0:02

Hello, and welcome to Noon Conference, hosted by Medality.

0:05

Noon Conference connects the global radiology community through free live

0:09

educational webinars that are accessible for all and is an opportunity to

0:13

learn alongside top radiologists from around the world.

0:16

Today, we are honored to welcome Dr.

0:18

Prashant Nagpal for a lecture entitled Cardiac CT: A Journey from

0:22

Origin to 2026. Dr. Nagpal completed his

0:25

radiology residency at the University of Iowa and his

0:29

cardiovascular imaging fellowship at Brigham and Women's Hospital, Harvard Medical

0:33

School. He currently leads the section of cardiovascular imaging at the

0:36

University of Wisconsin, Madison, specializing in advanced cardiac

0:41

and vascular imaging.

0:43

At the end of the lecture, please join him in a Q&A session where he will address

0:46

questions you may have on today's topic.

0:49

Please remember to use that Q&A feature to submit your questions so we can get to

0:52

as many as we can before our time is up.

0:55

With that, we are ready to begin today's lecture. Dr.

0:57

Nagpal, please take it from here.

1:00

Okay. So,

1:03

in today's lecture,

1:05

what I want to discuss is just how cardiac CT has evolved

1:09

from its start to how we use it now.

1:13

Again, before I start, here are my disclosures.

1:17

And

1:19

in terms of objectives, I want to

1:22

summarize the data that supports the role of cardiac CT for chest pain

1:26

evaluation. I'm going to focus a little bit more on chest pain because that's

1:30

where really the crux of the use of cardiac CT,

1:34

at least traditionally, has been, and why we see

1:38

such amazing growth in this field.

1:41

Along with that, I would like to summarize the emerging role of newer AI

1:45

applications and quantitative applications for coronary CT.

1:49

Some of the names that most of you may have heard is

1:52

CTFFR or CT plaque quantification.

1:57

And I would highlight a few examples of how we can use cardiac

2:00

CT beyond just coronary evaluation.

2:04

So again, my goal is to keep this talk a

2:08

little more basic and kind of go through the journey of how

2:12

cardiac CT has developed. So cardiac

2:16

CT primarily started as calcium scores exam.

2:20

So for those that do not know calcium scoring, it is

2:25

a non-contrast exam. No contrast is given, but we do

2:29

perform it with ECG gating. I'll touch base on what is ECG

2:33

gating a little bit as well in this lecture.

2:36

But it was basically an offshoot from the DSA data that if

2:40

somebody had calcified plaque, they found

2:44

more obstructive disease, and there was worse outcomes.

2:47

So the way we do it is

2:49

we basically do a non-contrast CT.

2:52

We kind of define what is calcium. So in this case,

2:56

in calcium score CT, anything that is more than

2:59

130 Hounsfield units is defined as calcium.

3:02

And then we tell the software that, okay, this is calcium in,

3:07

let's say in this case, this is left main.

3:09

So it measures the Agatston score of this.

3:11

But we obviously exclude other areas of calcification.

3:16

So anything pink here is 130 Hounsfield units.

3:19

But we manually, or these days it's almost automatic,

3:23

we basically exclude these areas and just measure the

3:27

amount of calcification in the coronary arteries.

3:31

So this was a basic, simple test. But if you look at one of the original

3:35

publication that came from Dr. Agatston on this

3:39

topic, and what it shows is that

3:43

even at that time, this was a very long exam.

3:47

So it required around 40 heartbeats to be done.

3:50

So the breath-holding time used to be 30 to 45

3:54

seconds. And

3:57

total procedure time from patient getting on the table,

4:01

kind of doing the full exam, it used to be around 10 minutes of

4:04

just a non-contrast exam. And

4:08

what has happened is that at the same time, it was

4:11

1990, when calcium score was kind

4:15

of coming out,

4:17

there were some people that started working on giving contrast and

4:21

then seeing if we can evaluate coronary using the

4:25

same methodology, but after giving contrast.

4:28

So in 2000, this was one of the first papers that came out that

4:33

you can do basically multiple slice

4:36

evaluation of coronary arteries.

4:39

It's funny, when I

4:43

now go back and look at these papers and I look at the image quality,

4:47

and when somebody gives us these papers, we know that

4:52

these images are probably the best images that could

4:56

come through. But this is how those images looked, and at that time,

5:00

these were considered very good image quality that we can achieve

5:04

diagnosis. But even with these, the problem initially

5:08

was that the heart rate had to be less than 60.

5:12

The radiation dose was, at that time,

5:15

approximately 15 years of background radiation.

5:19

And we'll come to how much is the background radiation

5:23

in US. But

5:25

this kind of continued, but what they showed is that even though the image

5:29

quality from today's standard was suboptimal, there was

5:33

reasonable correlation between CT, even in the

5:37

early 2000 era, and

5:41

DSA. So at that time, before the

5:45

64-slice scanners came on, the prerequisites

5:48

were that you have to have a slow heart rate.

5:52

That means less than 55 beats per minute.

5:54

Along with that, it has to be steady.

5:56

At one point, atrial fibrillation used to be an absolute

6:00

contraindication to do a coronary CT.So,

6:04

because your CT is acquired across multiple heartbeats, and if the

6:08

heartbeats are changing in interval, your data would not align.

6:13

Also, we were giving sublingual nitrate, which we still do, depending on the

6:17

patient's blood pressure. We give either 0.4 to 0.8 milligram.

6:21

And then this required breath holding for around 20 to 40 seconds,

6:25

which is very hard to do, especially for people who have disease.

6:30

At that time, the radiation dose was more than 20 millisievert.

6:34

For those that do not know, our annual background radiation

6:37

dose in the United States is around

6:41

2.5 millisievert-- Is around 3 millisievert, and

6:45

interestingly, outside of US, we say it is around 2.5

6:48

millisieverts. So for those who

6:52

are from outside the US, the annual background

6:55

radiation dose outside US is usually around 2.5 millisievert,

6:59

and in US it is 3 millisievert. So one

7:03

CT accounted for around six to seven

7:07

years' worth of background radiation that we get

7:11

exposed to. But there have been significant advances in our

7:15

field. And one of the advance that really changed

7:19

things, and this basically,

7:22

for layman, we talk about, oh, the CT went from

7:26

16 slice to 64 slice to 128 slice to now

7:30

320 slice scanner. This basically, the number of

7:34

slices that are increasing, that just means that the detector is

7:38

getting bigger. When the detector or the rows of

7:41

the detector keep getting bigger, you can

7:45

get imaging of a larger anatomical area in

7:49

one heartbeat

7:51

to a point that we have reached a point that we can

7:55

cover the entire heart in single heartbeat.

