0:00

Moving on to the technique. How do we obtain great images for a CTA

0:04

runoff?

0:06

So I like to think of it as what images are we going to get from the scanner

0:10

that we need for interpretation. Any CT scan starts with

0:13

obtaining a scout image or a topographic image, which looks like an X-ray like

0:17

this, and we use this for planning.

0:20

The first set of data set that we obtain routinely at our

0:23

institution is a non-contrast image.

0:26

Commonly, the field of view extends from the dome of the diaphragm to

0:29

the tip of the feet, and this is obtained without

0:33

administration of any contrast.

0:36

Next, we administer contrast using a bolus triggering technique.

0:40

We give at least four cc's per ml as the rate for injection.

0:45

Once we do a bolus triggering technique, we place a

0:48

region of interest in the proximal abdominal aorta typically and

0:52

place a region of interest within the aorta.

0:56

Once contrast is administered, there is a 15-second delay just

1:00

to make sure that the contrast reaches before it starts scanning the slice over and

1:04

over again. And then once an attenuation of 150

1:07

Hounsfield unit is reached, the scan is triggered, and after

1:11

an additional delay, a five-second delay to make sure that the contrast has

1:15

reached at least up until

1:18

the lower extremity arteries, the scan is then obtained in an

1:22

arterial phase. Now in order to optimize

1:25

our image quality, the arterial phase images are obtained

1:29

at a lower kVp so that we're closer to the K edge of the iodine.

1:33

This is now more intuitive because the modern scanners have an

1:37

automated tube voltage selection and as long as the

1:41

parameters indicate that this is a vascular task, it tries to pick

1:45

the best and the lowest tube voltage that would be diagnostic in

1:49

a patient. So obviously if it's a large patient, it's not going to

1:53

hamper your diagnostic quality by picking a higher kVp because it

1:57

wants to make sure at least that the images are going to be diagnostic.

2:01

So once that threshold is obtained in that monitoring slice of

2:05

150 Hounsfield unit, we obtain an arterial phase image again from that

2:08

same dome of the diaphragm to the feet.

2:11

And then we go back to the top so that there's some percentage of

2:14

overlap and scan again from the knees down to the feet.

2:18

Now why do we do a non-contrast scan?

2:21

We do it because sometimes it can be difficult to differentiate

2:24

contrast from low-density calcium.

2:26

So this is the same left lower extremity in the patient, and this is a

2:30

non-contrast image, and you can see that there is circumferential calcifications

2:33

along this anterior tibial artery.

2:35

If I was just reading the arterial phase without the non-contrast, it's

2:39

tempting to just say everything is patent and they don't have any disease at all.

2:43

But we know now from the non-contrast scan that they do have a significant amount

2:46

of circumferential plaque.

2:49

In the post-surgical setting, it helps problem solve between the presence of

2:52

surgical material and helps also assessment for active bleeds.

2:57

What's the rationale of immediate delay?

2:59

Now a lot of people think that these images come at the cost of a high radiation

3:03

dose, considering we have a non-contrast phase, an arterial phase, and then an

3:07

immediate delay phase. But really the benefits outweigh the risk,

3:11

and most of these patients tend to be older patients where radiation

3:15

dose isn't particularly a concern.

3:17

This also involves scanning non-radio

3:21

sensitive areas, and so when you account for tissue weighting

3:25

factor, the burden of radiation risk is really not high.

3:30

The rationale for obtaining an immediate delay is to basically

3:33

circumvent the problem of outrunning the bolus.

3:36

So the modern scanners are really, really fast, and

3:40

by the time we reach down to the toes, sometimes the contrast doesn't reach

3:44

all the way up until the feet arteries.

3:47

This is accentuated also in patients who have poor cardiac output.

3:51

And so having an immediate delay gives time for the

3:54

circulation and the contrast to reach up until the level of the knees, and we can

3:58

follow the contrast bolus. So for example, in this patient, if one were reading

4:02

just the arterial phase images, they might think that there is

4:06

a significantly diseased calf arteries, all of them basically.

