I'm an assistant professor of medicine and faculty of the nuclear cardiology area from the Yale University School of Medicine. I will be presenting to you today how to perform a PET myocardial blood flow evaluation as part of the PET curriculum. These are my disclosures. So the learning objectives of this topic is to describe patient preparation, camera setup, and quality control of 2D and 3D PET systems, describe how to acquire and process a cardiac PET perfusion study with myocardial blood flow information, learn procedures to optimize myocardial blood flow measurements, and detail information for important myocardial blood flow measurements as well. As we know, the integration of myocardial blood flow indices quantification to PET myocardial perfusion imaging has been established as an important diagnostic and prognostic tool. And with the advances in PET instrumentation, the ability to accurately evaluate for the presence and extent of coronary artery disease, diagnosed microvascular dysfunction really relies on high quality imaging acquisition interpretation. That's why it is very important to discuss this. These are very general slides, kind of a summary that may have been discussed in other topics. This is the very famous schematic, really showing or representing cardiac PET and inspect radio tracers. As we can see here, O15 water demonstrates close to linear uptake, whereas the linear extraction of technetium-99m tracers plateaued approximately two mLs per gram of myocardium. And then we can see here the other more common PET radio tracers like rubidium-82 and N13 ammonia kind of fall in the middle. This is a schematic showing some of the cardiac PET imaging protocols, and I think it's important to mention that every institution might have a slightly different presentation or approach, but this is very general. We have here in the top part, we have the rubidium, this one we have N13 ammonia, and this one we have the O15 water. As you can see at a glance, they're very similar, they probably vary more in terms of the timing that it takes to complete the study. In general, the regular imaging protocol for PET starts with a scout scan, which can be a computer tomography image or a low-dose radionuclide image-based, and that's just to localize the heart position. Then that is followed by the transmission scan, again, it may be a low-dose CT or radionuclide imaging, and then that's followed by the emission scan that starts with the bolus injection of the radionuclide. Then you go ahead, you obtain the resting images, you again may do another CT, you follow by the pharmacological stress, and then injection of the radionuclide to obtain the stress imaging. The following tables, they're a little busy, but the point of both of these tables is really to remember that there are three different types of crystals that are commonly employed in PET. Those include the VGO, the LSO systems, and the GSO that is in the next page. Really, what I just wanted to bring to your attention is that these different systems, depending on whether they're a 2D scanner or a 3D scanner, may have different dosing. Overall, the protocol is usually very similar. As you can see, the dosing here in the GSO systems that use 3D system is 20 millicuries. Just to remember that the focus of this talk is going to be on myocardial blood flow quantification. That is given in MLs per minute per gram of myocardium at rest and stress. What we talk about myocardial flow reserve is really the ratio of the blood flow at stress over rest, just so that we have those concepts pretty clear. I'm going to walk you through obtaining a PET myocardial perfusion imaging test and also quantifying myocardial blood flow, performing quality control and assurance to the point of interpretation and reporting. Important considerations in terms of the acquisition and reconstruction for a PET myocardial perfusion imaging test is patient positioning. Generally, we recommend supine with arms out of the camera's field of view. Images can also be obtained with the patient's arm resting at either side if the patient's unable to raise them, but we have to be very careful and report this at some point during the description of the study because this can generate artifacts as well. In terms of the dosing considerations, avoid standing in close proximity to a generator or the patient during injection. And in terms of PET, lead is really not useful to shield you from the radioactivity. We remember that larger patients will require higher doses, though 3D images will require less dosage than those obtained through 2D. So this is a picture of a typical positioning of a patient, again, supine with the arms up. They usually will have a blood pressure cup connected that allows us to monitor their blood pressure and hemodynamics during the test. And this is the usual setup of a rubidium generator. Now while we talk about acquisition and processing, we recommend LISMOD acquisition and LISMOD refers to event-by-event acquisition. It allows for higher efficiency of data storage, higher temporal resolution, and also higher flexibility in timing and reconstruction to quantify blood flow, perfusion, function through ECG-gated images, and also to be able to perform QA of breathing artifacts. It also enables reconstruction