Slides
Transcript
Alright, yes, so I’m AJ. I’m from the Mayo Clinic in Rochester. So, I’m going to be talking about photon-counting CT myelography and how it can help us. I don’t have any disclosures. I do want to thank several people, both my colleagues at Mayo as well as several of my mentors at Duke, who really helped me get this practice started at Mayo early on a couple of years ago. So, we’ll kind of go over what photon-counting CT is, how it’s different from traditional CT, and then we’ll go on to how it can help patients with CSF leaks.
I don’t want to go too much into the physics of this. I’m not a physicist, and I have a very rudimentary understanding of all this, but this is basically how a CT scanner works. So, if you’ve ever had a CT scan, you have some understanding of this. You go into this tube, you’re lying on a table, and what the tube contains are these x-ray sources on one side and x-ray detectors on the other side. Basically, the x-rays are emitted from the x-ray source. They go through the patient, and those x-rays interact with the atoms in the patient. Then more x-rays are emitted from the patient, and they hit the detector. Based on the information the detector gets, it’s able to create a cross-sectional image of the patient.
So, how’s that different from photon-counting CT? Well, most of it is actually the same. The x-ray source is all the same, the tube is the same. Really, the only difference is the detector and how the x-rays are detected. In traditional CT scanners, which are called Energy Integrating Detector CT, the x-rays encounter the detector and then the x-rays through a two-step process, are first converted to visible light photons and then they’re converted to an electrical signal. As part of that process, the light photons have to get summed up, so you don’t actually count every photon that hits the detector, and so you lose some energy information there.
What’s different about photon-counting CT is that the x-rays hit the detector and are immediately converted to an electrical charge, and that charge is read out by the detector. Every photon gets counted so we get more information, and that results in a number of different advantages that produce better images, and we’ll go over how that helps.
There are really three main things that I think help with regard to myelography, and probably the single biggest one is that photon-counting CT gives us better spatial resolution. We can create thinner imaging slices down to 0.2 millimeters. That compares to about 0.6 millimeters on most traditional scanners, so we get better resolution.
So how does that really help? Here’s a patient I had a couple of months ago, a very nice guy from Iowa. He had had eight years of headaches. This is his brain MRI. You can see it’s very abnormal. What Dr. Callen mentioned earlier about the radiologist reads mattering is really true. His brain MRI was initially interpreted as negative many years ago, and so he wasn’t actually brought to us for a long time. We finally got him to us, and he came through in an interesting fashion. He actually had a meningioma on his spine, a benign tumor, and that’s ultimately what brought him to Mayo and got him into our system. We got him in for a myelogram.
This is an image from his photon-counting CT myelogram. I’ll go through it here. It’s a little bit slow, but you can see this is a very clear CSF-venous fistula. This is the diverticulum here, and these are all the little veins that are draining the fistula, and you can see them with very high resolution. But these aren’t the cases where it really helps us. This is a very obvious CSF-venous fistula. You would see this on a DSM, and you would see this on an EID CT myelogram without any problem.
This is another patient – very similar-looking brain MRI. This was a 50-year-old woman with a very abnormal brain MRI. She had progressive dementia, and that was just the manifestation of her CSF leak. This is her photon-counting CT myelogram. These are these 0.2 mm images, and what you can see here is that there’s a little vein here, right there, in the internal epidural venous plexus. It’s basically right next to the bone, and that was the problem. You get these little veins that are right next to the bone, and if you don’t have high spatial resolution, they’re very hard to discriminate. This is where the high spatial resolution really helps.
What I found is that it doesn’t really matter how conspicuous the fistula is on the myelogram. The patient can be equally symptomatic. They may have very marked brain MRI findings, or they may have no brain MRI findings at all. It continues to be important to improve our myelographic techniques to find these fistulas.
This is another similar example. This is another patient where we scanned her on a photon-counting CT myelogram. You can see there’s a diverticulum here. So you can see there’s a diverticulum here, and right above it, there’s a little vein on these 0.2 mm images. The reason it’s hard to see is because the vein is right next to the diverticulum, so you just need good resolution to see it, even on a different image with very high resolution you couldn’t see the fistula. We really needed the 0.2 mm image to even see that finding, and so that’s where the high spatial resolution really helps us. Spatial resolution isn’t all about slice thickness. These are both 0.2 mm images from another patient, but it turns out that the way you reconstruct the image matters too.
