gold nanoparticles and selective contrast agents people are working on those lots of my colleagues are are active in that field. I've also seen examples of using k-edge imaging to using k-edge imaging to detect the platinum that exists in some of coronary stents and there was a very nice paper on that a while ago so we can do a lot with k-edge and multi-contrast imaging. I don't want to get into the details of that it's a fascinating field. Another advantage is because we're using all the spectral information is that we'd be able to do better metal artifact corrections or spectral corrections like cupping corrections in general and this would be a great help in image guidance surgery when there are surgical tools in the field of view most of which are made of metal. There is also applications I like to bring up in in our discussion session for example in oncology there is oftentimes we do inject once image twice or image multiple times for some type of applications for example in multi multi-phasing or biphasic imaging with photon counting we'll be able to inject twice with different timings and then image once and this could bring about lots of exciting applications for example in cardiac imaging we could have co-registered maps of delayed myocardial enhancement and coronary evaluation and in oncology we could have arterial and venous phase imaging of lesions acquired simultaneously. Here's an example that we acquired a while ago so in the top left you see a single energy CT and this is a movie of an animal that was injected first with gadolinium and we started imaging every two seconds for two minutes so you can see on the top right you can see the wash in and wash out kinetics of gadolinium contrast agent in terms of millimolar so it's very quantitative and different parts of this image and so after two minutes we did a rest and then we injected iodine so gadolinium by this point is at delayed enhancement stage and iodine is starting in first pass and then move through different stages of contract contrast kinetics. Bottom left you can see an early linear not noise corrected image but what's important here is that we can measure signal from iodine and gadolinium without the two interacting or interfering with each other. Lots of cool applications like this are in the future of photon counting CT. But one of the most exciting applications of photon counting CT is ultra high resolution imaging and it's very easy to show improvement in resolution with ultra high resolution CT it's like moving from an HD TV to a 4k TV you get smaller pixels so you can see better details. Here are two examples on the top you see conventional half a millimeter pixel size of a tumor and a stent in the bottom part you see ultra high resolution images of the same it's very easy to show the difference better resolution is easy to show so I will not talk about that a lot. However what I want to talk about is that high resolution acquisitions improve image quality even when you reconstruct your data at regular resolution and this is due to anti-aliasing because your sampling frequencies are better so you don't have as much aliasing and we've shown in papers that that could bring about up to 20 percent lower noise and this is again for free in terms of radiation you're acquiring your data the same way the patient is being irradiated the same way however your your sorry your acquisition is going to be ultra high resolution but your reconstruction is regular and that brings about 20 percent radiation dose 20 percent noise reduction which amounts to about 40-45 percent radiation dose reduction that is huge. Now of course if you acquire your data at ultra high resolution and reconstruct it at ultra high resolution you're going to have more noise because there is more energies there are more spatial frequencies involved so you see more noise but you can't just talk about noise as one number you have to look at the texture or noise power spectrum which is beyond the scope of this talk. And then again I want to talk about something else we have these better cameras these better detectors that can see tiny details and we know here's an MR example here's an MR cine image of my heart my heart is beating we know the heart beats it's very challenging to image the heart but that's not what I want to focus on I want to focus on these abdominal organs down here you can see those are also pulsating and this pulsation is not something that we cared about before I'm showing you another example of measuring how much my pancreas is moving during a cardiac cycle this is of course a little bit exaggerated but you can see the details and here's how much it moves in the in the axial plane my pancreas displaces about a millimeter during a cardiac cycle and now imagine that we want to acquire ultra high resolution images of my pancreas here's one pixel size of photon counting CT here's how much my pancreas moves so I believe that an ultra high resolution imaging submillimeter motion is the main challenge biological motions that could go unnoticed in standard resolution cannot be ignored at higher spatial resolutions we're not talking about respiratory motion that's usually not a big issue in CT scan times are very fast we're talking about cardiac contractions that directly or indirectly cause motion in the abdomen for example and there's an active area of research right now looking for new gating protocols that may be necessary even when we're imaging non-cardiac