Hello, my name is Dr. James Case, I'm with CBIT in Kansas City, Missouri, and this talk is going to discuss hybrid imaging pitfalls and artifacts. Here's my disclosure slide, please take a moment to review this. Hybrid imaging is when we pair two different modality systems into a single imaging device. Some examples of these most commonly used in cardiac would be PET-CT, less common in cardiac PET is PET-MRI, also SPECT-CT systems are also examples of hybrid imaging. What we would not consider hybrid imaging is SPECT or PET, where we use radionuclides for attenuation correction, or SPECT and PET with a non-diagnostic CT, which is solely used for attenuation correction. Really the key point of a hybrid system is that we're using two separate modalities in the same machine to capture two different types of data that we're going to use for our clinical diagnosis. Some of the major advantages of hybrid imaging is compared to radionuclide-based attenuation correction, as we have fast diagnostic CT scanning times, very high count studies for transmission attenuation correction, but probably the big one is that we have the clinical value of CT, and in cardiac, that means that we can assess calcium, and in newer CT systems, we can also perform different types of tests with the system like coronary CTA. Also in a number of institutions, they found this as a valuable instrument for getting higher quality technology into their laboratory by sharing the expenses with other departments such as neurology, oncology, and radiology. Some of the drawbacks to hybrid imaging are they intrinsically have higher radiation dose than radionuclide imaging. Secondly, the motion artifacts in hybrid imaging can be more complicated than what we have with radionuclide-based attenuation correction. Another problem that could come up is metal artifacts, because metal is a very efficient absorber of x-rays, it can create very challenging artifacts that can be impossible to read around without metal artifact correction techniques. And finally, there is considerable challenge in what to do with the CT data if we don't have the clinical expertise for interpreting radiology-type scans like detecting lung lesions and non-cardiac findings. Here are two examples of PET and PET-CT systems, kind of illustrating the differences between hybrid imaging in the PET laboratory. Up at the top is a dedicated PET system. These are typically older systems based on a platform where a radionuclide source circles around the patient and projects 511 KeV photons through the patient that are then used for attenuation correction. These transmission data sets have very limited clinical value, and really just for doing attenuation correction. The drawback of this is because of the low photon output, is these transmission times can be rather long, sometimes two to four minutes in length. Hybrid imaging, in this case what we use is a CT. So there's an x-ray tube positioned around the patient that orbits around transmitting x-rays through the patient that is then used for attenuation correction. Now these units, as you can see, have a larger footprint. They're heavier. They have higher power requirements than their dedicated partners. And because of that, they can have a more expensive build-out in addition to the additional expense to the CT scanner. The benefits are going to come in with that short transmission scan. Now those were PET examples of hybrid imaging. Here's some SPECT attenuation correction systems, and as I pointed out earlier, line source attenuation correction, like we see in the top example, is not considered a hybrid technique. The attenuation maps that are created have very limited clinical value. Similarly, there are some systems out there that use a very low dose CT that, again, has very little clinical value that's just for doing attenuation correction. When we talk about hybrid SPECT CT systems, we really are talking about true SPECT CT systems like 4-slice or 16-slice or even 64-slice type of machines when talking about SPECT CT systems. Now when we look at the variety of systems out there today, the SPECT CT scanner can also be used for doing CT. So this table here outlines some of the different techniques that are possible based on the number of slices. Now the number of slices in a CT is how much volume is captured in each one of the passes of the CT scanner, and that determines really what the final image quality of the CT map is going to be. So in some of the earlier systems that had four or six slices, we only can acquire at a very coarse resolution, one millimeter of in-plane resolution and three and a half millimeters of plane-to-plane. This really only gave us the ability to do attenuation correction with the CT without being able to quantitate things like coronary calcium. We can still visually assess calcium with these scanners, but we can't perform a quantitative like an Agson score with these types of systems. So they're really only valuable for perfusion and viability. When I say perfusion, I mean the full suite of perfusion, gating, and myocardial blood flow as well as FDG-based viability. But there really is very limited value for the CT portion of the study. By the time we get up to the 16 slice machines, we get to the type of resolution that we can look at some of the CT-like studies that can be performed, such as quantitative calcium assessment and some non-coronary CTA type of studies like runoff studies, neuro device infection, sarcoid, those sorts of things we can do with these 16 slice machines. Once we get to a 64 slice machine, we can come up with isotropic type of resolution. And by isotropic, I mean the same in-plane resolution as they have sliced. Now this allows us to open up the full suite of CT studies, such as coronary calcium, coronary angiography, as well as the regular studies that we would do with sarcoid device, neuro, whole body, oncology, et cetera. So hybrid imaging for attenuation correction. What artifacts that we minimize with hybrid imaging