We use these things called different imaging kernels that confer spatial resolution to the image. On the left here, this is the same slice from the same patient, but on the left, this is like a smoother kernel that we reconstructed with, and on the right, this is a sharper kernel. There’s this vein here, right next to the bone, that you can see here. It’s really hard to make out on that smoother kernel. On the sharper kernel, it’s much more obvious that there’s a distinct vein there, compatible with a CSF-venous fistula.
The thing is, it’s harder to make these sharper kernels on conventional CT scanners because the sharper you make the kernel, the image then a little bit more noisy and grainy. You can kind of appreciate that here. But the nice thing about photon CT is it mitigates the noise, so we can get away with using these sharper kernels and really maximizing our spatial resolution.
Another thing that’s really good about photon-counting CT is it has slightly better temporal resolution. What I mean by that, in this context, is you can scan the spine very quickly, and you can perform scans back-to-back for these dynamic CT myelograms that Dr. Carlton Jones very nicely described. That helps us too, as has been shown.
This is a patient who had a CSF-venous fistula that I could see on this initial scan that we did. I’ll try to scroll to the fistula here – it’s this vein here. I scanned that patient on this dynamic exam just 10 seconds later, and you could no longer see the fistula. Given that you have this image, you could probably make out, “Oh, that’s probably where the fistula is coming from,” but if I didn’t have both of these scans, I probably wouldn’t have called it on this image. Sometimes that high temporal resolution really does help us.
Moving on, and this is just kind of a counter-example – sometimes you see the fistula better on the later scan. This is one where I didn’t really see anything on the first scan I did, and then on another scan 45 seconds later, there’s this very subtle CSF-venous fistula going into the paraspinal vein.
Okay, and so the last thing that helps us about photon-counting CT is that it confers spectral imaging. Spectral imaging just means that we can detect the energy levels of the photons that are hitting the detector, and that helps us do these really cool things. One thing you can do with spectral imaging is you can make the contrast look even brighter than it actually is. So when we’re looking for really subtle CSF leaks, that spectral imaging can really help us.
This is an example of what I’m talking about. So, these are four images from the same patient, same slice. The only difference is this top left image – we’ve taken advantage of the spectral imaging to make the contrast look even brighter. You can see that there’s a little paraspinal vein here filling up with contrast, so that’s a very clear CSF-venous fistula. It’s harder to see when you reconstruct the image at different energy levels when you’re not capitalizing on this benefit of making the contrast look exceptionally bright. So, this is one where the spectral imaging really helped us.
This is another example. This was a young woman with a paraspinal venous malformation that you could see on her MRI. Sometimes, you get these congenital venous malformations next to the spine, and that can actually cause CSF-venous fistulas to occur. On the high-resolution image, we could see the fistula pretty well, but it was only on this spectral imaging, where we made the contrast look brighter, that I could see there was some contrast going into the venous malformation itself. That confirmed to me that the venous malformation was what was actually causing this fistula, and that affected how we treated this patient.
A question I often get is, “Can you actually use photon-counting CT to find leaks that were missed on other techniques?” The answer is, sometimes you can, and we’ve had many cases like that. This is an example of a patient who we initially did a digital subtraction myelogram on. I’ll just scroll through the images to show that there’s really no fistula there. There was this prominent bulb diverticulum here that we could see, but we never saw a vein fill. So, we repeated that study on photon-counting CT, and again, you can see that same bulb diverticulum here that we can wee, but we never saw a vein fill. So, we repeated that study on photon-counting CT, and again, you can see that same bulb diverticulum here.
As I continue to scroll down, what’s also present is that there’s a vein in the internal epidural venous plexus that’s filling up with contrast. And I continue to go down, and you can see it coming out laterally here, and I think you see that just to a little bit better advantage on the coronal image. You can see this is the bulb diverticulum – this is that vein. And so, you can imagine how without having cross-section imaging and high resolution, that would be a very hard fistula to appreciate.