organs like the pancreas now of course I want to talk briefly about the future where we can mix ultra high resolution and multi-contrast imaging here's a cool example that would that would fit into oncology so we had this study back in the day where we had microspheres or embolization beads some of them had iodine contrast agents in them and some of them had bismuth and we decided they were different sizes the bismuth ones were a little bit larger so here I'm showing an example of an animal study where we injected the right kidney with bismuth beads only the left kidney with iodine only and then we had a bunch of liver samples were which they were first embolized with bismuth and then iodine and on the top on the all the way to the right you can see our calibration vials iodine is red bismuth is blue so let's look at a couple of these examples on the left you can see the non-spectral image this is an ultra high resolution back in the day it was 250 microns now the resolution has improved and all of the images I showed you in this study are all reconstructed with filtered back projection of course when you use iterative reconstruction model-based reconstruction and AI-based reconstruction all of these will be improved but here's an example you can see tiny beads of gadolinium here no iodine and in the other kidney we can see a lot of iodine and no bismuth and iodine beads were smaller so we can see smaller details here these would be the exact same beads that are used clinically to embolize tumors and of course once we look at the liver you can see both bismuth beads and the distal parts of the of the vessels and iodine beads that were injected later so we're showing multi-contrast ultra high resolution imaging with a clear application in oncology just to recap I'd like to talk about the advantages of photon counting CT spectral information is very important and we can do material and tissue decomposition non-spectral information is actually very very important in day-to-day practice because we can get better soft tissue contrast I didn't go through the examples but you can look up the papers we've shown green matter white matter differentiability improves by up to 30 percent and in calcium scoring we've shown that we can reduce the radiation by up to 75 percent and we can correctly be imparting a metal artifact a lot better with ultra high resolution acquisition we can do that at lower radiation costs and we can reduce the noise by 20 percent without doing anything by just acquiring the data at high resolution but it is very important to do motion correction even in non-cardiac cases and the lack of electronic noise brings about better household unit stability and also lower radiation dose in low dose scans such as screening tubes and the and the best part of all is that we can have all of the above simultaneously thank you first of all thanks to the speaker for this excellent talk with tons of insight and thanks everyone for taking time out for this discussion now I would like to welcome everyone for this panel discussion around the topic of realizing photon counting CTs full potential with professor Luciani, Dr. Pantone and Professor Ami a warm welcome to the panel so let's start this discussion so let's start this discussion I'll start with professor Luciani you showcase some great outcomes and clinical benefits from existing state-of-the-art in the spectral space what are some of your expectations from this upcoming new technology and how will you define the success of this new technology oh thanks and it's a tough question that the as we know CT scan is I think the cornerstone of many diagnostic and actually interventional situations so I think one of the one of the main challenge is of success will be that CT scan is actually used on a routine basis and still answers as many questions as we now anticipate a CT scan can actually respond to that's the first thing the second thing of the main success of maybe additional technology is whether or not we can answer additional questions, whether or not we can provide new tissue analysis, and maybe we could go further in making a spectral analysis that maybe goes beyond the mere use of two materials, but why not more materials that could be very important, especially in oncology setting. So with photon counting and multi-contrast imaging, because we have more than two energy bins, more than two energies, we have four, six, eight, depending on the vendor, depending on the type of detector, we can separate multiple materials at the same time. So at least we'd be able to inject iodine and gadolinium, and we can time them well, so that one of these two contrast agents, for example, is in delayed enhancement phase for cardiac, and then the other one is used for coronary imaging. Similarly, for oncology, you could have, for example, these embolization beads that are filled with iodine, so they're enhancing, you can inject those, and then you want to assess the blood flow, the residual blood flow, and you can use another contrast agent, like gadolinium again, and I focus on iodine and gadolinium because I think those are the two available, readily available and usable contrast agents right now. So I want to ask you guys, both as an oncologist and a cardiologist, how valuable do you think it's going to be to have these co-registered simultaneous maps of coronary and delayed enhancement, let's say for cardiac and arterial or venous phases, let's say for oncology, how valuable