0:01

An important concept in cardiac CT imaging is temporal

0:05

resolution, which, quite simply, is the shutter speed.

0:12

So what is shutter speed?

0:14

Well, imagine you're on a train and

0:16

you're taking a picture of a tree.

0:18

The train is moving.

0:22

The shutter speed is how quickly your

0:23

camera takes a picture of the tree.

0:27

Keeping that same analogy, if you slow the

0:29

train down for whatever shutter speed you

0:32

have, you get a better picture of the tree.

0:34

So in a similar way, if you slow the heart rate

0:38

down, you get better use of your temporal resolution.

0:41

But the temporal resolution, shutter speed,

0:44

is essentially dependent on how quickly

0:46

the gantry, which is the CT gantry, rotates.

0:50

So it depends on the gantry rotation speed,

0:54

and therefore the gantry rotation time.

0:56

One of the interesting things about CT,

0:59

which is in contrast to MR, is that in CT,

1:05

temporal resolution and spatial resolution

1:08

are generally not related to each other.

1:12

Whereas in MR, they are directly related.

1:14

Not necessarily temporal resolution, but acquisition

1:16

time, because in MR, to improve the spatial

1:20

resolution, we need to increase the matrix size

1:23

which means we need to increase the scan time.

1:26

Thankfully with CT, we don't have that, and

1:29

that may be one of many reasons why CT

1:31

coronary took off, whereas MR coronary didn't.

1:36

So what are the factors affecting

1:38

the temporal resolution?

1:39

Well, as I mentioned, the most important

1:41

factor is the gantry rotation time.

1:44

The gantry has the detectors and has the X-ray source.

1:50

The X-rays go through the body, and the detectors

1:53

on the other side are looking, but the gantry only

1:58

needs to rotate halfway in order to see the whole

2:02

body, since some detector or the other would

2:05

have seen one part of the body and they can kind

2:09

of interpolate or extrapolate the other part.

2:13

So the actual temporal resolution is

2:15

gantry rotation time divided by two.

2:17

In reality, it's slightly more than two.

2:21

Because you have this fan beam that you need to take

2:23

into consideration, which is about 30 degrees, so you

2:25

have to factor in a rotation arc of about 210 degrees.

2:30

But essentially, to keep things simple,

2:33

it's gantry rotation time divided by 2.

2:37

Gantry rotation times used to be a

2:40

lot, and thanks to smart engineering,

2:44

gantries are getting faster and faster.

2:46

The rotation time is getting faster and faster.

2:49

Lower and lower.

2:51

So we're talking about 260 milliseconds now as

2:55

the best, but yes, it can go up to about 500.

3:00

The second thing is whether you have one

3:05

source of X-ray, one source of X-ray, and

3:09

therefore complementary detector or two.

3:12

The two sources are two X-rays.

3:16

At right angles to each other, and what you do

3:17

there is, instead of having to go through 180,

3:21

you only need to go through 90, and between one

3:25

source and the 90-degree place, the other source

3:28

going through 90 degrees, and you can just kind of

3:30

imagine it in your mind, you can see the whole body.

3:34

With a single source, you need to go through 180,

3:37

because once you've seen it through one end,

3:38

you don't need to see it through the other end.

3:41

With dual source, you need to go through at 90 because

3:43

you have one detector that sees 90 degrees, the other

3:46

detector sees 90 degrees, and together, they see everything.

3:51

So, the third thing is something that

3:56

probably isn't used as much these days because

3:59

the temporal resolution, by virtue of the

4:01

gantry rotation speed, has improved so much.

4:05

And that's called multi-segment reconstruction.

4:07

But I think it's worth understanding because it really

4:10

gets to the heart of multiple things in cardiac CT.

4:16

So what happens with a single-segment reconstruction,

4:18

which is the way to do it without invoking more

4:25

than one cardiac cycle, is that the gantry rotates and

4:33

CT sees one element of the heart in one

4:37

phase of the cardiac cycle in one cycle.

4:40

So let's say the left main, it sees 60 percent

4:43

of the cardiac cycle in one heart interval.

4:46

With multi-segment, what the gantry

4:48

is doing is it's saying, "Okay."

4:53

I have one detector that looks at the left main at 60%

4:56

of the cardiac cycle in heartbeat 5, and I have

5:00

another detector which looks at the left main origin

5:05

at the same phase of the cardiac cycle in heart rate 6.

5:08

So we'll make each of these just go through a partial

5:14

arc and improve our temporal resolution accordingly.

5:18

So, it takes that whole concept of partial scanning

5:25

and extends it to another level.