This is another example that Dr. Amrhein gave me. This was a patient who initially had an EID CT myelogram, a very good quality study, and they didn’t see a CSF-venous fistula. So, they repeated the exam on photon-counting CT. You can see that same diverticulum filling up, but in this case, on the PCD CT, there’s also a very thin vein that’s filling up with contrast, making the diagnosis of the CSF-venous fistula.
So, can you use it for leaks other than fistulas? That’s kind of all I’ve shown you so far, and you certainly can. This was a patient with this big extensive epidural fluid collection, so we know they have a dural tear somewhere – just a matter of finding it. We initially started with a prone dynamic CT myelogram using our EID scanner. We didn’t really see anything. I felt like this was a very good quality study. There was a lot of contrast at the spot where the leak ended up being, but we couldn’t get anything to leak out in the epidural space. When we repeated that on photon-counting CT, we saw this thin jet of contrast compatible with a ventral dural tear. So, that’s one where I felt like the resolution really helped. You know, if you look back at the EID CT, you can make out there’s probably a little bit of epidural contrast here, but it’s just so hard to differentiate from the adjacent bone without the really high spatial resolution.
This is that same patient, just looking at the sagittal image, and you can nicely see the leak here on photon-counting CT. I’ve started to do most of these on photon CT now, mostly because it helps our surgeons a little bit. We send a lot of these patients to see our surgeons, and they like to be able to see with high resolution exactly where the leak is coming from. So, you can kind of appreciate this one’s a little bit off to the right of midline, for example.
This is one where the temporal resolution really helped on photon CT. So, two scans from a prone dynamic CT myelogram. On the first one, you can see this ventral dural tear here, and then just 5 seconds later, you can no longer see the contrast flowing through that dural tear. So, it really helped us localize this leak accurately.
The last thing I would say, and I think this is one of the very important things, is it’s really not just all about imaging. I truly believe photon-counting CT helps a lot. I think it’s helped us make diagnoses in patients that we would have otherwise missed, but there is definitely more to it.
Here’s an example of that. This is one patient I had maybe a year ago. He had a normal brain MRI but a pretty sick guy, and he had a very classic presentation for SIH. You can see, I’m just going to show you his DSM looks pretty normal. You know, it was a good quality study. We used a lot of contrast. We didn’t see a CSF-venous fistula. So, we brought him back a week later and repeated the study on photon-counting CT, and so this is that exam. You can see that there is a very clear CSF-venous fistula here that, for some reason, wasn’t seen on the DSM. But the thing is, I can’t explain why I didn’t see this on the DSM. It’s not a subtle fistula – it should have been visible on the DSM. So, I think there’s something more to it. In a week’s time, this fistula seemed to open up, and for that reason is why I think we saw it on the photon-counting CT.
And, you know, I’ve had examples of just the opposite too. This was a patient from just a couple of weeks ago. I initially did a photon-counting CT on her, and you could see this big diverticulum here that filled very nicely with contrast, but we didn’t see a venous fistula here. This ended up just being a bod component to that diverticulum. But we knew she had a leak, so I just brought her back a week later and did a DSM. You can see that there’s a somewhat subtle but definitive CSF-venous fistula here on the digital subtraction myelogram. And just to prove it, I did a cone beam CT at that level, and you can see the vein here – the paraspinal vein filling with contrast. So, no doubt that’s a very real CSF-venous fistula. Again, I can’t explain why that wasn’t visible on the prior studies.
So, at the end of the day, I think we have to do everything we can to optimize our imaging techniques while recognizing that there are some factors that we don’t understand, and we have to continue to research those and pursue more knowledge about what will increase our odds of finding these fistulas.
Just to conclude, I think photon-counting CT is extremely promising. I think the main reasons it’s helpful are because it has higher spatial resolution, better temporal resolution, and these spectral imaging qualities that make it very easy to make contrast a little bit brighter. But keep in mind that there are other things we have to continue to study too, and I think that’ll continue to maximize the yield of these tests.
And I’ll stop there. Thanks so much for your attention.