is we are going to use the hybrid imaging for attenuation correction, which means that when we're interpreting the studies, we never call breast attenuation and we never call diaphragm attenuation. So that doesn't mean we have perfect attenuation correction. They have to be quality controlled. And when it fails quality control, the attenuation correction may not be working correctly. But we should never let slip into our reports that there's a breast attenuation or diaphragm attenuation artifact that's really inaccurate. The attenuation artifacts we do need to worry about are misregistration. Because the CT is acquired at a different time, the patient, if the patient moves between the transmission and emission data sets, it can introduce an artifact because of a misplacement of the data between the two types of studies. And that would be true of also dedicated PET and PET-CT. Now, some new sorts of artifacts that we see exclusively within hybrid imaging are breathing and breath hold. The PET side of the study is done during a free breathing phase over several minutes, whereas the CT, which scans across the patient's body, can be accomplished in a matter of seconds. So the way this is done is we have to do some kind of CT-specific breath hold or breathing protocol. And I'll talk a little bit more about breathing later on. But an inappropriate breathing protocol can introduce a different family of artifacts. Beam hardening, again, this is exclusive to hybrid imaging. We don't see that. And that happens because in the instance of metal or a lot of bone, you get a very efficient absorption of the x-rays that can create artifacts of its own. And then finally, noise when we don't apply enough CT dose in the largest of patients. The way we use hybrid imaging for creating attenuation maps is the first thing that we do is we acquire a CT transmission map. And I'm drawing a little drawing of the steps that we're going to do. So this is a regular CT. It takes a few seconds to cross the patient's body. And then after that, we dumb that image down so that it's really as if we acquired a line source attenuation. We reduce the resolution, and we translate all the attenuation coefficients in the CT image over into their equivalent, 511. And once we do that, then we forward project that to create a transmission cynogram, as if we had a line source of activity projecting through the patient. And that transmission cynogram can then be multiplied to the emission data and create our final attenuation corrected cynogram that we can then use to perform our attenuation correction. Now, when we look at this particular study, this is a normal perfusion study. But because we haven't done any quality control, there's a number of different things that might be going on with this particular study. In particular, one possible interpretation of this data is this is a normal patient with no CAD. It's a normal patient with no CAD. It's a normal perfusion study with a non-obstructive CAD, an abnormal study, but a non-responder due to something like caffeine interference or microvascular disease. The quality control process that we do is going to help us differentiate between all these different possible outcomes in the study. So a little bit tongue-in-cheek here, we always perform quality control, even if the appearance of the image is normal. We want to be able to do that to be able to figure out all these different possible interpretations. So I've added a bunch of kind of silly things that may not matter, but maybe the only thing that does matter is the patient comes into the laboratory with no known CAD. So pretty much we have no information walking into it until we start reviewing the data. The reading process that we go through in assessing whether or not there are artifacts present is the first thing that we're going to do is we're going to check the registration, and that is the registration of the emission data, that's the positron data, or the spec data in the case of a spec CT study, and make sure that the CT and emission data are in the same place in space. We need to be able to do that for both the stress and the rest. Most of the visualization software out there is only going to display one of those sets, so either the stress data sets or the rest data sets. We have to remember to look at both data sets to properly assess for misregistration. The second step that we do is we look for any interfering metal artifacts that may be introducing hotspot artifacts. Then finally, we check for breathing artifacts, any multiple positions of the diaphragm or inappropriate breath holds. Then finally, we inspect those CTs for any secondary findings such as calcifications and non-coronary type of results. So here's how we do this. The first thing that we're looking at is we're going to look at the stress overlays, and you can see there's a nice registration. We see the emission data and the transmission data. The transmission data set is of high quality. The emission data set is also of high quality, and the registration between the two is accurate, so the emission data is properly positioned in the mediastinum. We do this for the stress, and we also do this for the rest. Now we've confirmed that the rest transmission study also is of high quality, and the emission data is properly laid over the transmission data. Now the next thing that we look at is going to be the transmission, the CTs, and we start above the heart and go through the entire volume. As you can see, we're going from above the heart, up in the vasculature, all the way through the volume, down into the diaphragm region, and we do this to look for any kind of calcium and secondary findings. You may have to pass through this several times to find out whether or not we have an accurate result with that. Now what do we know? We haven't actually begun interpreting, but look at all this new information that we have by doing the quality step. We have a high quality study. There's no misregistration. There's a minimum of