is it to have those two things as opposed to doing two separate schemes? Would someone want to switch to photon counting because of that capability? Dr. Pantani, do you want to start with cardiac? Yes, this is very important, not for the combined evaluation of the perfusion and coronary artery imaging, but for the combination of scar detection, I mean, late enhancement and coronary artery imaging, because to be honest with the iodine contrast agent, actually the image of coronary artery, of course, but also myocardial perfusion is quite robust, still challenging the late enhancement with iodine contrast agent, because for example, for that aim, we need to inject high volume of iodine contrast agent. So if we combine iodine contrast agent with gadolinium, for example, I believe that the greatest advantage is to evaluate the scar with using the property of gadolinium rather than with the iodine concentration. This is very important because actually there are a lot of clinical scenario which this combined information are required. For example, thinking about the EP procedure, electrophysiology procedure of ablation imaging guide, there is a trend to avoid any kind of electronatomical mapping than just to use imaging as guide for all. And this is a one of the specific setting in which this kind of promising application could give a lot of interesting result. And for oncology applications, there's also some, I think, very nice insights. Actually, it's opening Pandora's box here, because if we focus on diagnostic strategies in oncology, that of course opens up to the field of new contrast agents that could actually focus on a specific item that you want to detect on the tumor. And I think that opens up the idea where CT scan is not just a diagnostic tool, but also a prognostic tool assessing the different components of a tumor. Now, whether or not that would be using different contrast agents, that's still an open question, I think. And I'm not a user of floating counting, so I cannot answer precisely. And of course, there's the issue of safety of combining different contrast agents. I think that has to be addressed as well. I see one domain where the actual question that you ask would fit very well, and that's interventional radiology. So maybe it's surprising to say, should we put a photon counting CT scan in an interventional room? But the actual situation that you question is a question that we have to focus on when we deal with chemoembolization. And if that helps to treat a patient or to optimize a patient on site, instead of putting the patient out, making an evaluation, and then whether or not you should redo the chemoembolization, that could be actual benefit. So probably the idea will be that CT scan and photon counting could be embedded in interventional rooms, and maybe that's a new subject. Actually, my follow-up question for this, we are talking about a lot of this advanced technology in terms of looking into different materials and things like that. What may be the right way to do implementation of PCCT in clinical practice? And this question probably is for the whole panel here, and we can have this discussion together. It's a tough question, right? It's a new technology, it's barely hit the market, and we're talking about how to best use it. I think it's too early to tell. I have some experience in using it. I think two of the main challenges are going to be data representation. The good thing about photon counting is that a lot of things that we want to have and have to think before we do the scans are already there. We don't have to think about acquiring the data in dual energy or more energies. It's already there. It's the choice of the radiologist post-scan to use them or not use them. So I think one of the main challenges is going to be, okay, we're used to seeing a CT image maybe in different views or different slice thicknesses, but now you're dealing with material maps of different kinds. Let's say you have a calcium material map, iodine material map, gadolinium material map, soft tissue, or, you know, sky's the limit. There's so many other things like calcium removed, iodine removed, all of these things that we're used to in dual energy or we're getting used to in dual energy. There's going to be an explosion of information just from one single scan, and it's going to be very tricky or very, you know, it's fun to think about it, but it's going to be a big challenge to see how to present this data to the radiologist so that reading scan time, you know, average reading scan, reading time for radiologists doesn't change. I think that's one of the biggest challenges. Yeah, I completely, I think that's something which we really need to address well as a new technology comes in. I think these are some valid questions that we have in front of us. Now, I think I would like to ask Dr. Fontone here. You have been using the state-of-the-art system today, revolution effects, and basically, you know, it can do a lot in terms of the, you know, high resolution scans, perfusion imaging, accommodating challenging patients, especially in the cardiac space. Where do you think PCCT will help? And what are some of the opportunities in front of us beyond what we can do today with the state-of-the-art systems? Yes, as you mentioned before, the portfolio of cardiac CT has increased a lot in the last years, and actually, we have moved from the pure evaluation of coronary artery stenosis to the evaluation of a lot of