of static, gated, and dynamic datasets. They're now available with nearly all cameras, which again will enable simultaneous dynamic and ECG-gated acquisition. So this is basically what LISMOD acquisition is. With this type of acquisition, the scanner coincidences are continuously recorded along with the information about the time after the start of the acquisition and the electrocardiographic signal and the signal about the breathing motion. Data can then be resampled in multiple formats at any time of the acquisition. So here we have panel A for static, B for dynamic, C for ECG-gated, and D for respiratory gating. And just note that for A, C, and D, we have to consider the pre-scan delay. So for example, in this one, we are reconstructing high-count static images by summing all the information after a predefined pre-scan delay that I mentioned here. It's usually a delay time after the tracer injection. Then the dynamic sequence are obtained by serial temporal sampling at different times after injection. So here the time after injection is basically time zero, and this is what allows for tracer kinetic analysis. And then for the electrocardiographically gated images are obtained at multiple phases of the cardiac cycle to assess for ventricular function. And lastly, the respiratory gated images can be obtained at different phases of the breathing cycle to correct for respiratory motion. So kind of summarizing what the process for image acquisition and reconstruction is, first we'll start with the low-dose CT or radionuclide scan for position. Then we will obtain the dynamic or list-mode acquisition in 2D or 3D mode. Then once we have our images, it's where the portion of checking for misregistration, patient motion or evidence of detector saturation. And this step is very important because in as much as one in four studies, it can have myocardial blood flow quantification issues just based on the fact that, you know, patient motion or any artifact. So now we talked about the acquisition of the images. Now, particular for the myocardial blood flow acquisition and processing, for that we require the measurement of the total tracer activity transported by the arterial blood and delivered to the myocardium over time. And the arterial isotope activity versus time is measured in the regions within the arterial blood pool. Those can be different depending on the software used, but usually are the left ventricle, left atrium or aorta. And again, there's a little difference between the acquisition processing in 2D and 3D systems. In 2D systems, it's usually single static scan. And I made a note here that because the dead time losses in 2D systems and random rates are low and change relatively slowly over time is that we're able to do that in 2D systems. 3D systems do use the dynamic scan with reconstruction of sequential short timeframes. Again, to estimate and quantify blood flow from dynamic PET images, what happens is that time activity curves are fit into a mathematical model describing the tracer kinetics over time. The most commonly used are the one tissue compartment model and the simplified retention model, but in general, they both are based on the concept of normalizing the late phase of myocardial activity to account for the total amount of tracer that was delivered by the arterial blood. And so this is an example of the one tissue compartment model. Again, the dynamic PET imaging starts with intravenous injection. So time zero also depict in this schematic here, and it goes first through the RA and RB as we can see here, then into the lungs and then the left heart. So we have here the LA and the LV blood pool and gradually is extracted from the blood pool and retained in the heart where is the myocardium retention. That's why you see in red where the LV blood pool and how there's a dramatic drop here to zero, whereas there is a plateau here showing the retention in the myocardium. So as we've discussed in previous slides, the quantification of blood flow is actually simple because it's done through software programs, but however, there's certain assumptions that are critical to accurate measurements. So let's start with a few. So first off, the data acquisition needs to begin prior to the tracer arriving at the heart. For example, for rubidium, the counts in the first frame must be zero. In the case of ammonia or fluropyridaz, the counts that remain from an earlier acquisition need to be subtracted. The second one is that for modeling the tracer uptake, a dynamic scan acquisition must start prior to the injection of the tracer, and it should continue throughout the study to observe the kinetic transport of the tracer from blood pool to myocardium. For retention methods, the blood pool tracer concentration can be acquired dynamically or as a single static frame. And this is very important, the confirmation of proper ROI placement, the proper and consistent location of the blood pool ROI must be confirmed, and you should always make sure that it's not touching any adjacent structures, and it is important to note that these placements may be different in different software programs, so it's real important to know and understand the requirements for the specific software program that you're using. Dynamic images should be free