7:58

These are called as volume scanner.

8:00

If people who basically read literature, sometimes the

8:04

literature comes out that single heartbeat acquisition using a volume

8:08

scanner.

8:09

Not only that, our scanners are becoming much

8:13

faster.

8:15

The technical term for that is called as

8:18

improvement of temporal resolution.

8:21

Temporal resolution is basically how fast you can image

8:25

something. So faster means you can image something faster.

8:29

That translates to that we can do imaging in

8:33

people in which the heart rate is slightly

8:37

higher, and even then the images are going to be clear.

8:41

We can do much smaller breath hold times.

8:44

So for example, almost all coronary CTs have become less than

8:47

five seconds of breath hold time. We are very lucky in my

8:51

institution that we can do a single heartbeat

8:55

acquisition. That means all we need is one

8:58

RR interval or one heartbeat. That means if somebody is

9:02

heart is beating at, let's say, 60 beats per minute, all we

9:06

need is one second of acquisition.

9:09

And

9:10

our image reconstruction, just like we see in technology all around

9:14

us, that,

9:16

for example,

9:18

our phones have

9:20

images that have significantly improved image quality over years.

9:24

Same thing is happening that we are getting newer types of image reconstruction

9:28

methods, which have led to improvement in image quality.

9:33

And now you don't have to give more radiation to improve the

9:37

image quality. So you can basically use newer algorithms to improve

9:41

it. So that has allowed us to decrease our radiation dose.

9:45

Our spatial resolution. So I know I talked a little bit about temporal

9:49

resolution, which is basically how fast you can image.

9:53

Spatial resolution is how fine details you can image.

9:56

So our spatial resolution is

10:00

improving. That means we can

10:03

visualize smaller structures with much more

10:06

accuracy, and that basically translates to our

10:09

coronary imaging directly because our coronary imaging

10:13

is done with--

10:16

Like our coronary arteries are few, sub 3 millimeter or

10:20

3 and a half millimeter. So then you can get very high-quality

10:23

imaging.

10:25

There have been some other advances, which I don't think it really

10:29

pertains much. How do you use nitroglycerin, whether

10:33

you do the sublingual nitroglycerin versus the patch.

10:36

There are certain advantages that come with patch versus sublingual.

10:41

There have been some publications on this that show that nitroglycerin

10:45

patches work the same way in terms of vasodilation,

10:49

except that it just takes longer time.

10:52

So we currently use some places, we do use patches,

10:56

and especially for areas in which the workflows are very

10:59

complex,

11:01

and we don't have nurses to give nitroglycerin.

11:05

The radiation dose per se is significantly

11:09

changed, for coronary CT.

11:13

For some

11:14

indications, we can do these even at a dose equivalent to a chest

11:18

X-ray. I would not say that in our program we have decreased our radiation

11:22

dose to that extreme. We still give very low radiation

11:26

dose, but not to an extent of, okay, it has to be equivalent to a chest

11:30

X-ray. But

11:33

there have been

11:34

so many

11:36

papers on this that you can now do

11:40

imaging with very low radiation dose.

11:43

And one of the main thing that has made it possible is that

11:46

earlier when CT started,

11:49

if you look at this in the bottom image, which has

11:53

ECG,

11:55

earlier, we used to acquire a large part of that cardiac

11:59

cycle. Nowadays, what we do is then this blue box that you see

12:03

corresponds to when the

12:05

X-rays are being acquired.So

12:09

we only acquire images, or the CT

12:13

tube is on only for a certain part of cardiac cycle.

12:16

So we don't get dynamic information, but we get static,

12:20

excellent quality information even at less than one

12:24

millisievert radiation dose. And as I mentioned, three millisievert

12:27

is kind of our

12:31

yearly background radiation dose.

12:32

So that's why when we talk about one millisievert, it

12:36

kind of equates to around one third of a year or four

12:40

months of radiation dose.

12:43

So this is exactly what we were just talking about, that we can

12:47

do imaging at less than

12:49

four months of radiation dose that we get

12:53

anyways in our lifetime.

12:56

There have been some studies that have kind of measured the effect of

13:00

changing technology, and one of the big ones that I

13:04

really

13:06

enjoyed reading was, and again, I think there is a newer version that's

13:10

going to come out of this study. So in UK, all

13:14

the NHS

13:16

scanners

13:18

have a lot of data on them, and they did a survey on

13:22

how dose changed on the same scanner from

13:25

2007 to 2017. And what they saw was

13:28

that there is significant, like significant from,

13:33

if you look at the number of DLP, which is also a measure of

13:36

dose, it decreased from

13:40

85 in 2007 to 195. That's

13:45

not 50% decrease. That's like 70% to 80%.

13:49

So it's around 78% decrease in radiation

13:53

over a decade.

13:55

So

13:56

personally, I was lucky to have experience on

14:00

both a dual-source scanner as well as wide detector.

14:04

So dual-source scanner is basically when we have two X-ray tubes, and

14:08

I will come to that, how is that useful, and wide detector is

14:12

what I was telling you that we keep on increasing the detectors to a point that we

14:15

can do a cardiac CT in one heartbeat.

14:20

The advantage that comes with all these is that we can

14:23

do prospective ECG gating in nearly all cases.

14:27

So except when we need dynamic information for any

14:31

specific reasons. For example, its most common indication is valve evaluation.

14:35

When we need valve information, that is

14:38

when we still do retrospective or a full cardiac

14:42

cycle. Otherwise, we just do a small part of cardiac cycle.

14:46

The good thing is we are able to achieve all this with very standard

14:50

protocols. So we only have four protocols that

14:53

suffice for most everything that we do.

14:57

Then we obviously do a sublingual nitroglycerin.

15:01

Heart rate control varies on what kind of scanner you have.

15:04

If you have a dual-source scanner, that means two X-ray tubes, you

15:08

can do much higher heart rates, whereas at a single-source scanner,

15:13

you cannot do much higher heart rate because the temporal

15:16

resolution or your ability to scan moving structures is

15:20

slightly lower.

15:22

Now, radiation dose for us has consistently been less

15:26

than three millisievert for a coronary exam.

15:29

That means we give less than one year of background radiation

15:33

dose. I'm going to show you one example of a

15:37

patient that came to us with a

15:41

BMI of 42. So that's like,

15:44

again, CT scan needs X-rays. X-rays have to go

15:48

through the body and be recorded

15:52

on the detector. People who have higher BMI, we struggle a lot.

15:56

Now, the second thing we struggle with is high heart rate.