additional information, mainly myocardial perfusion, but also quantitative plaque analysis, and a lot of details about the atherosclerotic burden that is allowed thanks to the introduction of the high-resolution approach. One of the limitation of the standard technology, anyway, is still the post-processing analysis of all this information that requires still a lot of human intervention and human activity in the post-processing. With this new technology, probably the scenario, thanks to the improvement of the image quality and the characteristics and the spatial resolution of the images allow to have a very clean data set that are more suitable, for example, for the post-processing of the AI solution that allow to make more automatic as possible the analysis of this data set that require the very minimal human interaction. This is important because we need to remember that the main job of cardiac CT in general is to rule out the presence of disease, stenosis or myocardial perfusion defect, whatever you want. And if we consider that the right case mix is composed mainly by patient who has not the disease, we can refer a lot of this patient to automatic post-processing to rule out and just to focus the effort of the medical doctor just in the small subgroup of patient in which to rule in. If we improve the image quality, thanks to the new technology, probably we can refer the majority of the post-processing to automatic process and just in this way to reduce our effort to involve human activity in the analysis of this huge amount of data. Yeah, I think the new technology that the data question comes in as well. And I think this can be a good question for Professor Amir. If you want to comment on data piece, how big is this and what are some of the challenges or how are we looking at this data management? I want to second Dr. Fontona's comments is that we're starting with this data set, with this large amount of data set and all these data should be AI ready. They should be machine learning ready. The outputs of the scanners are becoming more and more uniform across different vendors thanks to virtual monoenergetic images and using dual energy. I think photon counting with detectors that have more than two energies, I think six, eight, four, doesn't matter, it's more than two, will make that possibility, make it possible for cross vendors or even same vendor, different locations, different scanners of the same vendor to have similar image qualities. And that's a must if you want to have a reliable machine learning algorithm and we need to have it. I think it's going to be impossible for a radiologist to spend like four or five hours on a case because if you want to read all the images, it's going to take forever. So it's exciting because there's a lot of information, but I'm happy that we're at an age that machine learning is picking up and a lot of the things that we have to do manually or have to take a radiologist very valuable time can be pre-processed or post-processed by machine learning. If I can just pop in here, just I totally concur with all those comments, of course. And that also is a challenge for archiving because if we focus on oncology and I think, and cardiac would be probably the same, well, you have to deal with comparison and comparison comparing the same things and probably having the different contrasts, reconstructions available at any time. So we have to challenge PAX vendors and all server vendors so that the data is still available and easily and very rapidly available so that we can put it in routine practice. And that's also a challenge. And I totally concur with the AI ready design because that's the way to go to anticipate the data that we cannot manage on our own. Yeah, that's a great insight. And I think these are things which probably should be taken into consideration when these developments are happening. So really thanks for those insight. And I think building on that piece where in your current clinical practice, you may have different scanners with different capabilities. How do you see this new technology fits into your workflow or in general? Yes, I can comment for first. The setting, my hospital is an art institute so fully dedicated for cardiac disease indication. And for this reason, as you know, probably the field of cardiology is the field in which the technology counts more in terms of impact in your clinical activity. This is, for example, the reason why my hospital, the policy is to have all scanner with same level of technology in order to avoid different performance in cardiac application based on the scanner that you are using for the test for the exam. Because the impact to do something with the 64 slices scanner, for example, or last generation scanner in cardiology is terrific. And we cannot allow that there is one exam of high quality and one exam that is a low quality in the same institute. This is the reason why the approach is maximum technology for all scanner avoiding the scanner of first level and scanner of second level. I can comment maybe on spectral applications today at my institution. And I totally concur with what has been done and what's been said by Dr. Pantone because actually when you put a patient at my institution, we have actually three CT scans. So when you put a patient follow up for oncology, which is my practice, you have to make sure that the patient, if he or she is not on the same machine, benefits from the same technology. If not, you