Pitfalls: PET-CT, Part 2

Introduction

In the second lecture here, we are going to move away from the neck and chest and take a short divergence and talk about artifacts, and then move more inferiorly and talk about the abdomen and pelvis.

And then at the end we'll have a few more difficult cases for you to test your knowledge.

Majority of the artifacts that we're gonna talk about are unique to PET CT, which means that with dedicated PET imaging you would not see them.

Although there are a couple that I'm gonna go through that you might see on PET imaging as well.

The reason that we do see them on PET CT is because we're using CT for attenuation correction rather than with PET only, where we used point source.

So PET CT uses CT for attenuation correction rather than a point source.

And then the attenuation correction artifacts are unique to PET CT.

Some of them include artifacts related to intravenous contrast, oral contrast, and other high attenuation materials.

In terms of the PET CT artifacts that we are gonna discuss, we will talk about the unique artifacts secondary to intravenous contrast, oral contrast, ports and other high attenuation materials as well as we'll talk about breathing artifacts and some protocol issues related to breathing.

And then we will resume talking about some of the pitfalls in the abdomen and pelvis and take a look at normal and atypical organ uptake, as well as what some of the brown fat and muscle cases look like in the abdomen and pelvis, and then talk about some difficult cases.

Attenuation Correction Artifacts

In terms of attenuation correction artifacts, PET CT uses CT for attenuation correction, as I said, rather than a point source which was used in dedicated PET imaging.

And that actually has improved things quite a bit in terms of throughput because to do a scan for attenuation correction using a point source, we had to do different imaging and it took upwards of eight to 10 minutes.

Additional timing with CT using that for attenuation correction, it only takes us an extra minute, essentially.

So we do get some improvement in terms of time, however, we introduce artifacts that I'm gonna discuss that are unique to because we're using CT for attenuation correction.

And then we'll talk a little bit about non AC artifacts.

Again, as I said, breathing artifacts in arm position.

The actually physicist from the University of Pittsburgh Medical Center worked out some of the issues related to using CT for attenuation correction.

And what they originally did was they essentially they needed a way to segment the bone pixels or the pixels that were generated from CT up to the energy of the PET energy, which was different.

So CTs around a hundred KEV, whereas PET imaging is 511 KEV.

And not to get too technical, but essentially what they did was they segmented out soft tissue from bone, gave them different scaling factors, and then just scaled it up to the PET energy.

And then you get the attenuation corrected images from doing that using CT.

So what happened with that is that you get new artifacts, and this is we're gonna talk about intravenous contrast first.

This was one of the first descriptions from 2002, and what they noticed was when you inject intravenous contrast on a PET CT that you saw this uptake, if there were areas of high attenuation or pooling contrast, whereas on the non attenuation corrected images, you actually didn't see it.

And this is a case from our institution where it looks somewhat focal on the coronal image.

You look at the axial images and you see that this uptake appears to correlate to the area with pooling contrast in the left subclavian vein here, however, if you look at the non attenuation corrected image, you see that it actually does not exist.

So this was generated on the PET image during the attenuation correction process using CT.

Whereas on dedicated PET imaging, we would not see this type of artifact.

Occasionally it can be very focal though.

And so we see some normal thyroid activity as we saw in the last lecture.

And then we see some very focal FDG activity or what appears to be FDG activity.

And it looks like it correlates very nicely to contrast within the vessel.

Again, we look at the corrected images.

When we look at the non attenuation corrected images, we see that it doesn't exist.

So this is a focal artifact from intravenous contrast.

Now, keep this case in mind and then take a look at this case.

And it looks fairly similar.

There's a focal area of uptake.

It looks like it correlates to an area of contrast, but in fact, there's a small soft tissue density behind the vessel.

And so if you thought, well, this could be an artifact, you need to look at the non attenuation corrected images, and you see that it does actually exist.

So this is not an artifact.

And so this is gonna be a recurring theme with these attenuation correction artifacts, is that you need to look at the NAC images to verify, particularly when you have areas of high attenuation, whether it's oral intravenous contrast or devices.