of patient motion, particularly for compartment models, and that way the blood pool ROI needs to remain centered in the LA or LV cavity for the entire scan duration. Any motion should be manually or automatically corrected with the software algorithms prior to quantifying the myocardial blood flow quantification because it can really give different values. And then a detector saturation must be avoiding, and this can be achieved by limiting the injected activity concentration according to the scanner count rate ability, so it's again important to know your scanner as well. And all these artifacts really can cause underestimation of the blood input function and then overestimation of the myocardial blood flow. As part of the quality assurance of the dynamic series, again, dynamic time activity curves must include at least one zero value frame to ensure adequate sampling. Motion correction, it's important, and also check the shape of the time activity curves. You check for the peak height of the blood pool, attack at both rest and stress, and they should be comparable when similar radiotracers are injected. And also check for multiple peaks or broad peaks. That usually means that it was probably a poor quality injection. So now let's move to quality control, and this is a very, very important section, and it's really the essential first step in the decision to report your myocardial blood flow measurement. If you have a poorly obtained or poorly derived myocardial blood flow quantification, it's really problematic, and you need to be aware of all the different aspects or different sites where you can be having measurement errors to be able to troubleshoot. Again, all the softwares that are available for this quantification are based in mathematical assumptions. And if you violate those assumptions, the results will be unreliable. So I think the approach to quality control is important, but it is simple. And so I call it the review, correct, and report approach. Review, review, review is going to be very important, and I'm going to be repeating it throughout the rest of this presentation because I think that's probably one of the most important aspects when you are performing flow quantification. So again, review. Review co-registration of your admission and transmission scans. Review them both at rest and stress, and review the placement of the regions of interest. So regions of interest, the myocardium, and also the LA, LV, or aorta, and review both rest and stress, blood flow, and myocardial time activity curves as well. You know, and if at any point the quality fails, and you should be able to correct it. If correction is not feasible, then just don't report them because then you will be giving problematic information to the ordering provider. So what are important aspects of the correction of registration of admission and attenuation correction scans? So always use views as plain. Always check in the transaxial, coronal, and sagittal views for, you know, any motion or any issues with the registration. We recommend that any required correction of misalignment must be performed on the PET scanner console with repeat reconstruction after proper alignment. This is important because misalignment may be due to patient and or respiratory motion. And there is, it can result in imaging artifacts on the relative perfusion side of things, and then a corresponding regional decrease in blood flow. And so you would be prone to sending a patient for angiogram that may even have a normal test if you just check for these things. Again, review, review, and review. Blood pool should be in the same location for stress and rest. As we mentioned before, it should not touch the walls of either the left atrium or the left ventricle to avoid spillover. So this is an example of appropriate placement in rubidium 82 PET. We have the myocardial ROI, and so that it should also accurately encompass the myocardium during all the frames that will be used for determining the tracer uptake. And this is standard ROI. This is in DLV. Again, it's not touching the LV wall or other structures or the LA. Same thing here. It's just, it's correctly positioned. And the software will usually support manual adjustments of the boundaries if necessary. So while you're reading, you notice that there's some discrepancy. You might be able to change it directly through the software. And, you know, this is a, I think a great example of what motion artifacts can cause on a myocardial blood flow. So in this schematic here, we have the same patient, no motion. And we have the patient here, um, and this other side with motion, same images and everything. So I'm here in this one in panel. So panel a no motion, panel B motion of 12.8, 10.2, and 6.8 millimeters in the septal inferior and apical directions respectively. Um, again, you can see, um, here the, um, the green box, the LV, um, blood pool, um, volume of interest. Um, these are the time activity curves in patient with no motion. And this is, um, with motion, uh, where blood spill over and the myocardium is less uniform. And lastly, we have here, um, the polar maps, um, and you can see, uh, here the, the myocardial blood flow quantifications around two with no motion, um, it's about 7.9 MLS per minute per gram with motion. So, um, between