compare apples and carrots. So basically with my experience in spectral CT, all my CT devices had to have spectral CT so that whatever device would the patient be placed in, they would have the same type of quality of examinations. And that goes back to your first question, Mac. If it's going to be a success, it's going to be available everywhere. If it's only on one machine so that only a scarce number of patients can access this technology, this will not be a success. So we have to think together of how we can make that achievable. And that of course brings the questions also of cost and economic, I would say productivity of the system. How can we handle devices that are high-end system that cost more than standard system? And how can we show that they benefit overall to the care of the patients? Absolutely. I think that's where the real understanding is needed that when we come up with this new technology or when we're talking about this new technology, it should not just be incremental shift from what you're already doing versus it should really show some great benefits. So I really agree with your comments here. So I think this was a great discussion. There are many such rising questions like this and we would continue to have this discussion, I would say. But I completely agree with all of the panelists here in terms of their comments and really thank you for giving such insightful comments in terms of making sure that there is a good understanding versus what the current state of the art can do versus what the expectation of a PCCD and some of its challenges. So with this, I would like to thank all of you for this very engaging and insightful discussion. And I hope our viewers enjoyed this webinar. We will now open the session for Q&A and with all the panelists here. So thank you very much and appreciate your time. Thank you. Welcome to the Q&A session. We already have quite a few questions lined up, so let's get started. The first question, I think this is for Professor Luziani. This is more about what you have highlighted in your presentation on FAT imaging. So the question is, what are the protocols that you're using and how have you implemented that in your department? Yes, thank you, Mac, for the question and thank you for the audience. Pretty simple answer. I mean, as long as you switch on the GSI acquisition mode, then GSI FAT maps are right away accessible through the GE server or the GE application. And that's what one of the reasons GSI has to be always on, is that you never know before you start whether or not you're going to be needing those specific tissue maps and especially the FAT map. And you saw in the examples how that can be useful, both for quantification of steatosis and quantification or detection of liver lesions within steatotic liver. So GSI on, and then the map is right away accessible just as the iodine map. It's just a decomposition, material decomposition map. There's actually a follow-up question I would say, and probably this can go out to both Professor Amir and Professor Luciani. How will it help you if you had improved spatial resolution with spectral detail? So mostly talking about the material separation. Yeah. Now, of course, combining both would be just extraordinary for oncology and for all of us who are involved in oncology evaluation and the evaluation response to treatment. We know how the boundaries of the tumors are, where the information should be provided, where we have this invasion, microvascular invasion, where we have response to therapy, which probably should be more detected in those regions. And this is the regions where today we could have most gain in the use of both spatial resolution together with the contrast that's provided by spectral CT today and tomorrow by photon counting. But Professor Amir, please. I would like to second that. It's absolutely correct. There's a lot of information in the edges of tumors, as you mentioned. And right now, in almost all technologies, you have to pick between using ultra high resolution if it exists, it only exists in very few scanners, or spectral imaging. And even when you use spectral imaging, you have to use larger voxels because of signal to noise issues. So it'd be amazing to be able to see examples, you know, I showed you some results from prototype scanners, but it would be amazing to see some of these in real life, where we can actually quantify the amount of iodine or perfusion on the border of a tumor at high resolution in the order of 100, 250 microns. So that would be definitely important. With the caveat, and I touched on that briefly in my presentation as well, that motion is going to be an important issue. We tend to forget about motion in the abdomen, everybody knows about cardiac motion. But when you get to these high resolutions, you'll have to do some new corrections to make sure your images are crisp. Yeah, in fact, there was a question in the Q&A section that what are those motion correction challenges with PCCT, especially when there is high spatial resolution more than doubled? Yes, so I can give you an example. Of course, we know about cardiac challenges of motion in the cardiac imaging field. But I would like to talk about an abdominal organ like the pancreas. So the pancreas, and I showed some examples of my own, my pancreas, is that it moves up and down as a diaphragm moves up and down because of the cardiac motion. It also, the pancreas also sits right on top of the aorta, so it also moves