15:59

Now, this example is a patient with high BMI

16:04

and high heart rate. And we did the CT, and this is

16:08

how these images look like.

16:10

So when you look at these images, these are really high quality

16:14

exam, almost everything fully diagnostic, and

16:20

despite having two things that were not working in our favor, heart

16:24

rate and BMI, the radiation dose was 1.5 millisievert,

16:28

and the images were super diagnostic.

16:31

All segments fully

16:34

evaluable. Now, obviously,

16:37

when we can do it in patients who have high BMI, high heart

16:41

rate, we can obviously do it much better on

16:45

a patient that have normal BMI and

16:49

relatively low. This was still 96 heart rate.

16:52

Ninety-six for a coronary CT is in no way considered low.

16:56

It's also a high heart rate. But if you look at the quality of

16:59

images that you can get out,

17:03

even at high heart rates, is just phenomenal these days.

17:07

So

17:09

the reason I'm showing you is not because I want to show off

17:13

image quality improvements. What I'm trying to show is that at high heart

17:17

rates, we are able to do these. What that entails too is

17:21

that you are able to do coronary imaging with less

17:24

and less roadblocks. Patients coming in with high heart rates, we are not

17:28

wasting time giving too many medications. We are scanning them fast.

17:32

And when we can scan them fast, it opens the door

17:35

for imaging people in the emergency room or giving

17:39

people access to imaging much faster and sooner.

17:44

This is just an example of a coronary CT we did.

17:48

Radiation dose, again,

17:51

one third of the background yearly radiation dose.

17:54

Contrast, only 50 mL, and we could do full

17:57

evaluation of the entire coronary tree.

18:00

And

18:02

this does not just apply to

18:05

normal vessels. This is a patient with diseased vessels, and still we get

18:09

very high resolution.One

18:13

basic thing that has led to growth of coronary CT

18:16

is excellent resolution. And when

18:21

we talk about this,

18:23

I know I've mentioned this a couple of times, we always talk about resolution in

18:26

two ways. One is spatial resolution, how small structures

18:31

you can see. And I think it is very good for people to understand the

18:34

difference between spatial resolution CT

18:38

versus spatial resolution on angiography.

18:42

And

18:43

on the CT,

18:45

the

18:46

spatial resolution is 0.2 to 0.4 millimeter,

18:50

whereas invasive angiography has superior spatial resolution

18:54

slightly, but it is around 0.2 millimeter.

18:58

Where CT is slightly slower, it's mainly temporal resolution.

19:02

That means how fast the images are acquired.

19:06

It is basically around 66 millisecond for a

19:10

two X-ray tube machine, and around

19:14

most people who do cardiac imaging, we don't do it

19:18

if the temporal resolution is lower than 165

19:21

millisecond. One, and invasive angiography tends to be much,

19:25

much faster. That's why images are more crisper.

19:28

You see them very well. And one basic difference is also you

19:32

are giving contrast in the peripheral vein on an CT,

19:36

and whereas invasive angiography, you are putting a catheter and truly injecting in

19:40

the intracoronary lumen. So

19:44

basically, sometimes the contrast enhancement is superior.

19:48

So these were the basic

19:50

overview of

19:53

how our CT has grown. Now, coming back to

19:57

how do we visualize these images, and what are the common caveats on

20:01

how do we get to these images? So

20:05

to start with, how do we visualize these images?

20:08

So the one thing that we do is that these images are acquired

20:12

in basically a plane that is, we call it axial plane.

20:16

These are basically slices through our body, and this is

20:20

how they look. But interestingly, these are so high

20:23

resolution that we can reconstruct these images in any

20:27

different plane. So the data is only our axial data set.

20:32

This image, which is showing you entire right coronary artery in one

20:36

view, is just a reconstruction from those images.

20:39

This is just image post-processing.

20:42

Then,

20:45

in terms of image visualization, the image on the left,

20:49

this one, is called as a volume rendered.

20:51

That means it

20:53

basically is giving a volume to entire thing that is

20:56

enhancing or is bright, and then you start seeing these

21:00

structures. These are predominantly for overview of

21:04

these. We don't use it for diagnosis.

21:07

Somehow what we have seen is that the surgeons really like this to

21:11

understand the anatomy really well.

21:13

So this is basically a volume rendered image for

21:17

this and for the LAD. And this image in which you

21:21

kind of see the entire lumen pretty well, this is called as

21:25

curved planar reformats or CPRs, not the CPR,

21:29

that cardiopulmonary resuscitation.

21:32

This is basically the radiologist's CPR.

21:34

So this is curved planar reformats.

21:37

And this is, again, you can visualize these volume rendered images multiple ways.

21:41

So if you are looking at images and you're confused which image I'm looking at,

21:45

this is still a volume rendered image. Okay.

21:49

So we were talking about how do we do these

21:52

exams. So the most common way we do these exam is

21:56

we do prospective

21:59

triggering. So prospective triggering is if you

22:03

look at this middle row and you see this ECG

22:08

here, and in this ECG trace, we

22:11

only acquire at a certain point of cardiac cycle.

22:15

It does not give you dynamic information because you are only acquiring at one

22:18

point. We try to acquire it when it is most

22:23

stable or less motion-y. So in this mode,

22:27

it's also called a step and shoot mode.

22:29

So basically, we shoot X-rays, then we move the patient table,

22:33

then we shoot X-rays, we move the patient table.

22:36

If we do it only at one part of the cardiac cycle, so for

22:40

example, if you are only acquiring images at, let's say, 60%

22:44

or 70%,

22:46

then this is called as prospective gating with no padding.

22:50

Padding means if you are acquiring not the full

22:54

heartbeat, but you are acquiring more than just

22:58

one

23:00

definite point of the cardiac cycle.

23:01

So you can do padding like 70% to 80%, or you can

23:05

do even bigger padding. Obviously, your radiation dose keeps on increasing

23:09

as you pad more, but you can do 40% to 80% padding.

23:14

Now, retrospective gating is basically when you turn the tube

23:18

on the entire time. So it allows you to get dynamic information because the heart

23:22

is moving all the time. So we can do

23:25

retrospective gating with tube current full on

23:29

all the time, or you can just change it that, let's say, you

23:33

want diastole to be full dose, and then as the heart is

23:37

beating during the systole, because that's not your still phase,

23:41

you kind of decrease the radiation dose.

23:42

This is called as retrospective gating with MA modulation.

23:47

So we can do things both ways.

23:49

Now,

23:51

again, for those who are into reading about

23:55

coronary literature, they may see a word called as a flash

23:59

mode or high-pitch helical mode. In this,

24:03

what is happening is that the tube is rotating very

24:07

fast, and this is only available on the

24:11

scanners that have two X-ray tubes.