Lymphangiogram Effect

Now this is an artifact that you would see on both PET and PET CT imaging, and this is a lymph angiogram effect from injecting FDG and blowing the vein so it infiltrates into the soft tissue.

And that FDG activity is then taken up into the lymphatics and you get this sort of linear area of uptake.

But this is actually real.

And you can see it actually in the lymphatics here.

And then what you do get are multiple lymph nodes in the left ULA and even into the mediastinum.

So these patients will often have to be reimaged at another time, particularly if you're worried about a breast cancer or lymphoma and that kind of thing.

Oral Contrast Artifacts

Oral contrast, the same thing that you see with intravenous.

If the higher the attenuation of the oral contrast, the more the degree of artifact that you see.

So here we have very high attenuation material in colon, and we see these areas of apparent very intense FDG activity.

If we look at the fused non attenuation corrected images, we see that again, it doesn't exist.

So this is by definition an attenuation correction artifact.

How do we can we use other types of oral contrasts that might correct it or reduce the amount of artifact?

Yes, in fact, this is a group from Essen, Germany that studied using Voluma or a negative or water-based oral contrast.

And in fact, if you use that, you see that the effect actually is essentially resolved.

And this is comparing the same patients with a barium based oral contrast versus Voluma.

And you can see that essentially it just resolves.

So just keep it in mind.

If you do have, if you use oral contrast on your PET CTs, which most people do these days, that you will have some degree of artifact and sometimes it's superimposed on physiologic FDG activity as the colon and even small bowel can have physiologic uptake.

Just to show you that depending upon the scanner that you use, the artifact may or may not be as significant as I'm showing.

But this was the standard algorithm was to classify CT pixels as either bone or soft tissue.

And they used a cutoff of essentially 300 Hounsfield units in the beginning.

Now unfortunately, oral contrast can be in the range of 700.

And if you have, say, a patient that had an upper GI and has a residual barium, it can be even much higher than that.

So again, most of the vendors have adopted this modified algorithm, which just not to get too technical began, but to they segment out the bone pixels, they segment out oral contrast pixels with known 511 KEV values, and then they rescale it using the original algorithm.

And essentially all that does is it corrects for the artifact.

So what you have here is physiologic FDG activity, and then you've got superimposed artifact that it corrects for.

And again, as we see a very high attenuation material in the stomach, this is from a patient that had an upper GI, the Hounsfield units we're actually measuring very high.

We see on the non attenuation corrected image, there's nothing on the standard algorithm.

There's intense activity, whereas on the modified algorithm, there's very little.

How about this?

Does this patient have a colitis?

And of course, on the CT, there's this patient probably shouldn't have been scanned.

But we do see extremely high attenuation contrast in the colon causing streak artifact.

And this is mostly artifacts.

So if we look at the axial image through that area, we see that there is some physiologic FDG activity, and this is the non attenuation corrected image.

Whereas of course, this image was the corrected image.

Benefits of Oral Contrast

Benefits of using oral contrast.

And like I said, most people that do PET CT do actually use it, delineation of the GI tract mucosal versus submucosal lesions.

Sometimes looking at adjacent mesenteric nodes can be much easier.

Of course, peritoneal carcinomatosis, you can see it much easier on CT than with PETs.

A lot of times with PET, it's you see sort of a dirty appearance.

And I'll show you a couple examples of that.

And then identifying small pelvic and peripancreatic lymph nodes, physiologic versus pathologic FDG activity in the GI tract, it becomes a little bit more easy to identify, although again, oral contrast can cause an attenuation correction artifact.

So it's important to always look at the non attenuation corrected images.

And then of course, a detection of PET negative lesions.

So again, having nice oral contrast on board, we can see that there's some nodularity of the omentum, that's much more much easier to identify.

And then, as I said, with the PET portion of the exam, we only get a little bit of activity because of the size of these nodular implants.

And once they become larger than about one centimeter, they will become much more intensely FDG avid.

But early carcinomatosis can be very difficult to identify on PET imaging alone.

Another case here showing nodularity of the omentum mild FDG activity.

And so when you look at the PET images, you don't see you sort of get this dirty or haziness.

A lot of this is physiologic bowel activity and it's very difficult to tell.

So it's having the contrast on board does help delineate those types of cases.