5:28

And it's able to do this because each part

5:32

of the heart, when the pitch is low enough,

5:36

each part of the heart in each phase of the

5:38

cardiac cycle is seen by more than one detector.

5:42

So there's information that we can use, and we

5:46

can just use what one detector has seen through a

5:50

small arc, another detector seen through another

5:53

small arc, but splice the two arcs together.

5:56

Okay.

5:58

And that's essentially what you're doing, and you

6:00

see here that with single-segment partial scan

6:07

reconstruction, one heartbeat gives you one part.

6:12

Next heartbeat you go down, next heartbeat

6:15

you go further down, so you're kind of

6:16

going through the anatomy in three different

6:19

heartbeats, same part of the cardiac cycle.

6:22

With multi-segment reconstruction, you're now

6:25

using those three heartbeats and three different

6:30

detectors to reconstruct the same place.

6:35

That's the key difference.

6:37

You can do that using two, and you use

6:40

that doing three or four heartbeats.

6:42

There's obviously a limit to how many you

6:45

can use because you only have so many

6:47

detectors, and you can only go so slow.

6:50

But what it does is it reduces the temporal

6:53

resolution by a factor, and that factor is the

6:56

number of heartbeats used to reconstruct the image.

6:58

So if it's two, divide it by two.

7:00

If it's four, divide it by four.

7:03

So, the first thing that will

7:05

be obvious over here is that.

7:08

For this to work, the heart rate needs to be

7:11

very, very regular, because if the heart rate

7:15

is not regular, what you think is 60 percent of

7:18

the cardiac cycle in beat one, 60 percent of the

7:21

cardiac cycle in beat five may be very different.

7:23

And for those to be different simply means

7:26

they're at different phases of contraction

7:27

or relaxation, and therefore when you stitch

7:30

them together, you get misregistration.

7:32

So that's one problem.

7:33

So it has to be very regular.

7:35

The second is that for this to be useful,

7:39

the heart rate needs to be a little bit fast.

7:42

You know, in the order of 80 to 90, or even

7:45

70 to 80, um, because the pitch is low.

7:49

And if the heart rate is slow and the pitch is low,

7:52

we're talking about dramatically increased scan times.

7:56

Does it make a difference?

7:59

It definitely makes a difference.

8:00

It makes a difference both in terms of increasing,

8:04

you know, freezing the heart, you know, that's a term

8:06

used for having a high enough temporal resolution

8:09

that you can overcome the motion of the heart.

8:13

But most importantly, creating an image without

8:16

stair step artifacts, creating a more uniform image.

8:20

Because whenever you have stair step artifacts, there

8:22

is a tendency to overcore stenosis at that stair step.

8:28

So yes, it definitely makes a difference, again, going

8:31

from single on the left to multiple on the right.

8:37

And all of this, I think, gets to a very important

8:40

physiological point with cardiac imaging, which

8:44

is that the diastolic component of the cardiac cycle,

8:50

which is a component where the heart is moving less.

8:55

There are certain parts of the diastolic component of

8:58

the cardiac cycle where the heart moves even less, but

9:02

as a general rule, the heart moves less in diastole than in

9:05

systole, of course, because systole is contracting.

9:07

The diastolic component of

9:09

the cardiac cycle is variable.

9:14

When the heart rate is low, the diastolic component

9:17

is longer relative to the actual heart rate

9:22

or RR interval when the heart rate increases,

9:25

the diastolic component shrinks and it's the systolic

9:28

component that is relatively fixed, and you can see

9:31

in the graph there how the length of the diastolic

9:35

component varies with the heart rate, and that has a

9:40

consequence, and the consequence is that when you have

9:44

a gantry with a rotation time of 250 milliseconds,

9:48

what that gantry rotation is doing, what it's

9:51

doing is it's basically seeing 250 milliseconds

9:55

of what has happened during that cardiac cycle.

9:59

And that 250 milliseconds may not be much

10:04

of a, maybe the same if you're in diastole.

10:08

So what happened in the beginning

10:09

of the 250 is the same as the end.

10:12

But if the heart rate is fast,

10:16

and the diastolic time is short,

10:18

shrunk to 100 milliseconds.

10:20

Now that 250 milliseconds is not only

10:23

seen in diastole, but also seen in systole.

10:24

And that's where motion artifacts come in.

10:28

So, yep.

10:28

So everything nicely ties up when you think

10:30

about cardiac imaging, always think a little

10:33

bit about the physiology, the heart rate, the RR

10:37

interval, and how everything fits in together.

10:40

Thank you.