24:14

But the problem with this is that you'd need really stable heart

24:18

rates to use this for coronary imaging.

24:20

We rarely use it for coronary imaging.

24:22

We only use it for our aortic scans if we were to use

24:26

it.So,

24:29

it is not that while we are trying to image

24:32

coronaries in the best phase when there is least

24:36

motion,

24:38

we still try to add a little bit of padding.

24:40

Padding, as I said, is that I would not just acquire

24:44

a 70% phase. I would try to acquire, let's say,

24:48

60% to 80% phase. The reason we do that is we can

24:52

still have some cardiac motion. So, for example, in this case,

24:56

what happened is we were looking at this coronary artery, and at

25:00

proximal LAD, you see it seems like there is some stenosis.

25:04

And this was kind of our best diastolic phase that was

25:08

75%, and it looks

25:10

narrow at this point. So what we did was we kind of

25:14

looked at it in other parts of cardiac cycle, and we just looked at it in

25:18

80% and this area opened. That tells you it was just

25:22

motion that we are visualizing. That's why we try to add a little

25:26

bit of padding to our

25:28

scans that we do.

25:30

Now, irregular rhythms or high heart rates, which used to be

25:34

contraindication earlier, are no longer contraindication.

25:38

We try to handle them very well. So obviously, the rules

25:42

still are that the lower heart rate is better.

25:45

If you have very regular heart rate, you can just keep

25:49

acquiring the data and then tease out the good part of data so you

25:53

can use retrospective gating. Though it leads to a lot of radiation, you can still

25:56

use it.

25:58

And

25:59

one thing that we learned, so in our practice, we don't do retrospective

26:03

gating at all. If the heart rate is higher, we actually do

26:06

end-systole, because end-systole tends to be much more consistent.

26:11

So if somebody asks you, if somebody's heart rate is

26:15

60 and then you make them exercise and it goes to

26:19

90, what part of cardiac cycle changes? It's the end-diastole.

26:23

So as the heart rate keeps increasing

26:27

from, let's say, 50 to 100, it's the

26:31

end-diastole that keeps on getting smaller and smaller.

26:34

End-systole remains consistent.

26:36

So that's why in people with high heart rates, we try to

26:40

do end-systolic acquisition. We also aim to

26:44

use highest temporal resolution.

26:46

That means we do not slow down our gantry.

26:49

Usually. Usually, we don't. Slowing down gantry

26:53

slows down our acquisition or slows down our resolution, but it

26:57

also allows us to put more X-rays if needed.

27:00

So we sometimes, very rarely, slow down gantry only if we

27:04

are doing visualization of coronaries in somebody with a

27:08

BMI of more than 50. We want more X-rays to go in.

27:13

Then obviously, if you are dealing with somebody with irregular heart rate,

27:17

make sure that you avoid other artifacts like

27:21

contrast delivery issues, breathing issues, and other things.

27:25

So

27:26

again,

27:27

sometimes it doesn't happen in our practice.

27:31

Many practices do that if this is a difficult patient with high heart

27:35

rate or irregular heart rate, they do a calcium score, see if there

27:39

is high calcium, then they don't even do coronary CT.

27:43

In our practice, that doesn't happen.

27:45

We have never canceled a coronary CT

27:49

because of high calcium. Because what we have found is that

27:53

almost any amount of calcium, you can get

27:56

diagnostic information

27:59

in most of the cases. So,

28:02

in

28:03

high heart rate acquisition, as we talked about, you want

28:07

end-systolic triggering, or you can do wide padding so that you

28:11

can really get high-quality imaging.

28:14

So,

28:16

even in high

28:18

heart rate, the reason I was showing you that last example

28:22

is because this was one case in which we did-- that typically coronary

28:26

acquisition happens at late diastole, most commonly because

28:29

we try to do it at slower heart rates.

28:33

But this patient, we scanned at a higher heart rate, and if you look at it,

28:37

the right coronary artery is absolutely non-diagnostic in late diastole.

28:42

So we basically changed the images to

28:45

end-systole, and suddenly the images are fully diagnostic.

28:50

So if you look at this patient, this is a 58-year-old male.

28:54

We scanned this one at 62 beats per minute.

28:56

In no way high heart rate. And it was

29:00

regular.

29:02

And suddenly we saw this, and we were like,

29:06

"Huh,

29:07

what happened here?" So our thought is whether the

29:11

patient is having coronary motion, but hopefully,

29:15

the motion does not happen to this extreme that there is everything

29:19

double, aorta is double, left atrial appendage is double,

29:23

or there is some kind of misregistration because patient had

29:27

some form of arrhythmia, which even though it's a regular rhythm,

29:31

but patient through something atopic or something.

29:36

Or is this patient breathing? This doesn't look like breathing artifact.

29:39

Breathing artifact should also move sternum and other

29:42

structures. This is localized to cardiac structures.

29:46

So the sternum is pretty crisp. If a patient was moving, per se,

29:50

it would be one of those. So what we saw when we saw

29:54

this kind of image, we go back to scanner,

29:57

and we look what has happened. And what we saw was this

30:01

patient was having some kind of a strange rhythm in which

30:04

what happened was the patient's

30:08

normal waves were kind of identified as QRS

30:11

complexes, some of these random ones.

30:14

And it was basically a scanner

30:18

reading this ECG issue. So what we did

30:21

was we went there, and we looked at it, and any

30:25

place where it randomly found

30:29

these areas, we kind of labeled these that, okay,

30:33

we want you to reconstruct from this, butDo not

30:37

reconstruct from this ectopic kind of a beat.

30:40

So the patient was throwing these PVCs, which

30:44

basically all of these green were listed as

30:48

PVCs and deleted, and suddenly our data was

30:52

like this, and now this same scan became fully

30:55

diagnostic.

30:57

So these days,

30:59

what we do is that we also use other techniques like arrhythmia

31:03

rejection, in which what we do is that we

31:07

do

31:08

basically a practice

31:10

heart rate, and we measure the RR interval.

31:14

If the RR interval changes by any more than 10%, it

31:18

gets rejected. So that's basically on in

31:21

almost all cases that we do. One thing

31:26

I would like to show in specifically for atrial fibrillation,

31:30

we

31:31

do atrial fibrillation cases. Obviously, we do them with prospective

31:34

gating. But one thing we should do is that you

31:38

should never reconstruct these images in a

31:42

percentage basis. Percentage basis is basically scanner looks at the

31:45

RR interval, and first R wave is counted as

31:49

0%, second is counted as

31:52

100%, and it starts reconstructing a certain

31:55

percentage.