these two scans, there was about a 15%, uh, difference, um, with motion. So that can be the difference between a normal study and, and abnormal study. So again, uh, what to do, or what do we need to know, uh, about the time activity curves and how they can be used for quality control as well. So they're different, uh, depending on the, um, the model used, but they do share some, uh, quality requirements, um, the time activity curve must begin prior to the infusion or injection of the tracer, um, it usually will demonstrate an initial increase in blood pool counts. Um, and then you'll notice a precipitous drop in the count, uh, once the, um, there's an uptake of the tracer and then it demonstrates a steady, um, uptake of the tracer, and then it followed by a plateau counts without any abrupt changes. Like I showed you with the one, uh, tissue compartment model figured I showed before. Um, there are a few software dependent considerations. So for instance, for the net retention model, um, again, the myocardial ROI is generally determined from a single myocardial scan. Um, so always again, review, review, and this one I used to inspect, but the same thing, uh, check the ROI to ensure that it's accurately tracing the myocardium and excluding adjacent non-cardiac structures that may contain tracer. Um, the blood phase curve should demonstrate the following important near zero counts in the first dynamic frame of the rubidium acquisition, a strong peak of activity between 25 and 75 seconds after starting, uh, of the infusion and good blood pool clearance prior to the myocardial uptake frame. And I have exactly, um, the, you know, the three points that I mentioned, um, in this, um, schematic, um, you have here the myocardial ROIs, and then you have here the blood pool ROI. So if you can see in the previous one, we had like a little box. Uh, no, here we have a little dot on the three areas. And again, it starts at zero. If there's a peak, then it drops. And then myocardium here are shown by the blue. It's a plateau that that is, um, the myocardium retention. In the compartment model, um, the myocardial ROI needs to be reviewed for correct placement in each frame of the acquisition. So remember that the net retention is a single, uh, scan. This one are, um, several frames. Um, and so it is important to check for, uh, the correct placement, particularly in the blood pool phase. Um, and the dynamic time activity curve will show the blood pool and myocardial activities. So it's kind of like, I guess the excellent thing of, uh, the addition of myocardial blood flow quantification for pet myocardial perfusion imaging is that you're not only taking photos, but you're actually taking a video of, um, of what's going on, uh, inside the heart. Um, again, very similar. My blood pool and myocardial activity pool should start at zero and have single peaks. Um, the rest and stress are very similar. Um, the uptake and myocardial blood flow maps should have similar regional distribution and again, review, review, review the blood pools, pillover maps carefully. Um, so again, this is a quality assurance data for a, um, myocardial blood flow analysis with, uh, rubidium 82. This is a one tissue compartment model. We're checking here, the orientation, uh, the left ventricular lung axis to the center of the short axis. Um, we are contouring the, uh, LV tissue and making sure that everything is centered, um, appropriately and really not overlapping the myocardium. So you can see how everything is traced. Again, um, the time activity curve when reviewed, it starts at time zero, which is the time of the injection. And there, uh, is a peak. Um, this is the myocardium and this is the blood pool. Then you have a peak and then the blood pool drops significantly while the myocardium plateaus. So, um, another consideration to when you're reviewing, um, so you, we went from positioning the patient, having an idea of the dosing, acquiring your images, um, and also the information needed for the tracer kinetic modeling and quantification of blood flow. Now you're in your station sitting with your software, reviewing whether what you're looking at in terms of the blood flow quantification is accurate. Um, and something that always should be considered, um, if maybe the blood flow is not making too much sense is adjusting, uh, for the rest myocardial blood flow, uh, because, um, resting blood flow varies linearly with the product of the heart rate and systolic blood pressure. So, um, the rate pressure product correction that we call, so that should be something, uh, to consider. So to summarize the quality control tools, again, correct registration of emission and transmissions can check, you know, review, review, review, and correct blood pool myocardial ROIs. And, um, the time activity curves, uh, again, need to be reviewed and corrected as necessary. This is, um, basically what we have mentioned. Okay. So now we, uh, we were able to obtain our blood flow. We did quality control, quality assurance. Now, how do we display a typical PET MPI with MBF quantification? So always examining, uh, the transaxial coronal and sagittal views for alignment, um, that reoriented images should be displayed as the short axis. So you slice perpendicular to the long axis of the LV from