Ports and Other High Attenuation Materials

Looking at ports and other high attenuation materials.

So here we have a focal area of apparent FDG activity correlates very nicely to this chemotherapy port.

Now the differential for this, if you inject FDG into a port, you could get this appearance, which is why we don't generally inject into the ports.

We usually use a peripheral vein, some inflammation if you just put the port in, but it's usually more around the port, not in the port.

And then the third option would be that this doesn't actually exist and that it's an artifact.

So looking at the uncorrected or attenuation corrected images, we see that again, there's a photon deficit here showing that it is an artifact.

Same thing here on the right side.

Again, there's very high attenuation port here.

It's there on the attenuation corrected images on the non AC images, it's absent or photon deficient.

And these artifacts you will see with other orthopedic devices or other types of implants.

So again, without attenuation correction, we see a photon deficit standard algorithm.

And then again, sometimes with these modified algorithms, we can correct for abnormal or types of structures that are implanted.

Dental and Orthopedic Hardware Artifacts

Other dental hardware and amalgam can cause artifacts very similar.

So here we see areas of apparent uptake, a lot of streak artifact on the CT apparent uptake.

We look at the non attenuation corrected images and it's photon deficient.

Orthopedic hardware in the spine can be difficult.

The differential for this type of uptake would be loosening infection versus artifact.

And it's important again to check the non attenuation corrected images because you don't wanna talk about loosening or infection if in fact it doesn't exist, if it's just an artifact.

And then focal areas of apparent increased uptake.

So this is a patient that had lung cancer and this was the only other potential abnormality.

We can see that there's a very high attenuation material in the spine, and this patient had a vertebral plasty.

And again, we see that it doesn't exist.

Now, early post procedure, you can have mild to moderate FDG activity from inflammation, but that generally resolves.

This is a patient who had colon cancer and has a large surgical clip with apparent activity around it.

And again, we look at both the attenuation corrected and NAC images, and we see that the majority of this activity here is from an artifact.

And then interestingly, calcified lymph nodes can also give you an attenuation correction artifact.

So I'm gonna show you two cases of patients with lung cancer, where in this particular case, it's a subcarinal lymph node that appears to have fairly intense uptake.

And so the differential for this appearance would be granulomatous disease, which I guess would be somewhat, there's still an inflammatory component versus an artifact.

When we look at the uncorrected images, we see that it it's not there.

So this is indeed an artifact.

And then this is a patient that had a left sided lung cancer, and the only other abnormality was a contralateral lymph node in the inferior right paratracheal region.

If we again look at all the images, we see that it's there on the attenuation corrected not there.

And so this potentially could have a major impact in terms of their surgical management.

So a single contralateral lymph node would make them stage three B without it, the patient went for surgery.

Respiration and Respiration Artifacts

Let's switch gears and talk a little bit about respiration and respiration artifacts.

Because the CT is done as essentially a single snapshot, whereas PET imaging is done as time elapsed photography, if you will.

So CT you get a single shot where with PET imaging, we collect the data over a single area for about four minutes.

And so we get essentially a summation of the respiratory motion in that, and the majority of it is at end exhalation.

So there's actually a mismatch between the anterior aspect of the chest wall between the CT and PET images, which can lead to on the corrected images, a vanishing of the anterior chest wall.

So there's a number of ways that you can do the respiratory acquisition.

So you can do what a lot of people do is tidal respiration with older CTs.

This can cause a fair amount of respiratory variation near the diaphragm here.

So you see this mushroom artifact.

The second would be to have the patient breathe tidal respiration until you get to the top of the lung, have them hold their breath until they get through the liver, and then tell them to breathe normally again.

And this will actually reduce this artifact by about 60 to 80%.

With older scanners, it can be fairly significant where a portion of the liver appears to be in the lower portion of the chest.

And then at the extreme, you can have actually half the liver where it appears to be displaced into the lungs.

This was a paper that we did describing this modified respiratory artifact where you have the patient breathe tidal respiration, then hold their breath wherever they are.

Generally the best alignment will be at end exhalation.

And then have them start breathing once they get through the liver.

And again, this will reduce the artifact by about 60 to 80%.