31:57

Now, problem with percentage basis is that

32:01

when somebody's in Afib and the RR interval is changing,

32:05

their absolute millisecond beyond which this

32:08

percentage sits also change. So what we do in these patients

32:12

is we do millisecond-based reconstruction.

32:15

That means we look at the QRS complex R wave,

32:19

and then we tell the scanner, just do 500 milliseconds

32:23

after this R wave. That way it is consistently in the

32:26

same part of the cardiac cycle, and

32:30

it prevents misregistration. So

32:33

this is kind of some very neat tricks we do to

32:37

improve our image quality.

32:41

I briefly touched upon this, that coronary CT in obese people is a

32:44

very big problem, and especially in morbidly obese,

32:48

because our image quality is poor when those

32:52

X-rays have to go through a lot of tissue.

32:55

This is sometimes how image quality looks.

32:57

This is kind of one of the good image quality

33:01

in a patient, let's say, for very high BMI.

33:04

Now,

33:06

what we do is that we increase penetration.

33:09

That means we increase the voltage of our X-rays.

33:12

We increase our mA.

33:14

We make sure that there are no heart rate issues, there are no other

33:18

issues. We also sometimes slow down the gantry

33:22

rotation speed, just to put more tube,

33:26

like X-rays, in the patient to increase the

33:29

quality. We also rarely, but we also

33:33

increase the slice thickness slightly because thinner

33:37

slices have increased noise, thicker slices have lower noise.

33:41

And this is just an example. I know in no way this is an ideal

33:45

exam, but this patient came with a BMI of

33:48

around 51. And

33:52

the problem was, I thought that this is going to be a fully

33:55

non-diagnostic scan. We did it with

33:58

very good heart rate control, and we slowed down the tube a

34:02

little bit to push more X-rays, and what we got was a very

34:06

high-quality exam.

34:09

So this is one of the work that actually,

34:13

I know when I made these slides, was under review.

34:15

Now it is accepted, but we basically looked at

34:19

how we can use newer

34:22

reconstruction algorithms. There are new DL, stands for

34:26

deep learning. So there are new deep learning algorithms that can improve the

34:30

image quality, even in high BMI patients, to an

34:33

extent that we can do very high-quality imaging

34:37

in these patients.

34:39

So same is true for patients with stents and

34:43

high calcium, that we can do very high

34:47

quality, but we have to be careful of certain things.

34:51

Stents, obviously more than three millimeter are usually okay.

34:54

Less than three millimeter, we are equivocal.

34:57

Less than two, unless it's a photon counting CT, we

35:00

don't

35:02

allow that in our practice, at least if the question is stent

35:05

evaluation. But for people who may give SCCT

35:09

boards one day, it's basically considered three millimeter is the

35:13

magic number in boards if they ask you.

35:17

So,

35:20

then

35:22

high kVp

35:24

in patients with stents and calcium, we do high kVp, which basically increases

35:28

penetration. We do multiple phases. We don't do just one phase.

35:32

We kind of add padding. Now, for

35:36

high BMI patients, we use softer kernels to

35:40

decrease the noise. But for

35:43

stents and calcium, we do slightly sharper kernel, like bone

35:47

kernels, to see these

35:49

things better.

35:51

And there are few new options, like for photon counting,

35:55

we have some new techniques, or we can do ultra-high definition

35:58

modes. This is just our experience

36:02

with a photon counting CT that we have

36:06

at our site that we did a study in which we looked at

36:10

stenosis characterization between regular CT.

36:14

These are called energy integrating. That's like the regular CT that we have.

36:19

Not just we, that's the industry standard.

36:21

And then we compared it with

36:24

photon counting CT, and what we saw is our

36:28

resolution of calcium is significantly more superior

36:32

on the photon counting CT. And our stent resolution

36:36

is also very high. This is not a three-millimeter stent.

36:39

This was a two and a half millimeter stent.

36:41

I was a little worried whether we will be able to see it or not, but

36:46

amazing stent resolution.Okay.

36:49

Beyond that, we can get excellent

36:52

image quality

36:54

even with prospective cases. When you

36:58

add padding, you can look at some dynamic information.

37:01

This is basically a prospectively gated exam.

37:05

And we get

37:07

the images to the point of how you see in echo.

37:11

And by changing where we inject, for example, we

37:15

normally look at left-sided chambers, but if you trigger the

37:19

CT in a way that the contrast is mainly in the right heart, you can

37:23

get excellent right heart visualization, which basically

37:26

allows us to do

37:28

a lot of preoperative planning for pulmonary valve structures.

37:33

I want to share how this high

37:36

quality and these skills

37:39

change to

37:42

clinical utility. So this is basically a 25-year-old patient,

37:46

came in with chest pain. And a young patient, chest pain,

37:50

very unlikely to have coronary artery disease, but

37:53

again, we do some low-risk patients as well.

37:57

And we did this, and what we saw was that this patient's

38:01

coronary artery was coming from the wrong cusp.

38:04

So if you see this image, this is the right coronary artery.

38:07

Normally, it comes anteriorly, but it is coming from the left side, and this is

38:11

just a volume-rendered image showing the same thing.

38:14

Normally, it comes off this cusp, but it is coming from the left cusp.

38:18

So this is basically an anomalous coronary artery.

38:22

Now, on CT,

38:24

this does not just stop here. So we

38:27

can tell this is anomalous coronary artery. This is inter-arterial course.

38:32

We can also evaluate the ostium really well.

38:35

So obviously, this is how the ostium looks up closer to the

38:40

origin. And on CT, we could

38:43

look at the endoluminal view the way a surgeon would see this, and

38:47

we can even tell the difference of

38:50

how this is the normal left main, this is the anomalous coronary

38:54

artery, how far they are, and how they look.

38:58

So this is basically a beautiful round origin of the left

39:02

coronary artery lumen, and this is basically an oblique or

39:05

slit-like origin of this. So,

39:10

this is one of my favorite papers to review anomalous

39:13

coronary arteries. If you want to do it, there are five types.

39:17

I won't go in detail, but I just wanted to put it out there.

39:21

It's from 2017, an excellent paper for this.

39:26

And it even has some nuances of how do we differentiate

39:29

inter-arterial versus subpulmonary.

39:32

That is differentiated based on the level of the anomalous coronary

39:36

artery. Like if it is at the level of the pulmonary valve, it's called

39:39

inter-arterial. If it is below the level of the pulmonary valve, so this is the

39:42

pulmonary valve, and the anomalous coronary artery is below it, it is called a

39:46

subpulmonary. It is a very important difference

39:50

because inter-arterial is basically

39:54

what we call as

39:56

the malignant or basically high-risk course,

40:00

whereas subpulmonary is not a high-risk course.