4:10

But because we have an immediate delay, we know that these are widely patent and we

4:14

just outran the bolus in that first pass of acquisition.

4:18

The clues to diagnosis of this is that there's typically a natural

4:21

gradation of the vessels. If multiple vessels are

4:25

involved, then that's another clue.

4:27

Looking at the contralateral limb is also helpful to

4:31

see if it's an outrunning of bolus versus a genuine

4:34

thrombosis.

4:36

Which images do we send to PACS? So typically, we send our non-contrast

4:40

images in PACS at two and a half by two and a half millimeter slice

4:43

thickness and interval. We also send thin slice images, so 1.5

4:47

millimeter with a 0.5-millimeter increment.

4:50

We send arterial and immediate delayed phase images at the

4:54

same slice thickness and slice increment as the non-contrast image.

4:58

And in addition, we send small field of view images of

5:02

individual legs, so the right leg only for the arterial and the

5:06

delayed phase, and then the left leg only for the arterial and the

5:09

delayed phase.

5:11

These are typically always sent from the scanner to our PACS

5:15

system, and our aim is really to have the highest

5:19

balance of spatial resolution to signal-to-noise ratio, and the

5:22

slice thickness choice really helps affecting that, and we're going to see a few

5:26

examples in the next few slides of why that's important.

5:30

So the rationale for thin slice is an appropriate matrix.

5:34

Here, this is a larger matrix, and you can see that we have both the

5:38

legs in that imagery construction, whereas in this one, we only have a single leg

5:42

in this imagery construction. And really, the spatial resolution is

5:46

so much more better for the smaller field of view

5:49

image versus the larger field of view image.

5:53

And that's why when interpreting the calf artery, sometimes just looking at the

5:56

small field of view images would be a lot more beneficial than trying to interpret

6:00

them in the large field of view images

6:03

What are the other image reconstructions that we send?

6:06

So we send a maximum intensity projection images,

6:10

which means that the brightest voxels in the image are

6:13

interpolated to a predefined slice thickness, and then those

6:17

can be laid out. This helps like a good one image

6:21

visualization of the entire vascular tree.

6:25

We can even do bone subtraction, as was done over here.

6:28

The 3D lab then can also work on making volume rendered

6:31

images. They can suppress bone and just highlight assessment of the

6:35

arteries.

6:38

The reconstructions that we don't typically routinely use include

6:42

the curved planar reformats as well as the straight-line luminal

6:45

reformats. So in curved planar reformats, we're

6:48

basically placing seed points and the entire

6:52

vascular tree can be obtained in this single mid slice, and it's laid

6:56

out. It helps specifically if it's a tortuous vessel.

7:00

The straight-line luminogram, again, kind of has the same purpose, but instead of

7:04

having that curved view, the vessel is

7:07

just outlined as a straight line, and we can assess for

7:11

patency and stenosis and thrombosis and measure lengths in a

7:15

much more easier fashion. But these are not typically sent, and we don't

7:19

typically use them, but it's good to know that we have these in our toolbox if we

7:22

need them for interpretation.

7:24

Now, while

7:26

we obtain images, a lot of the times we are bogged down by

7:30

artifacts. The most common, especially because they're elderly patients, is

7:33

having prosthesis. So photon starvation and beam hardening from metal

7:37

implants, embedded projectiles, or other metallic devices

7:41

can cause these streak and beam hardening artifacts,

7:45

which are then also accentuated at the lower kVp images, which is what

7:49

we typically tend to do in this vascular task where we want to

7:53

assess the arteries.

7:55

So we use the newer metal artifact reduction algorithms that can

7:59

improve inconsistencies in the data, in the projection data space,

8:03

and can result in less artifacts and help kind

8:07

of unravel a lot more of the anatomy that we can

8:10

confidently assess. So this is what the same patient looks like

8:14

without MAR and with MAR, and we're going to play both of them side

8:18

by side to see what effect they had.

