0:01

Okay, the next topic we're

0:02

going to review is T1 mapping.

0:04

T1 mapping is used for direct quantitative

0:06

measurement of the T1 time of tissues.

0:09

This is useful because your standard CMR sequences,

0:13

although you can put an ROI on a sequence and get

0:15

a signal intensity measurement, that measurement

0:18

is not standardized and does not actually

0:19

correlate to the real T1 time of the tissues.

0:22

So the nice thing about T1 mapping techniques

0:24

and also T2 mapping techniques that we'll

0:26

talk about later, are that you can actually

0:29

measure directly the T1 times

0:32

of the tissue in milliseconds.

0:34

Now, why is that useful?

0:35

T1 mapping may detect diffuse scar not

0:38

visualized on late gadolinium enhancement

0:40

images or abnormal sphingolipid deposition,

0:44

um, which is specific to Fabry's disease.

0:46

So the idea here is that your T1 can

0:49

be altered in the presence of fibrosis.

0:52

And what happens with fibrosis is

0:54

that you get replacement of myocytes

0:57

with fibrotic tissue and collagen.

0:59

That fibrotic tissue increases

1:02

the amount of extracellular fluid

1:04

in the heart muscle, and then that extracellular

1:07

fluid volume can actually be measured by T1 mapping

1:10

because the extra water content elongates the T1

1:13

time and results in basically an altered T1 map.

1:19

In the case of Fabry's disease, actually, you have

1:21

deposition because of the metabolic abnormality

1:23

associated with Fabry's disease you actually have

1:25

all these sphingolipids, which deposit

1:27

themselves within the cells themselves.

1:29

And those are primarily composed of fat.

1:32

And as we know, fat has a very short T1 time.

1:36

And so, um, with Fabry's disease

1:37

actually is really interesting.

1:38

You can see a shortened T1 time, uh,

1:41

due to all the sphingolipid deposition.

1:43

It's important to note here that T1

1:45

mapping is really used for diffuse processes.

1:49

So when you have, uh, what's called

1:51

interstitial or diffuse fibrosis,

1:53

you get collagen interspersed with myocytes.

1:56

Um, so these things don't really show

1:58

up on the late gadolinium enhancement images.

2:01

Late gadolinium enhancement is great for identifying

2:03

focal fibrosis, which means that basically all the

2:06

myocytes are dead and they've been replaced by scar.

2:09

That's where the gadolinium infiltrates,

2:11

and we see that on the post-contrast images.

2:13

With diffuse fibrosis though, because

2:16

it's a uniform process across the heart

2:18

and you have a mixture of

2:20

myocytes and interstitial fibrosis and collagen,

2:24

you're, you're not actually going to pick up

2:25

anything on the late gadolinium enhancement images.

2:28

So basically, the way this works, it's pretty simple,

2:31

is that the techniques are mapping the T1 recovery curve.

2:35

So you perform a 180-degree inversion of the spins,

2:39

and then you wait for that T1 recovery to happen and

2:42

sample the signal after the different inversion times.

2:45

So just basically wait at a whole bunch of

2:47

different points on the recovery curve and image.

2:49

And then if you map that out pixel-wise with

2:53

a graph, then you can extract the T1 time.

2:57

And, um, lastly, the T1 mapping can be performed

3:00

with two different approaches, either the TI

3:02

scout, which is our standard sequence that

3:05

we use to identify the correct inversion time for

3:07

the late gadolinium enhancement images or dedicated

3:11

sequences known as modified Look-Locker sequences,

3:15

MOLLI, or you may even see the term shMOLLI, which

3:18

is SH-MOLLI which means, um, basically shortened MOLLI,

3:21

uh, which is in pretty wide use today.

3:25

The only difference between these two

3:27

sequences—

3:28

they're both valid.

3:30

The problem with the TI scout is that it's not

3:32

normalized to cardiac motion, meaning that you're

3:34

going to get images all throughout the cardiac cycle.

3:37

Whereas the MOLLI is nice because each image

3:39

is in the same part of the cardiac cycle.

3:41

So you can, it's much easier to do a

3:44

segmentation and basically

3:47

measure the times across the multiple images.

3:52

So here's an example of using

3:54

Look-Locker for T1 mapping.

3:57

And in this case, you can see, actually,

3:59

it doesn't look like the myocardium position changes

4:02

much from phase to phase, which is surprising,

4:05

but generally, usually, you're going to see some.

4:08

This patient must have a very slow heart rate.

4:10

You'll usually see that the heart contracts, actually,

4:13

during the acquisition, which is what makes it a

4:15

little more difficult than the MOLLI technique.

4:17

Um, but what we get here is you,

4:19

you actually end up graphing the signal.

4:21

You do an ROI and you can either do a full

4:24

heart ROI, or maybe what's most often used is

4:26

just a, an oval-shaped ROI in the mid septum.