40:03

Here is another example. This was a 19-year-old

40:07

basketball player. Patient was having chest pain and shortness of breath.

40:11

So again, young people.

40:13

And this echo usually is the first thing.

40:16

So this echo was done, and I don't know amongst

40:20

the readers, people who do echo, I don't do echo myself.

40:23

I'm a radiologist, and we do CTMR. But what was wrong in this

40:27

echo was there was a very large coronary artery coming

40:31

off, and there was some reversal of flow in this.

40:34

So our echo readers read this as that the coronary

40:38

origins look normal, but the coronaries are

40:42

dilated, and because there was a reversal of flow, they thought

40:45

this patient has some form of fistula. So we did a CT.

40:50

Obviously, on CT, we saw large coronary arteries. It's clear.

40:54

But if you slowly look at the origin of the coronary arteries, so

40:58

if you look at them, I know on echo, they look normal.

41:03

But on CT, this is a very

41:07

abnormal course because the left coronary artery

41:11

is arising from the main pulmonary artery.

41:15

And this is how you can see echo was looking at this

41:18

image. And on this image, obviously, if you don't see

41:22

this fat plane, it looks like this is the sinus of Valsalva, this is the right

41:26

coronary artery, this is the left.

41:28

But

41:30

there is a fat plane, which obviously echo could not see,

41:34

and this is a patient that actually has a diagnosis

41:37

of

41:39

basically ALCAPA, and the management is very different

41:43

for these patients.

41:46

This is another very good example of how I think CT

41:50

really has grown as a diagnostic

41:54

test for so many difficult cases that we've seen.

41:57

This is a 48-year-old, came in with shortness of breath, dyspnea on

42:01

exertion. Patient had no prior medical care.

42:04

We did a CT in this patient. The CT was

42:08

done out of ER. And what we saw that the

42:12

contour of the heart was really abnormal.

42:16

And,

42:18

obviously, this contour was something that triggered the

42:22

CT because it was seen on an X-ray and a bedside echo that was done

42:26

on ER. But what we also saw

42:30

was other findings that

42:33

echo did not see.

42:37

If you see this patient's IVC, it's supposed to go in the

42:41

right atrium. But in this case, the inferior portion

42:45

of the interatrial septum is kind of absent, and this

42:48

IVC is draining into both right atrium and left

42:51

atrium.So if you see it on an axial image, this is

42:55

how it looks. So this patient has what is called as an

42:59

inferior sinus venosus ASD, along with a

43:02

large pseudoaneurysm

43:05

of the LV because of an old infarct that was

43:09

never

43:10

managed.

43:12

So CT is excellent for these kind of

43:16

things, especially problem-solving.

43:18

Now, quickly coming back to acute chest pain.

43:22

This is where the heart of CT is. So

43:25

chest pain, second most common cause of ER visits.

43:28

The most common cause is abdominal pain.

43:31

It leads to millions of ER visit just within US.

43:36

And we also know that cardiovascular disease is the highest

43:39

mortality and morbidity amongst different etiology.

43:43

So,

43:45

interestingly with CT,

43:47

the

43:49

thing that stands out, it has very high sensitivity. Okay?

43:53

Sensitivity means it finds the disease when it should.

43:57

The problem with CCTA is that sometimes

44:01

the specificity, whether it is significant or not, just off

44:05

of pure CT, I'm not talking about CT with FFR, it used to be a

44:08

little low, especially when we look into more than 50%

44:12

stenosis.

44:15

But again, if you want to rule in the disease, whether there

44:18

is significant disease or not,

44:21

CCTA is excellent. Only problem is it

44:25

does not have very high specificity.

44:28

But what is happening is that our CT is changing, and this is

44:32

what is driving all the newer guidelines that's coming out because our data is

44:36

changing. This data that I just showed you was a data

44:40

from 2018, not long back. From a research standpoint, I

44:44

consider 2018 to be recent

44:47

data. But what is happening on side by side is that our

44:51

technology keeps on changing every year.

44:55

So what our temporal resolution used to be, temporal resolution,

44:58

again, ability to scan, how fast you can scan.

45:02

We could get 250 millisecond or quarter of a second

45:05

resolution, which now has changed to much

45:09

less than that, like less than one tenth.

45:12

And one thing that really drove this is basically to

45:16

get an image in CT, you have to go 180

45:20

degree across the patient. Now,

45:23

when you have one X-ray tube, to get 180 degree, let's say this

45:27

tube moves around, fully around the patient in 0.3

45:31

seconds or 300 millisecond to get

45:34

180 degree,

45:36

you get 150 milliseconds. So this CT's temporal

45:39

resolution is 150 millisecond because that's how much time

45:43

it takes to go half around the patient.

45:46

Now, if you put two X-ray tubes, this X-ray tube rotates

45:51

90 degrees, and this simultaneously rotates 90 degree.

45:55

And what you get is, if it is 300 millisecond

45:59

and

46:00

rotation, but to get 180 degree rotation, you just

46:04

need quarter rotation. So the temporal resolution

46:07

automatically doubled.

46:11

So basically, you get 75 millisecond, and you can do

46:15

very high heart rates with these

46:17

exams.

46:19

Few things we know from data, and again, this is not recent data, this is

46:23

from 2011, that you can have--

46:27

CT gives us a lot of prognostic information.

46:30

So if you look at this, normal people, survival probability

46:34

for multiple years, and these are people who came in with chest pain.

46:38

These are not people who are general population.

46:41

These are chest pain people. If CT was normal versus

46:45

as you keep on having more and more disease, your

46:49

survival probability keeps on decreasing.

46:53

And even in non-obstructive disease, so when we

46:56

say that there is disease, but the stenosis is less than

47:00

50%, if you see here that

47:04

if patient had one vessel disease versus two vessel disease versus

47:08

three vessel disease, all non-obstructive, still,

47:12

your risk of

47:15

mortality is significantly higher.

47:17

So patient who has

47:20

three vessel disease has six-fold higher

47:24

mortality as compared to a patient that has

47:27

no disease.

47:31

Okay.

47:32

So

47:33

one question is, is normal CCTA, so let's say if somebody comes to

47:37

ER and we do a CT, can we say that normal

47:41

CCTA is an effective gatekeeper? So

47:45

to look at that, there have been extensive studies, and what we have

47:48

learned is that people who have normal CT from the

47:52

ER, their event rate in

47:56

those patients,

47:58

by event means cardiac event, cardiovascular event,

48:02

is less than 0.3. Whereas if you use a nuclear

48:05

scan to do the same triage, the event rate would

48:09

be 1.1. So what we have learned is that coronary

48:13

CT,

48:15

obviously, is very sensitive to find things.