apex to base, uh, the vertical long axis. So vertical slicing from septum to lateral wall and the horizontal long axis, um, which you slice from inferior to interior. Um, all slices of all the datasets should be displaced aligned adjacent to each other. Um, and again, just a reminder that the standard segmentation model divides the LV into three major short axis slices. So the apical will have four, the mid will have six same with the basal, and then you'll have the true, um, apical in the middle. Um, and this can be also be software dependent because some softwares have the ability to segment the LV in different numbers. So again, this is, um, uh, how we display, we have the short axis, the horizontal long axis, and the vertical long axis. You really can choose between a linear grayscale or monochromatic color or multicolor scale, which is the one that we used here. Um, data from the individual, uh, short axis tomograms can be combined to create a polar map display that I will show in the next slide. Um, and depending on the software, uh, reconstructed perfusion images can also be displayed in 3D static or cine mode. And this, um, again, it's your preference what color scale you'd like to use. So this is basically when you're ready to read, uh, again, polar maps, uh, we have the flows per segment, um, and you have your time activity curves. Again, if this was a video, you will be able to see, um, the uptake, uh, and then the retention. Um, these are beautiful curves here. Um, and, you know, again, this is the, the flow reserve that's, uh, reported and the blood flow. So now we're ready for interpretation. Um, and so to get to this point, you position your patient correctly. You reviewed all the images you obtained. You did all the quality control, um, for the images, for the blood quantification. You checked your curves, um, and you are now ready to interpret those numbers that you, um, that your software is, uh, telling you that the blood flow is. So, um, we, uh, this is, uh, kind of the, this is the numbers that we use. We know that, um, a, a flow reserve, which is probably the, uh, the, uh, one of the measurements that has the most data is associated with an excellent prognosis when it's above two. So, you know, any flow reserve above two, the interpretation is normal and the relative risk for, uh, cardiac related events in that patient is typically low. Um, and when it's slightly decreased, so less than two to 1.7, then it jumps to the intermediate risk, um, less than 1.7 to 1.2 is high. It's abnormal. And then less than 1.2, the perfusion defect is highly abnormal, um, and has a very high relative risk of cardiac events. Um, and lastly, less than 1.2, the perfusion defect may be considered non-diagnostic study, and we'll touch a little bit on that. So the relative risk in that sense is a little bit indeterminate. So just, um, uh, we do recommend that the approach to interpretation of blood flow should be additive to the information that we already get from cardiac perfusion. So perfusion and function first, and then, uh, flow reserve incorporation and later, uh, again, through review, review and review. Um, we, the recent ASNIC practice guidelines, um, do recommend evaluation of restress, um, blood flow and reserve global, um, also provascular territory and segmental within, uh, uh, territory as well. Remembering again, that the data and literature is primarily based on global, uh, and the, uh, my cardiac blood flow reserve and, uh, small variations, uh, in each segment of a territory, uh, may exist. So when we, um, talk about the interrogation of the individual segments in the vascular territory, this is, uh, you know, in our software, this is, this is where you can see, um, the different variations, um, in each segment. Um, it really will provide more specific, uh, information about this vessel branches, but again, the position of the myocardial blood flow measurements do decrease, um, with smaller segments due to statistical noise. So you need to be very careful not to over-interpret in those small, uh, regions. Okay. So the, uh, this is, uh, important, I think, to, um, remember that again, the blood flow indices should be interpreted in the clinical context as, uh, the presence or absence of coronary, uh, artery disease changes really the approach to the data, but it's useful in patients with non-OCD and all those with, uh, CED. So just to mention a few things, you know, when it's useful or the clinical value, you know, in patients with no known CED, you know, if you have normal perfusion, you're high, you have a high negative predictive value if your MFR is normal, if you don't know, for example, regular aspect, um, and you're not sure whether this is an artifact or not, and then, uh, or it's a very mild defect, but you also have a, an issue with the, uh, flow reserve in that vascular territory that helps you confirm that that is perhaps coronary artery disease, uh, may help you predict more severe disease, you know, one vessel, a normal perfusion versus two to three vessel, you know, depending on the flow reserve. And it may also help you identify balance, um, CED or microvascular disease, you know, with patients with known CED, it's still useful. Um, again, it may be abnormal after CABG, um, and patients with already