4:30

And then you graph those points across time.

4:33

So this is the intensity of that ROI, and this is

4:36

the time, the inversion time, and from that graph

4:39

can be extracted a best-fit line, and the best-fit

4:43

line there is basically defining the T1 recovery

4:46

curve, and from that curve, the program is able to

4:49

calculate a T1 time, and that's done on a pixel-by-

4:52

pixel basis when you actually create a T1 color map,

4:56

and then when you do an ROI, it actually takes an

4:58

average T1 time for the pixels included in that ROI.

5:02

The same process here is shown

5:04

with the MOLLI T1 mapping.

5:07

The difference here is that the

5:08

number of data points is much smaller.

5:10

You have just a few data points here

5:12

to calculate your T1 recovery curve, unlike

5:15

quite a few more data points on the Look-Locker.

5:18

But like I said, the advantage here is

5:20

that they're all from the same part of

5:21

diastole, so actually a little bit easier.

5:24

Okay, uh, just some notes about T1 mapping and

5:27

how they're expressed in the literature and how

5:29

one might describe them in clinical reports.

5:33

There are basically two different flavors

5:34

of values that you get from T1 mapping.

5:37

And there are different camps

5:39

out there in the research world.

5:41

And some people are really behind one technique

5:44

and other people are really behind others.

5:46

I don't necessarily have a strong opinion on this other

5:48

than the fact that I think the native T1, the first

5:51

technique is, is just easier.

5:53

So, um, my general bias is to try

5:56

and do, keep things fairly simple.

5:58

So I would tend to favor that

6:00

technique because of the ease of use.

6:01

But basically, um, the two different

6:04

techniques are native T1 and ECV.

6:05

Native T1, you'd perform a non-contrast

6:08

T1 map, and you measure the T1 time.

6:12

Generally, for looking for diffuse

6:13

disease, we do measurements of an

6:15

oval-shaped ROI within the mid septum.

6:18

ECV is extracellular volume, and this

6:22

is actually a measurement that uses

8:24

post-contrast and pre-contrast images.

6:27

It's expressed as a percentage, and that percentage,

6:30

what that percentage means is the percentage of

6:32

tissue that is composed of extracellular space.

6:35

So, in a certain given quantity of myocardial

6:38

tissue, the ECV is the percentage of that

6:41

myocardial tissue that is extracellular.

6:44

If you subtracted that from one,

6:46

the remainder would be the percentage of that

6:47

tissue that is cellular, meaning myocytes.

6:50

And so basically, if your tissue is composed

6:52

of a combination of myocytes and extracellular

6:55

space, the more fibrosis you get,

6:58

the relatively greater percentage of that

7:00

tissue is composed by extracellular space.

7:03

So increasing ECV percentage equals worsening fibrosis.

7:08

And the thing that makes this tricky is

7:10

that it's a map created from pre- and post-

7:13

contrast images and uses the—

7:16

basically, the difference of their T1 times

7:18

normalized to hematocrit to calculate this.

7:21

And so what is, what makes this complicated?

7:23

Well, one is now you're doubling any errors by using

7:26

two sets of images, a pre- and a post-set of T1.

7:30

Um, you're going to match those images.

7:32

So the patient has to be consistent in terms of

7:35

where their heartbeat and where you're acquiring

7:38

the images in diastole, and the patient, you know,

7:40

can't have moved in the bore of your scanner.

7:42

Um, and so that can present problems.

7:44

And then the hematocrit is another issue.

7:47

Hematocrit isn't necessarily

7:48

widely available for every patient.

7:50

So do you do an extra test of hematocrit the day of?

7:52

Is that necessary?

7:54

Generally, you like to have a hematocrit

7:55

that's fairly recent, within 24 hours.

7:58

It can fluctuate over time.

7:59

Um, so that, that creates another potential hurdle.

8:03

Okay.

8:04

So, uh, on this next slide, I'm going to

8:06

present exactly how ECV is calculated.

8:10

Basically, it's a subtraction technique

8:12

that's normalized to the hematocrit.

8:14

This is all happening on the backend.

8:16

So, you know, you don't have to do this

8:17

manually, but this is what's happening.

8:20

If you use any analysis software, generally, they're

8:22

going to ask you to contour the pre-contrast and

8:24

the post-contrast imaging, and then

8:27

provide a hematocrit, and this is what's happening

8:30

in the background on a pixel-by-pixel basis.

8:32

The T1 times of the myocardium

8:33

and the blood are being compared.

8:35

The T1 times of post-contrast are subtracted

8:38

from pre-contrast, and same with the blood,

8:41

and they're normalized to hematocrit.

8:42

And from that, you get an ECV number.

8:45

Here's just an example from the

8:46

literature of a case where they measured

8:49

the T1 time before and after contrast.

8:51

So you see that post-contrast, the T1 is

8:53

quite a bit shorter, 429 milliseconds versus 982.