Occasionally you can have lesions which are either in the artifact or adjacent to the artifact.

This is actually a ground glass opacity adjacent to this mushroom artifact that we see.

And so it's important to not blow it off as purely artifact and to look at all of the images.

You see that there is uptake within this ground glass opacity.

And this was a BAC or the new terminology adenocarcinoma in situ.

Okay, so then your options then are, you can do the CT at full inspiration as we generally do for CT of the chest, but again, you're gonna have this vanishing chest wall where the corrected PET images may be missing part of the data because it's using the CT for attenuation correction and it's using the wrong attenuation values.

Alternatively, you can have it with tidal respiration, where again, you're gonna generate these artifacts.

If you have a 64 slice scanner, you probably won't have much of an artifact.

Whereas if you're using four eight slice scanner, which many PET CT scanners are these days that are out there, you will get significant artifact or you can do the modified breath hold.

Having them hold their breath through the chest, have them breathe normally after that.

So those are the options.

Other Protocol Issues

Other artifacts in terms of patient size.

As the patient's body mass goes up, the quality of the images tends to go down.

So you can see that on these larger patients, it can be very difficult to see if this is all these little nodular areas here, these focal areas, this is what carcinomatosis looks like on PET and it's very difficult to tell whether this is actually pathology or just artifact.

Whereas with the thinner patients, it's pretty easy to see that the nice delineation of muscular activity and so forth.

In terms of other protocol issues, immobilization.

Generally all patients should be encouraged to remain still during the exam.

And as I think I mentioned in the first lecture, movement between the CT and PET portions of a PET CT will result in misregistration.

So generally we encourage using head holders and other devices to make sure that patients don't move between the exams.

So if you do the CT with the patient awake and they fall asleep and turn their head during the PET portion of the exam, you will have gross misregistration of the images.

So arm positioning is important as they can cause significant artifacts and obscure pathology.

In general, consider scanning the patient with arms up.

For most patients, unless they have head and neck carcinoma, then you want to consider doing them with arms down or do a double CT examination with the arms up first and then do a limited scan through the neck area.

And so this is the type of artifact that you get with the arms down.

You see some beam hardening going through the posterior aspect where the arms are located, same patient with arms up, and we see that that actually goes away.

Now in some patients it can be fairly significant like this where you have severe artifact from the patient's arms being at their side and that can introduce artifact into the corrected PET images from using the wrong attenuation values from these high attenuation streak areas.

Physiologic FDG Uptake in Abdomen and Pelvis

Alright, let's move on now and talk about some of the physiologic FDG uptake in the abdomen and pelvis.

Vaginal Canal

The vaginal canal can sometimes be problematic.

This was a patient that we did young woman for lymphoma, and this was she was originally done on a dedicated PET study.

And when I saw these two areas of activity, I knew that one of these is gonna be bladder.

Wasn't sure if there was a potential cervical mass or what was going on.

So we did two bed positions through the area of interest.

And we see that there's physiologic FDG activity within the vaginal canal.

She actually has a tampon in place.

So the differential for this type of activity would be menstruation.

So if they're actively menstruating, you inject them with FDG activity.

Some of that FDG activity will end up around the tampon.

Or when I presented this at the RSNA several years back, somebody said, well, couldn't that just be urine?

And I suppose urine could have a similar appearance around it.

So expiratory FDG around the tampon, this is a patient, again, similar appearance.

We see the urinary bladder and then we see some activity inferior to that.

And she is 53 years old and had several children and coughed several times during the exam.

And we see that this is stress incontinence.

So the differential, again, blood urine or much less commonly tumor endometrial activity.

Uterine Activity and Fibroids

The rule of thumb is in premenopausal patients, it's generally physiologic.

So this is a nice example, 29-year-old with otherwise fairly normal appearing endometrial canal.

There's focal intense FDG activity in the endometrial canal, and as long as you have a decent history that she is menstruating, there's no need to re-image her or bring her back.

However, in post-menopausal woman, very similar appearance on PET, although we see that the endometrial canal is expanded, there's some nodular lesions and this is endometrial carcinoma.

So premenopausal and menstruating, no need to follow up.