8:21

So even just kind of starting out, we can see that this metal artifact

8:25

is obscuring assessment of the contralateral as well as the ipsilateral

8:29

popliteal artery in this patient.

8:30

You can see over here how this popliteal artery is a lot more

8:34

obscured, whereas with MAR, the degree of artifacts is

8:38

much less,

8:40

and assessment of this popliteal artery is a lot more,

8:44

and there's not as many streak and beam artifacts that are kind of hindering its

8:47

pathway.

8:49

In terms of looking at that same ipsilateral popliteal artery, so this is

8:53

without MAR, and this is with MAR, and you can see on this sagittal view

8:57

that a lot more of the popliteal artery length is

9:01

obscured by the degree of artifacts by the knee

9:04

prosthesis. Whereas with MAR, you can confidently

9:08

interpret that the popliteal artery is patent for a longer

9:12

segment, and just a short segment remains

9:14

uninterpretable. So it's not perfect, and it's not going to completely

9:18

eliminate your artifacts, but it does help reduce them quite

9:22

significantly. Next is dual energy and

9:25

photon counting detector systems for which we can leverage

9:29

many material decomposition techniques to enhance the

9:32

quality and diagnostic utility of the CT

9:35

angiograms. You can create mono energetic imaging, so the closer

9:39

you are to the K-edge of iodine at 40 keV, you can see

9:43

that the column of contrast looks much brighter.

9:46

Whereas at a higher keV, like 100, the contrast is

9:50

much duller. This also comes at the expense of noise.

9:54

So the lower the keV, the brighter the contrast, but higher the image noise.

9:58

The higher the keV, the lower the contrast, but lower the image

10:02

noise. And this property, in an ideal patient, you could also

10:06

use to reduce the amount of contrast volume that might be administered to these

10:10

patients and can facilitate lower contrast media dose protocols,

10:14

especially helpful in patients with chronic kidney disease.

10:18

Now, though the application of material decomposition and

10:21

contributions of iodine to each voxel can be subtracted to

10:25

create virtual non-contrast images, you always want to make

10:29

sure that you look at source data to not be fooled by any

10:32

artifacts. You can also create pure calcium images that

10:36

overlay all the calcified plaque burden in the artery.

10:39

You can create a pure lumen or an iodine map where just the

10:43

iodinated contrast is visualized, and so this gives a very

10:47

nice problem-solving way, especially if someone didn't have a

10:50

non-contrast image, to really just see what exactly is calcified plaque and

10:54

where exactly is the lumen. And it can also serve as

10:58

another strategy for radiation dose saving in that if you are

11:02

able to create pure calcium and virtual non-contrast images, you might be able to

11:05

forgo acquisition of a non-contrast

11:08

series. We're now going to look at different slice

11:12

thickness and different detectors and their effect on assessment of the

11:16

lower extremity artery. So this was a scan in the same patient.

11:19

Once it was scanned on a photon counting detector, and the other time point it was

11:23

done by the conventional energy integrating detector.

11:26

They both have the same slice thickness as well as a similar

11:29

field of view. But you can see, because of the inherent

11:33

construct of the photon counting detector, whereby they have

11:37

sharper reconstructions, smaller detector elements, absence

11:41

of septae, the visualization of especially the

11:44

smaller branches are much more conspicuous on the photon

11:48

counting detectors images compared to the energy integrating

11:52

detector images.

11:54

This is, again, the same patient scanned on the photon counting

11:58

detector, and I'm just going to show the ultra-high-resolution

12:02

reconstruction compared to the standard reconstruction from that same data set.

12:06

So this is a 1 x 0.8 millimeter slice thickness, and this is 0.2

12:10

x 0.2, again, with a relatively similar field of view.

12:14

Because of the absence of the septa and

12:18

the smaller detector elements, again, the fine detail is even

12:22

more clarified on the high-resolution

12:26

image data set compared to just the regular photon counting detector CT.