coronary events. Um, but for example, in patients with cardiomyopathy, if normal, it helps exclude CED, um, and, you know, uh, then post-PCI, again, it may be abnormal, but it's usually, uh, useful if you have, uh, pre-PCI data available. So now you're ready to report your blood flow. These are just sample sentences that can be, um, found in the ASNIC, uh, uh, document that was, uh, recently released. Again, um, you may choose to add, uh, phrases in the description, whether, uh, I think it's always important to mention whether there's a rise in blood flow or no rise in blood flow, which might, um, signify that the, there was not a good vasodilatory effect. Um, you can, in the conclusion, you can, um, you know, talk about if it's normal, um, how it indicates lower risk of coronary disease beyond the normal perfusion, or, um, if the flow reserve is abnormal, truly abnormal with a normal myocardial perfusion, the patient's still at a slightly higher risk category than patients with no known CAD, or if you think the perfusion defect, um, correlates with the, uh, regional reduction in flow reserve that can help also guide, uh, angiography and further, uh, testing. Um, and you know, if, if you reviewed, corrected, um, and you decided that technically you can't report it, it is okay to say, you know, I'm not reporting in this patient due to technical patient specific concerns that can affect accuracy and inappropriate clinical decisions. Um, just remember that there, uh, there are four scenarios where you can have a common discordance between perfusion and flow. Um, when you have normal, uh, visually perfusion, but, uh, low or very low MFR, um, it can be multi-vessel CD. It can be a diagnosis of coronary microvascular disease, or it can be a combination of moderate diffuse epicardial disease with microvascular disease. But most important, it can be that it's, there's a circulating pharmacologic stress inhibitor, um, such, you know, maybe the patient forgot that they had tea for breakfast and they didn't think that that was, you know, caffeine containing product. So always consider this first. Um, if there's still uncertainty, uh, you may want to repeat the stress portion that same day. Um, obviously if you have a lot of coronary calcifications, then, you know, uh, it really might change, uh, it might be any of the, uh, three items above. So to finish, I just, uh, want to quickly go through two cases just to kind of like review everything and put everything together that we learned. Um, so our case number one, uh, it's a 49 year old woman with obesity presented with intermittent angina. She has this lipidemia and the question was assessed for ischemia. So she underwent in a remedium 82 pet MPI. I'm not sure if you're able to see the perfusion images, but they're pretty normal. I think the one thing that maybe we can mention, there's like some GI uptake, but this is in the rest images. And it really doesn't cause any issues with interpretation. Um, as you can see the reserve, um, this stress and rest and the flows are pretty normal. Um, the rest is slightly elevated, which is not unusual to see in patients with obesity. And then we have here, um, the CAT scan for attenuation corrections showing no coronary calcifications. So how do we, um, report that? Um, so no ischemia during recognizance and stress, and we're talking about the perfusion. Then we talk about function, which was normal that I didn't show in the case. Um, then we talk about, um, the flow reserve and the coronary calcification. So really an, an, uh, normal study. And then, uh, for the second case, we have an 80 year old man, no, uh, CED with total, uh, with the CTO of the RCA who presented with worsening excretional dyspnea concerning for an adrenal equivalent. So we have the perfusion images here, um, and we can see a, um, moderate size, uh, and we have conceived here, uh, severe reversible perfusion defect of the, um, inferior and proceptor walls in when the distribution of the known CTO, and we can see here the flows are mostly diminished in the RCA. So how do we report, um, this study? So again, um, we really recommend doing like a general statement. So in this case, we called it an abnormal recognizance and stress perfusion test. Then proceeding with perfusion would describe, um, what we're seeing and the defect. Um, we go again with function and any well motion abnormalities like in this case, and then we report, um, the flow, um, you know, and if we think, for example, in his case, that the myocardial flow reserve is abnormally low on the RCA territory, uh, that suggests single vessel disease. Okay. So, well, um, this brings our, um, talk, uh, to conclusion, uh, again, during the past, uh, 45 minutes, I walk you through obtaining, uh, pet, uh, study, uh, acquiring it, processing it, uh, doing the quality control quality assurance of the, um, myocardial blood flow quantification. Um, I hope you remember the review, review, review, correct. And then you will be ready to report, um, accurate, uh, uh, myocardial blood flow quantification indices. Uh, thank you very much.

PET myocardial blood flow evaluation

patient preparation

camera setup

quality control

cardiac PET perfusion study

coronary artery disease

microvascular dysfunction

Yale University School of Medicine