Postmenopausal is almost always going to represent endometrial carcinoma.

Other uterine activity and specifically fibroids.

Fibroids can have variable FDG uptake ranging from almost no uptake to fairly intense.

This is a patient with breast cancer and has hepatic mets and right internal mammary lymph node here, but there's a large mass in the pelvis.

And this is a large fibroid with very minimal FDG activity.

This is a patient with several fibroids, small ones that are FDG avid.

There are others that are mild to moderately FDG avid.

This one has intense FDG activity in general.

The only other thing in the differential would be a leiomyosarcoma, and this is my only case from probably over a hundred thousand cases.

So I don't think you need to raise this as a strong possibility.

With fibroids being so common, we generally will say that it correlates to fibroids on CT, particularly if it's a premenopausal patient.

Ovarian Activity

Ovarian activity, same advice.

So premenopausal we see symmetrical, almost symmetrical, slightly asymmetrical in the right activity, corresponding to otherwise normal appearing ovaries.

This is almost certainly going to be physiologic.

This is a case that I showed in the prior lecture with her primary lesion was a cervical carcinoma, and these were areas of brown fat as I showed you.

However, in her pelvis she had, this is her primary mass, and then she's got right ovary, left ovary with asymmetrical focal intense uptake.

Now, she was 33 years old at the time of the diagnosis and the person that read the exam read it as asymmetrical focal intense uptake in the ovary, suspicious for metastatic disease.

Because her five year survival, if she does have metastatic disease to the ovaries, is about 15% versus if it's localized to the cervix about an 85% survival.

They, before they even treated her, they did a laparoscopic resection of the ovary and this turned out to be a corpus luteal cyst, a hemorrhagic corpus luteal cyst.

So for her prognostic information, they even took it out before they treated her.

Other Benign Processes

Other benign processes.

So let's say this patient has a primary lung carcinoma and you see focal uptake in what appears to be a right adrenal nodule.

Well, the way that the majority of people are doing PET CT, it's either with intravenous contrast or without.

There are very few that are doing with and without.

So with intravenous contrast, this is essentially an indeterminate nodule.

If we had a non-contrast CT and we could measure the Hounsfield units, we could certainly raise the possibility of an adenoma.

So this was read out as a primary lung carcinoma with a right adrenal lesion with uptake suspicious for or compatible with stage four disease.

The patient had an MRI with it and out of phase imaging and showed that this was an adenoma.

So adenomas, just like thyroid adenomas, just like even colonic adenomas can have variable uptake anywhere from mild to intense uptake in general.

In general, adrenal adenomas will have mild to moderate uptake rather than intense.

And just a study unenhanced attenuation greater than 10 Hounsfield units, SUV max greater than 2.5, and the presence of mets in other parts of the body were highly correlated with metastatic disease.

However, like I said, if you have a non-contrast CT that's probably the best thing you could get.

Focal intense FDG activity in the colon, just like breast and thyroid should be further evaluated with colonoscopy.

This, however, the differential does include polyps or adenomas.

This turned out to be a colonic adenoma and as I mentioned in the first lecture with the acute intramural hematoma, this is actually a patient who had a car accident several weeks before their PET CT and has a perinephric hematoma that you can see this crescentic shaped perinephric hematoma that has intense FDG activity in it.

So for whatever reason, whether it's macrophage deposition or extravasation locally into this area, these areas can have fairly intense FDG activity.

Another patient here with focal colonic activity, although you might recommend a colonoscopy in this patient, there are some other findings that actually might give you the answer.

There's a tiny little extraluminal gas and some stranding.

So this is actually acute diverticulitis.

Same.

So you've got intense uptake your expiratory FDG in the bladder, and then you've got this area inferior to the bladder.

And as we saw, you can have this from stress incontinence or other etiologies, but one thing you need to consider is a cystocele.

So this is a large cystocele.

Difficult Cases

Okay, so we got about five or 10 minutes left.

We'll do a few more difficult cases.

Okay, here's the first case.

So same thing number one, cancer number two, benign.

Okay, so good, 87% did get correct answer.