intrascan motion. We have a CT of high quality and a small amount of coronary calcium. So now we can begin interpreting the study. So now looking at this, we have confidence that we're looking at at a high quality study, minimum of calcification, maybe a little bit, and a normal perfusion test. And we look at the gating. So we have normal wall motion and thickening and an improvement in ejection fraction at starting at 72%, going up to 76%. And it's nice to look at these wall motion in both color and black and white. The color is good for assessing thickening, whereas in black and white, it's better for assessing the wall motion. Then finally, we look at the myocardial blood flow. And in this case, we have a high quality blood flow study. All of our lights are green in terms of the peak to plateau, very little background activity at the beginning of the study, a sharp peak indicating that the ROI placements are accurate and a nice normal blood flow reserve. So where do we sit with the interpretation of this study? Now what we know is this is a high quality scan, no misregistration artifacts or breathing, no metal artifacts. The CT is of high quality. There's a small amount of coronary calcium, normal perfusion, gating, and blood flow. So you can see we have a tremendous amount of information in there, but probably the most important is because we went through those quality control steps. We have a high degree of confidence in our final conclusion with this study. So now what we're gonna do is we're gonna look at a SPECT-CT study. So this is a low dose tetraphosmin study at rest, followed by a high dose of tetraphosmin at stress. The patient was done with attenuation because of breast implants and a history of breast attenuation. And this is done on a dual head anger system with a six slice CT. So looking at that particular study. And now with PET studies, we don't look at rotating images for looking for patient motion, but with cardiac SPECT we do. And as you can see in the study, there's a minimum of motion within the stress and rest studies and good counts within the rotating images. And just like what we do with PET, we're gonna inspect the overlays for accurate alignment. In this case, what we have is we can see a good overlay of the transmission study at stress. But if we look at the rest, the rest overlays, I wanna highlight this particular artifact. And go ahead and draw this. We can see that the heart is sitting out in the lung field. Now the lung field has a lot lower attenuation coefficients than the soft tissue, about anywhere between a 20% of the density up to about 30% of the density of the soft tissue. And when we apply those attenuation coefficients, which are lower, what we do is we reduce the attenuation corrected values. So it can appear when we have misregistration between the two, as if there is a drop in counts or a fictitious defect. So looking at how that appears in the emission data set. And luckily for this study, the artifact happens within the stress study or within the resting study. But what we can see is right in here, what we have is an artifact, which is making that apex very difficult to read. As we're trying to read this small reduction activity here at the apex like this, we don't really have a good rest comparator. So when we see these sorts of things, we would ask our technologist to go back and reprocess the study and correctly align these images with one another. So how frequent are these types of artifacts? They tend to crop up pretty often. PET CT, they're very common. As little as one centimeter can introduce a change in the perfusion artifacts. So just to shift as little as about that much can introduce a change in the perfusion appearance. And this can change in PET CT, the diagnosis in a very high proportion of cases. And in this older publication, that can be anywhere between 30 and 40%. It can also change blood flow values. So the take-home with this is always when interpreting these studies, inspect for the proper registration at both rest and stress for the correct alignment between the two. And since it is a very simple artifact to correct for, when you do see misregistration in those images, send those back to the technologist. And then finally, this needs to be done on the dynamic as well as the emission and gated data sets. So again, it may take time to go and redo all the reconstructions, but it's a necessary step to be able to have a diagnostic readable study. The next type of artifact is a breathing artifact. And the challenge with breathing artifacts is that it not only can move the relative position between the heart and the lungs, it can also change the shape of the lungs if it's done incorrectly. So some things that can introduce a breathing artifact are stuff like coughing, ininspiration breath hold, or just bad luck catching the diaphragm at a place that we don't want it to be. It can be difficult, if not impossible, to fix with shifting. When it is suspected that there may be a CT that has been compromised due to a breathing artifact, you do, it is strongly recommended that you acquire a second CT, so both a stress and a rest CT, so that in the case of where a patient has taken a breath or there's a bad breath hold, we do have a backup CT to correct for the images. It's not a big deal to substitute a stress CT map for doing the attenuation correction for rest and vice versa. But having both CTs can save the day in more often than not. So here's an example over at the left of this image, and I wanna draw your attention to this. This is one where the patient has taken a breath during the study and pulled the diaphragm downwards. You can see the red arrow points to the diaphragm here at stress, it's up in a higher position than where it is at rest, where the diaphragm has been pulled down lower. So there's an inspiration, more of an inspiration type breath hold at rest, which drags all of the soft tissue down and then pulls it out of alignment with the rest of the emission data. So what does