12:30

So really, we have to kind of balance all these different

12:33

pieces of acquisition parameters when we're tailoring our

12:37

scan to get the biggest utility from that scanner for

12:41

that patient, for that diagnostic technique.

12:44

Photon counting detector has enabled ultra-high-resolution imaging that

12:48

has up until now not been possible, going all the way down to 0.2

12:52

millimeters.

12:54

And the reason why that becomes interesting is because in the same person, again,

12:57

scanned on both the photon and the energy integrating detector, you can see

13:01

how here you can see that there's just this calcified plaque

13:04

along the popliteal artery. Whereas here you can really better

13:08

characterize this plaque and assess that there's a lot more of this

13:12

non-calcified component plaque.

13:13

So from a therapeutic standpoint, from a research

13:17

standpoint, plaque quantification and radiomics may be able to be

13:21

improved with this photon counting detector technique with better

13:24

visualization and assessment of the vessel wall and plaque.

13:29

Now, coming to slice thickness. So same patient underwent a

13:32

scan, and we have two different kinds of reconstructions, one at 0.4

13:36

millimeter and another at one millimeter, and this is just the zoomed-in

13:40

images of the same.

13:42

This is the origin of the anterior tibial artery here, and you can see that

13:46

there's a big chunk of calcified plaque at the ostium of the anterior tibial

13:49

artery. Now, because of the blooming artifacts from the

13:53

calcified plaque, even though this is a thin slice images, you can see

13:57

how one might be tempted to say that this is a high-grade stenosis of the

14:01

anterior tibial artery. Whereas in the 0.4 millimeter, you can

14:05

see that there's a lot more of the contrast column in the anterior tibial

14:09

artery. And looking at it in a different plane, in an axial view,

14:13

instead of calling it a severe stenosis of the anterior tibial artery, I think

14:17

we're more confidently able to say that there's just a mild amount of

14:20

stenosis. So because of the reduced blooming, it

14:24

increases reader confidence, improves accuracy, and really the

14:28

sharper image and the lumen clarity is unparalleled

14:32

with lower slice thickness images, as well as with photon

14:36

counting scanners.

14:39

Kernel selection, again, is important.

14:40

Now, even though one might have a photon counting scanner, if you don't

14:44

appropriately set up your protocols, you might set yourself up for failure.

14:48

So again, on just a photon counting scanner using just the smooth

14:51

kernels, one might think that this is just calcifications.

14:55

In fact, when you did automated

14:58

voluminated rendering, it just looks like calcifications.

15:01

But if you reconstruct this with a sharper kernel, even though it comes with an

15:05

edge enhancement fallacy, you can see that this was actually a

15:09

stented segment rather than calcification.

15:12

And obviously the automated volume rendered imaging also was

15:16

fooled based on that kernel selection and would've just thought that this might

15:20

all just be calcifications as opposed to a stent construct.

15:24

Therefore, thinner slice images, sharper reconstructions are

15:28

particularly helpful, in my personal view, for looking at the

15:32

infrapopliteal arteries. Even though the images might look a little

15:36

plasticky, you might take a little bit of time to get used to looking at.

15:39

There are noise reduction filters and sharper kernels which can help with this.

15:44

Now, despite us getting the right kind of imaging and all the necessary

15:47

reconstructions, there are still certain pitfalls that remain.

15:52

One is blooming from the dense calcifications that might limit your luminal

15:55

assessment, or you might overestimate the degree of stenosis.

15:58

Widening your window width and level is sometimes helpful, but sometimes

16:02

you're just not able to, and an MR angiogram potentially

16:06

offsets in patients with heavily calcified arteries,

16:10

particularly for calf arteries. You might have venous contamination, which may

16:14

again limit assessment just a little bit, at least of the

16:17

infrapopliteal arteries, because each calf artery comes with paired

16:21

veins, and so kind of parsing that one artery

16:25

in the middle of two veins which are opacifying, you might miss

16:29

significant stenoses.