So this could be lymphoma, but it has a pretty nice appearance, this soft tissue round around the aorta and around the proximal iliac vessels for possible retroperitoneal fibrosis.

So during the inflammatory phase, it can be very intensely FDG avid.

I read this case and suggested that as the diagnosis and they treated her with a course of steroids.

And here's the follow up after with steroids.

Looks like she's got hydronephrosis too here.

Yeah or Yeah, yeah, yeah, yeah.

She did have a little bit of hydro here.

Which is another common secondary process that occurs secondary to the fibrotic changes.

So this is a patient with, I'll give you that, lung cancer and I wanna know, is this cancer or not?

Okay, so 78 got the correct answer.

Obviously this has significant implications for the patient.

If they have a single metastatic lesion to the left iliac bone, they become non-surgical in stage four.

Paget's disease can have a variable appearance on PET imaging anywhere from no uptake to fairly intense uptake.

And it's important to look at the anatomical imaging correlation.

The cortex is thickened, the trabecula is also thickened.

I'm not giving you all the images, but this was Paget's disease.

How about this?

So this is multifocal HCC, so 86 percent got the right answer.

And yeah, so the important thing here is that this this is not certainly not a cyst, it's not a hemangioma, it's not a focal nodular hyperplasia.

So there really aren't that many etiology other things left.

HCC generally has variable uptake.

About 50% of them may have very little FDG activity.

So pretty good appearance for multifocal HCC.

How about this?

I'll give you the history.

The patient had a right hemicolectomy about six or eight weeks before this.

Great.

Okay, so 91% got the correct answer.

And this is an area of omental infarct.

We see the omentum here is fatty attenuation, and during the acute phases this will be FDG avid and over time will resolve.

So that's good.

How about this?

Okay, good.

So 85% got the correct answer.

So we see speculated mass in the small mesentery with some calcifications only sort of moderate FDG activity.

This would be compatible with carcinoid, although other things could certainly give you this appearance such as treated lymphoma.

Although we wouldn't expect there to be uptake in treated disease.

Fibrosing mesenteritis.

Okay.

How about this patient had radiation to the pelvis?

That's the history.

Okay, so a hundred percent got the correct answer.

So this is we see the vagina here, we see the rectum here, and there's essentially communication between the vagina and the rectum.

So this is a rectovaginal fistula, which certainly can give you inflammatory changes in that area.

Patient had a history of lung cancer.

Okay, so actually only 21% got the correct answer in this case.

So interestingly the person that read it at UPMC got the correct actually read the CT was got the correct diagnosis.

So this patient did not have cirrhosis about a year and a half before this.

And we see that there's these areas almost rounded areas within the liver.

Some with very minimal FDG activity, some with more moderate areas of uptake.

And she had essentially a rapidly progressive cirrhosis.

There aren't that many things in a patient with breast cancer.

You really need to consider pseudo cirrhosis.

So whether she was sometimes is described as post-treatment changes or from the mets themselves actually causing the desmoplastic reaction.

How about this?

No history of malignancy, but a history of Cushing's.

So essentially the patient had a surgical note that they took out both adrenal glands.

The right adrenal gland measured almost seven centimeters and this was done a couple years later.

And we see that morphologically, this looks almost like a very giant adrenal gland intense uptake.

This is a case of massive adrenal hyperplasia.

This is actually this was benign.

How about this?

So we have a cystic lesion in the anterior aspect of the left kidney with some nodularity.

This is a contrast enhanced study.

There's some enhancement of that nodule.

Even though there's very little FDG activity, this is cystic renal cell carcinoma.

Okay?

This patient has a primary lung carcinoma and potential metastatic lesion on the axial PET CT.

You see that this is somewhat linear and although we don't see an actual fracture line, the patient had an MRI, which did show a fracture line.

So this is a sacral insufficiency fracture.

This patient has a history of breast cancer and this is a follow up PET CT.

Good.

So yeah, so this is another one of those cases where it really doesn't make a whole lot of sense.

The method of spread just it seems very odd to not have hepatic disease, osseous disease, but to have extensive splenic mediastinal and upper abdominal adenopathy.

This is a rare case where we actually did biopsy the spleen, and this was granulomatous disease.

Okay, thank you.