48:18

And not only that, once it says that the patient does not have

48:21

disease, the outcomes of these patients are

48:25

significantly better.

48:27

One very good

48:29

study that has been done is called the SCOTH 心 HT study.

48:32

And in that study, what has happened is that they

48:36

followed

48:38

the Scotland population for five years,

48:42

and now very recently, this year, there was a 10-year

48:46

database from the same study, which kind of confirms this long-term

48:50

benefit. What they showed is that people

48:54

who got CT versus standard of

48:56

careAnd these are chest pain patients.

49:01

So with CT patients, the risk

49:06

of any event was significantly lower, whereas other

49:10

standard of care, the risk of incidents was significantly higher.

49:14

The reason this is because when you get a CT, you find

49:18

disease even if it is non-obstructive, whereas the conventional

49:22

care focused on obstructive disease.

49:26

So if you look at this data, and

49:30

one thing that really stands out to me, the death from coronary artery

49:34

disease was lower in the CT range. Non-fatal

49:38

myocardial infarction, so basically

49:41

even MIs that do not lead to death, is

49:45

significantly lower. And I think that this

49:50

completely is related to

49:53

increase in preventive therapy. So patients, when they get a CT, you

49:57

identify early non-obstructive plaque, and then people

50:01

are put on preventive therapies appropriately,

50:04

and they have longer-term survival from cardiovascular disease.

50:09

So

50:10

in terms of CT, the advantages that we get is that

50:14

you

50:15

can see every spectrum of disease, from no CAD

50:19

to severely stenotic CAD, and this is where it shines through.

50:23

And

50:25

for many, many years, nuclear scans were considered the standard of care.

50:29

Now, if you look at this from that standpoint,

50:32

there are three patients. All three are

50:36

50-year-old patients with atypical chest pain.

50:39

The one on the left has completely normal coronary artery disease.

50:44

This one on the right has three-vessel

50:48

non-obstructive coronary artery disease.

50:52

And this one on the left has

50:55

at least two-vessel disease, and on top of

50:59

that, this has low attenuation plaque

51:02

with

51:04

positive remodeling or basically expansion of the vessel, so also known

51:08

as high-risk plaque. So what we realized is that

51:12

if

51:13

all three patient would have gone through a nuclear scan pathway,

51:17

they would have the same nuclear scan, and it would have been read normal,

51:21

but the outcomes of these patients and the

51:25

long-term risk of these patients is significantly different.

51:29

So this is where CT really shows its strength,

51:33

and there have been multiple

51:36

studies done on in EUD, and all kind of

51:40

point to the same thing. CTs are faster

51:43

and they're safer, and they lead to

51:47

decreased long-term poor outcomes.

51:51

One thing I want to just share with you is that with CTs,

51:55

not only we identify plaque, not only we grade stenosis, we

51:59

also look at high-risk features. So there are main

52:04

four high-risk features that we talk about.

52:06

One is called as positive remodeling.

52:08

That means there is expansion of the vessel at that level.

52:12

Other is called as low attenuation plaque.

52:15

That means there are very low or hypodense areas

52:19

within, which basically people used to say these are fatty

52:23

rests inside the plaque. Then there was a napkin ring.

52:27

Napkin ring is when you see a small rim of high

52:31

attenuation around it. So that is a napkin ring feature.

52:34

And then there is spotty calcification.

52:36

That means you have calcified plaque, but within that, there are spotty

52:40

calcifications. All of these four are

52:43

considered high-risk features. And what we have

52:47

noticed is that out of these, at least in ROMICAT

52:51

study, which was done, the relative risk was highest with

52:55

spotty calcification.

52:58

But any high-risk feature is associated with a relative risk

53:02

of 32 as compared to

53:06

normal. Now,

53:08

what we have noticed is that anybody who has

53:13

two features like low attenuation plaque and positive remodeling,

53:17

any amount of features lead to

53:20

increase in poor outcomes, and two feature plaques are really,

53:24

really poor. And this was another study,

53:28

also a relatively recent study, in which what they did was they

53:32

measured the amount of low attenuation plaque in the same

53:35

SCOTHART or Scotland population.

53:38

And what they found is that if somebody has low attenuation plaque of

53:42

more than 4% versus less than 4%, they

53:46

really diverge in the incidence of MI

53:50

and death.

53:52

So basically,

53:54

just having higher amount of high-risk plaque

53:58

leads to poor outcomes in patients.

54:01

So we can now quantify plaques. We can tell

54:05

exactly how much plaque people have.

54:08

And

54:09

what we have learnt,

54:11

not just from quantification of non-calcified plaque, but even

54:15

from coronary artery disease studies, like even

54:19

just from calcium score, is that anybody

54:23

who has the total amount of plaque same, whether it is

54:27

obstructive or non-obstructive, if the total amount of plaque is same,

54:31

then their risk of events is same.

54:35

So is more associated with

54:39

symptoms.

54:41

Obviously, more stenotic the lesion, more are the chances somebody

54:44

has chest pain. But it is not

54:48

associated with

54:51

higher risk. The risk is equal if the

54:55

total amount of plaque is equal into patients, whether it

54:58

is obstructive or non-obstructive.

55:00

This is a little difficult concept to understand, and I

55:04

think it took me some time because we conventionally always

55:08

thought70% or more stenosis is really, really bad,

55:12

and if somebody has 20% but diffuse disease, it's not

55:16

bad. That is not true. And one last thing I want

55:20

to talk about is FFR. And

55:23

stenosis

55:27

is basically anatomical visualization.

55:30

What we can do now is we can add physiology to it.

55:34

So FFR is computational flow dynamics modeling

55:38

that what it does is that it looks at how the

55:42

stenosis is and how the model of the coronary artery

55:46

is, and they predict whether this vessel

55:50

is significant or not. And

55:54

FFR, just to give you a perspective, has

55:58

normal values. Basically, if it is more than

56:01

equal to 0.8, is considered okay. So in

56:05

this vessel, if this is 0.89, 0.85, this is okay.

56:09

If it is less than

56:11

or equal to 0.74, it is abnormal. So in this

56:15

example, if there's a drop happening to 0.7s,

56:19

like 0.72, 0.74, this block or this

56:23

stenosis is significant.

56:26

And there is a gray zone, between 0.74 and

56:29

0.79 is considered a gray zone. And in

56:33

those, we basically evaluate how the patient is doing, and

56:37

based on that, we decide. But one thing that we have learned,

56:41

I remember when I started the data part, I

56:44

said coronary CTs are very sensitive.