in and out as the aorta is pulsating. And then when the blood pressure waves, systolic blood pressure waves gets to the pancreas, it starts blooming or expanding and slowly. So you have this triphasic motion of the pancreas, and the delay between these three phases of motion depends on the age, size, and height of a patient. It's an open question. That's something that I'm actively doing in my lab right now. So we need to know all of those things in order to get the crispest, best images of the pancreas. And it's one of the organs where there's a lot of half a millimeter and quarter millimeter questions that would take the patient to surgery or no surgery. So it's new. I think we're at the stage that we know we need to look into how to correct these motions. I haven't seen anything yet, you know, because these scanners are not widespread right now. We hope that, you know, within the next year when we see more people use these scanners that these questions will arise and we are actively looking into correcting this motion. At first stage, we're trying to measure the amount of that motion. Thanks, Prof. Amir. I think this is a specific cardiac question asking about how can you use the extra resolution in cardiac imaging? And probably this is for Dr. Pantone, if you want to address that. The issue of spatial resolution is one of the most challenging situations for cardiac imaging, basically because we have two settings, spatial resolution is the other thing. And I'm talking about classified lesion and evaluation of intrastent restenosis. With the actual technology, we find this setting very challenging. We have a lot of solutions to improve the performance in these two clinical scenarios. But of course, if we move in the direction of micrometric spatial resolution, probably these two challenging situations are completely overcame, are completely resolved. So I believe that that is the direction for cardiac imaging and is a quite mandatory direction to resolve the actual problem that we have with the start-up technology. Thank you, Dr. Pantone. I think there's one very interesting question and logical question that says, is there an additional gain from PCD CTs, multi-energy application compared to dual energy? I think- Short answer is, of course, yes. Even when you're using it for, even when you're using photon counting CT for dual contrast, sorry, single contrast applications like iodine imaging, you have more measurements, you have more samples of the attenuation of iodine or attenuation of whatever object you have in the field of view. So by default, you have more measurements, so you'd be more robust to noise, you'd be more robust to some other artifacts that, you know, I don't want to go through the details. But yes, there is an advantage in having more of those. There's an advantage in doing metal artifact correction, beam hardening correction, just because you have a better estimate of the spectral response of the object. And of course, on top of that, if you're using multi-contrast imaging, the advantage is clear. You need to have more than two energy bins to be able to recover more than two contrast agents. But it's going to be valuable. That's why I commented on using it in the IR or in oncology, where you're going to have a lot of metal surgical tools or other things in the field of view. You will see a clear advantage for photon counting. Thanks, Prof. Amir. There's a two-part question I would say, probably, Prof. Lutsiani, you can answer this. The first part is, can spectral sequence be used for tissue characterization and planning in oncology? The second part to that question is, how do you do the timing for double contrast with one scan? Probably, that's more of a future application. But Prof. Lutsiani, if you want to quickly take on the first one. Yeah. The first question and first answer, yes, indeed, that's very interesting. Today, as Prof. Amir said, we can mainly distinguish two materials. Distinguishing iodine from adjacent tissue, iodine from water, provides quantitative evaluation. I use that in my practice to especially detect quantitatively the fibrotic contents within tumors and distinguish that from necrosis and distinguish that from active tumor. That's the better way to distinguish those three components that we know consist of many tumors. Yes, this is true that this can be used. Is it as standardized as RESS's criteria? Of course not, because it's new and it has to be evaluated. Now, the question is, can we go further from that to maybe three materials? That would be very interesting. Probably, that connects to the potential of photon counting detectors, because then we can combine more than two materials and better assess directly within one image and maybe provide quantitative assessment within one image of those three components as a single one. I think that's the way to go. That's why I'm so enthusiastic in seeing the benefits of photon counting detectors in the future coming months and years. Thank you, Professor Luziani. Prohami, you want to comment on the second piece for me? Sure. It's an important question. I showed you an example of dual contrast injection. But when we go to specific applications, that's also an active area of my research right now and we're writing a grant on it, on what is the best sequence of injection of these two contrast agents, let's say iodine and gadolinium for cardiac imaging, and which one should you inject first and what should be the delay between