8:57

And that's because of the GAD effects.

8:59

And so, you know, if you took the difference there,

9:00

you'd get, you know, 400 something in difference, and then

9:03

you'd normalize that to hematocrit, and divide it by the

9:05

blood pool, and so on, and you get your ECV number.

9:09

So this is actually my, uh,

9:10

first chance to show you guys this paper,

9:13

which came out in 2020, it’s super useful.

9:17

It's in the Journal of

9:19

Cardiovascular Magnetic Resonance.

9:21

What's fantastic about that is this is

9:23

an open-access journal, so anyone could

9:26

get it just by searching for it online.

9:28

It's called Reference Ranges or Normal

9:30

Values for Cardiovascular Magnetic

9:32

Resonance in Adults and Children.

9:33

There was an older version from several

9:35

years ago, and now it's been updated,

9:38

and it's almost like a book in a way.

9:40

I mean, it is an article, but it's an incredibly long

9:43

article because they've included so many different

9:46

normal ranges for any type of thing you're looking for.

9:49

So normal LV sizes, normal LA sizes, normal

9:53

aortic sizes, and so on and so forth.

9:56

So, inside of this really great reference.

9:59

You can get native T1 as well as ECV

10:01

ranges from a bunch of different publications.

10:05

And so, I'll just focus on the normal ranges

10:08

here as well as the standard deviation

10:10

and the upper and lower limits of normal.

10:12

And you see that for native T1 times, depending on

10:15

the technique, around the high 900s for 1.5 T,

10:20

and around the low thousands for 3 T.

10:23

Um, and so it's important to know that if

10:25

you are doing 3 T, the T1 time is longer.

10:28

And then ECV percentage somewhere

10:30

around that 25, 26 percent range.

10:32

So upper limits of normal hover around the 30% range or so.

10:38

So generally, the number we keep in mind if

10:40

we're doing ECV is around 30, whereas for the

10:42

native T1 times, it's somewhere around, you know,

10:45

1000 to 1100 or so in one and a half Tesla, and

10:50

over 1200 or 1300 or so for, for three Tesla.

10:54

Both native T1 and ECV have

10:57

been studied, as we talked about.

10:58

And, um, one note here, if you look at articles

11:01

out there, although we have these normal ranges,

11:03

it's important to note that the T1 is a bit

11:06

of a tricky — the N NT2, we'll talk about later,

11:09

it's a bit of a tricky sequence in that you

11:11

can compare to normals that are published,

11:14

but there's always a little bit of local variation.

11:16

And so all the guidelines out there would recommend

11:19

that you establish your local reference ranges.

11:21

How do you do that?

11:22

Well, they've suggested that one should

11:25

basically start paying attention to your

11:27

cases, get T1 in a whole bunch of cases.

11:30

If you can ideally find maybe about 20 cases that

11:34

are normal, you know, that let's say it's a rule

11:37

out sarcoidosis case and you don't find any cardiac

11:39

sarcoidosis, you can use that as a normal and then

11:42

use that to establish a local reference range.

11:46

Okay.

11:46

This is just a nice graph that if you ever go to

11:49

a talk about T1 mapping, you'll see this over and

11:52

over, and over again; it comes from this paper,

11:55

which is a guideline statement from the same

11:57

journal JCMR, which again is open access, so super

12:00

helpful, um, which is a consensus statement that

12:02

was put out in 2017 about how to use T1 and T2

12:06

mapping. This is the statement that I mentioned

12:08

that says you should really create your own

12:10

local references for what the normal values are.

12:13

And I just point this out because I think,

12:16

you know, oftentimes when we think about

12:17

T1, especially, there's some questions to,

12:20

you know, what should you really use it for?

12:22

What do the numbers mean?

12:24

My personal opinion is that the most important

12:27

diagnoses that are useful for T1 are Fabry's disease.

12:30

You can use it because if you have Fabry's

12:33

disease, you will see a really, really low T1

12:36

time that's quite a bit separated from normal.

12:39

And then the other one is amyloidosis.

12:41

So amyloidosis, the T1 times here, very,

12:46

very different from normal, very elongated

12:49

in the case of native T1 values

12:51

and very elevated in the case of ECV.

12:54

So in my opinion, um, these two diagnoses

12:57

have the best separation from normal patients.

13:01

Um, whereas although these look

13:03

on this graph that they're well separated

13:07

from normal patients in sort of real

13:08

world clinical experience, these diagnoses—

13:11

so HCM, dilated cardiomyopathy, etc.—

13:14

there's really so much overlap with normal

13:16

patients, it's really difficult to make much

13:18

sense of what that means. So the use of the

13:21

T1 number to define, you know, the extent of

13:24

diffuse fibrosis, and so on, is a little more difficult

13:27

in these sort of more chronic diseases that aren't

13:30

so significantly different from normals.