that look like? Well, the artifacts can be quite profound. So the bottom study, the rest, is that misregistered and breathing artifact all combined into one. So we need to be able to correct for these sorts of emission artifacts. Now, the way we would correct for this, because the in-inspiration breath hold is not only moved the heart relative to the soft tissue mediastinal transmission counts, it's also warped the shape of it. So the correct course of action here is not to just shift the transmission coefficients relative to the emission counts, but actually to use a different map. So we're gonna use that stress transmission map that doesn't have that inaccurate breath hold for correcting for the image. So if we go ahead and reprocess this data set, we can use that stress transmission map to fix the resting study, as well as do a regular attenuation, misregistration correction on the stress and get a normal looking appearance to the study. Now, breathing artifacts come in a number of different families. Here are two different types. One is what I call is the mushroom effect on top. And what has happened here is the diaphragm on the far left upper panel that I've just drawn a line on. The diaphragm is caught in an in-expiration level. And then as the CT continues to progress, the patient breathes in, and the diaphragm sinks down to an in-expiration position. So the diaphragm gets caught at two different places during the CT scan. And that can introduce some rather challenging artifacts as we can see on the top row. Another one is where the breath hold takes place and creates kind of this jagged appearance on the lower example. And this can happen with like coughing where there's a sharp change in the position of the diaphragm. And this can create kind of a chopping artifact in the image. And this is where that second scan, second transmission can come in and help out with doing the attenuation correction. When you see these situations where the diaphragm is in two different positions, or if you see kind of a sawtooth appearance to the edge of the medial stynum in these coronal views, send those studies back to the technologist for reprocessing and ask them to consider using the other transmission scan for correcting for these artifacts. Because these are hard to detect, many laboratories will routinely acquire a transmission study at both stress and rest. As long as it's a low dose, low MAS study, we're not contributing a great deal of additional radiation dose, but it can turn a non-diagnostic study into a very readable study. The next artifact that I wanted to discuss is implanted metal devices. And this can, many of our patients these days do have a lot of metal within their body. It can come from shock wire, surgical clips, pacemakers. All this stuff contributes to the star artifact you can see at rest and stress here, on where the septal wall would be. There's a bright on the CT map of very high density metal that's blocking the CT x-rays from passing through that metal. And it creates regions of high and low counts. And when we translate that into our final images, it can appear like a hard knot of extra counts. And it somewhat depends on where the heart ends up relative to that in the transmission study. So here we can see that hotspot being introduced has more of effect on the stress images down in here. You see on the top row, then on the bottom row of the rest, it's a little bit there, but less pronounced. But what ends up happening is we have a drop in counts in the anterior relative drop in counts that can make this more difficult to look at. And when we go ahead and apply transmit the metal artifact correction, let's draw it off there. We can make this a less pronounced artifact. And you can see now the anterior wall is of the same brightness as the inferior wall. There is very little downside to applying metal artifact correction. And it should be applied on every patient whether or not it's needed. It has very little downside. Here's an example from a more significant where there was a more significant artifact. And you can see that in hot knot that's created along the metal wire passing through the infraceptile regions that when we apply metal artifact correction, it goes away. So finally, in summary, the acquisition side, we need to do QC on with hybrid imaging. And probably the most important thing I would recommend is that to practice whatever breathing protocol that you're doing with your patient. Some protocols have in expiration breath hold, others use a free breathing breath hold. But whatever breathing protocol you use, it shouldn't be a surprise to the patient. Practice with the patient prior to using it. Prior to bringing them into the imaging suite. Another recommendation is to go ahead and do a second transmission scan at low dose so that when these problems crop up, we can have a second CT scan that may be the one that provides a good result for both studies. In processing, reconstruct using the best CT. The technologist should be trained to go ahead and look at how the CTs look and switch between stress and rest CTs to use the best CT map available. Check the registration, both at the stress and the rest, and both of those should be registered correctly. And then on metal artifact correction, I would recommend using it on all cases. There's very little downside for most metal artifact correction. Finally, the reading process that you should follow needs to be done on every single patient, regardless of how the images look. First, inspect for registration, then inspect for breathing, then finally inspect for any metal that might be interfering with the reconstructed images, and then finally inspect that CT all by itself for any secondary findings such as coronary calcium. And then, and only then, should we begin interpreting the study. Thank you for watching this video and enjoy the rest of the workshop. Thank you.

hybrid imaging

artifacts

PET-CT

SPECT-CT

radiation dose

quality control

metal artifact correction