the two? Those are all very interesting and important questions that we would love to figure out in the next year. Thanks, Prohami. There's a very interesting question here on, you know, it says, are the scanner going to be more green? Probably highlighting that it's going to be more environmentally friendly or so? Tough question. Let me just pop in. It's very difficult to assess the green aspect of that. And for me, it's very difficult because you also have to put in relationship all the other aspects of managing a patient. And it's not just by doing one CT. If you only think of that machine, okay, then we have to focus on the quantity of data that's provided. And indeed, spectral CT provides heavier load of data. So you could think, okay, that's less green than a CT scan that does not produce that much data. But if you have a better management of the patient, if you can reduce the numbers of other examinations that are done, well then, is that more green? And I think that's why I think it's very difficult to assess that. But there's one thing that already can be achieved is reduction of iodine contrast agent injection. And reducing the dose means also that you reduce the impact of providing iodine and taking iodine from the earth. And if we think of that, that's not only a reduction of risk for the patient, it's also a reduction in terms of wasting some unnecessary iodine contrast. Thanks, Professor Mitsiani. There's one more very interesting question here asking about what are some tasks in your practice that still may remain unaddressed even with potent counting CT? Probably Prof. Amir, you want to take that first and then we can have a discussion around both oncology and cardiology? That's a very important question, right? In my mind, there are still questions, there are still applications where even when we switch to photon counting, we will see some small improvements, but they will not solve the issue. An easy one that comes to my mind is the issue of iodine quantification in virtual non-contrast images. As long as we don't take advantage of the K edge of any of these contrast agents, we always are going to run into an issue when we're trying to separate iodine and calcium from each other, or iodine and calcium from the background. Photon counting may help with CNR, SNR, so we have better contrast between these materials, but it's not going to solve the inherent problem with iodine that its K edge resides in the lower end of the spectrum of diagnostic CT energy, so we're always going to run into that. That's why I think it's important to look into other contrast agents, and the one that's most readily available is gadolinium. So that's one of the things I don't think is going to go away anytime soon, unless we move on to other contrast agents. Thanks, Brock, for me. Dr. Fluzziani, Dr. D'Antonio, you want to comment something on that? In terms of, I was wondering whether we could also speak of native contrast, and in oncology, being able to assess the bone marrow is still a challenge for CT. You know how it's difficult to detect bone marrow invasions, small invasions, small metastases throughout the bone marrow, and how MRI outperforms CT scan today. Now the question is whether or not with photon counting CT, we could have this distinguish of three different materials, maybe fat, iodine, and calcium, and maybe that could help in detecting those issues, and then that would put CT, again, maybe at the same level of MRI for detecting those lesions. That's my guess, I would say, where one of the benefits in oncology could go. Yes, I fully agree with my colleagues, but specifically in the setting of cardiology, what is still missing in the clinical practice is the possibility to evaluate the histology of the tissue of the myocardium, because now there is a very common use of biopsy in a lot of cardiac disease after you have done a diagnosis with non-invasive imaging. So one of the most interesting questions for the future of photon counting is if with this excellent spatial resolution we have any chance to move our non-invasive imaging from an anatomical level to an histological level, that is the point, because there are a lot of settings in which this could be reasonable. We have, of course, to check the performance of this scanner on the ground with this specific count, but if the spatial resolution is in the range of micrometry, we are absolutely in the range of histology, and therefore probably a lot of potential information we can have from this new generation scanner. Thank you, Dr. Pantone. Thank you, Dr. Luciani, on answering that question. So in interest of time, we'll close it here. There are many other questions which have been listed, but we respect the time and close this session here now. I would like to thank everyone who is tuned in for this webinar, taking out their precious time and listening to our panelists and speakers. Thanks to the speakers, who I know Dr. Luciani and Dr. Pantone, you're quite late at your side, but thank you so much for making your time. And Professor Amey, thanks so much for a very insightful talk on the PCCD itself, and we hope that we'll see all of you again, and thank you again for everyone's attention. You're welcome. Have a great day. Thank you all. Good day. Bye.

Computed Tomography

Spectral Imaging

Photon Counting CT

Oncology

Cardiac Applications

AI and Machine Learning

Data Management

Precision Medicine

Radiation Dose Reduction

Diagnostic Capabilities