56:48

They lacked specificity. Now, FFR

56:52

has changed that. If you look at data now, in

56:56

patients with stable coronary artery disease who went

57:00

through every test. They went through a nuclear scan, they went through a coronary

57:04

CT without use of FFR, they went through a PET

57:08

scan, like a cardiac PET, and then they

57:11

added FFR to coronary CT.

57:15

And what we found is that once you add

57:21

FFR, the combined sensitivity and specificity analysis,

57:26

also known as diagnostic accuracy of any test, it was

57:30

the highest. If you look at the blue, it's the highest accuracy

57:33

amongst all tests, followed by a PET, like a

57:37

cardiac PET like we expected, and then the third was

57:42

basically coronary CT. So

57:45

with that, again,

57:48

I have a lot more data. Again, this is just a repeat of

57:51

SCOT-HEART that continued and now has given us

57:55

10-year data on this as well, and it's the same thing.

57:59

It continued from five years to 10 years, that the

58:03

benefit of coronary CT keeps continuing.

58:06

And one thing that

58:11

basically we learned from this is that when you do a CCTA,

58:15

more people go on preventive therapy and better are their long-term

58:19

outcomes.

58:21

So,

58:22

I won't go more in the data, but if you have any

58:25

questions from what I have. So let

58:29

me at least give you my information.

58:34

One second.

58:36

Oh.

58:37

And again, if you have any questions, please feel free to reach out to me.

58:41

This is my email. This is my contact.

58:43

You can reach out to me with any other things.

58:47

I want to have some time for questions, and I'm happy to stay over to just

58:52

answer any questions that you may have.

58:55

Well, thank you so much for that comprehensive lecture.

58:58

We will open up the floor for some questions.

59:01

If you've got any, please go ahead and put those in the Q&A box.

59:06

You may have covered everything, that nobody has questions.

59:10

There is so much to cover on cardiac CT.

59:12

I think I could have gone on for another one hour.

59:17

But again,

59:19

I'll conclude with this, though. But please feel free to put in

59:23

questions. That doesn't mean I'm logging off.

59:26

But one thing that I've learned, it's a

59:30

test that includes some nuances. It's a test that

59:34

needs good training to learn. But it's a test

59:37

that has significant impact on patients.

59:40

It is a test that really is changing

59:44

guidelines as we speak. Almost every new guideline that comes out,

59:48

if you look at the chest pain guidelines from

59:51

2022 to lipid guidelines that came out

59:55

last week,

59:57

every guideline has increased its confidence and

60:00

evidence for cardiac CT, which basically translates

60:03

to more number of tests going forward, and I

60:07

think more

60:10

impact of this test in management of our patients.

60:14

Again, when I say this, I know whenever I teach

60:18

cardiac imaging, I always say whether you are a cardiologist or a radiologist, it

60:22

has significant impact in our lives.

60:25

And I am sure eventually, most of us, the way cardiac CT goes,

60:33

or the way cardiovascular disease happens, most likely it impacts

60:37

all of our lives in our lifetime as well.

60:41

So,

60:43

first of all, thank you for the comments.

60:45

And there is one comment on how much iFR

60:49

is available and accessible and how much it costs.

60:52

That's a very, very good question.

60:54

So, one thing I know,

60:57

iFR traditionally is a term that is used for

61:01

invasive FFR. The FFR that I was showing is

61:04

CT FFR, which is a non-invasive FFR, basically,

61:09

measured from CT data.And

61:12

how much it is available. So, at one point,

61:16

FFR was very limited because there was only one

61:19

company that was FDA approved in US,

61:24

and

61:25

it was kind of a little bit of monopoly at that time.

61:27

But what has happened is, one, they have shown

61:31

amazing data over the years, that the impact of FFR has

61:35

been amazing. Even though it was only one company

61:38

doing most of the work, the impact was amazing.

61:42

Now, these days, we have multiple companies that are offering it, so

61:46

the access to CT FFR is

61:49

significantly improved over years as we go.

61:53

At the same time, it still is something that hospitals have to do a

61:57

contract with the company, send the cases to them, they process.

62:00

It's not like a box that you can put it in your hospital, so

62:04

many places I know they have reservation sending their

62:08

cases to an outside company, even though they do it with a lot

62:12

of security review and it

62:15

is not unsafe, but you can imagine it involves

62:18

steps of a CT leaving your

62:23

premises and going to a company. Even though you can structure it in a very safe

62:26

way,

62:27

there is still a little bit of practicality to it.

62:31

Now, it also comes to your second part of the question, how much

62:34

does it cost? Because there is an extra cost to it.

62:38

You are sending it somewhere, they charge something, and then

62:42

you basically charge the insurance.

62:44

And right now what we see with Medicare covering FFR,

62:48

90% of insurances that we deal with

62:51

basically reimburse us for

62:55

FFR.

62:57

But some

62:58

insurance companies still don't, and that's one limitation.

63:02

I see it as myself as if I

63:07

were to have an option of healthcare insurance provider, I would

63:11

always go with somebody who supports FFR, especially if I need

63:15

it.

63:18

To the point, how much does it cost?

63:21

Medicare reimburses $1,200, out of which

63:25

most goes to the company. So you can imagine that

63:29

almost it costs around $1,000 extra to

63:33

get FFR.

63:36

So,

63:38

to this point,

63:40

it is mostly available in US or in

63:44

big European countries. Many places, I remember last year

63:48

I gave some FFR talk in Asia,

63:53

and one country in Africa, they did not have FFR.

63:56

That's a problem with a technology like this.

63:59

It's not universally available.

64:04

There are a couple questions in the Q&A box, if you're able to open

64:08

that up.

64:10

It's the little box with the question mark.

64:13

Oh, awesome. I now know where is Q&A.

64:17

Okay.

64:19

One is, "Which software is the most commonly used software for

64:23

calcium scoring? Which is more reliable?"

64:28

Again, software is calcium scoring, is

64:31

a very standardized protocol. I would say

64:35

reliability, almost any software that allows

64:39

calcium scoring, it would be reliable because we have standardized

64:43

that you can do it only at a certain slice thickness.

64:47

You can only do it with a threshold of 130, so accuracy is

64:50

very high. In terms of software, it depends what is available.

64:55

Many PACS, like regular PACS systems, have it.

64:59

In our institution, we have three

65:03

softwares that can do. Our PACS can do it straight off

65:07

the

65:08

reading station. We have access to what is called as Vitrea

65:12

that does calcium scoring very accurately.

65:15

We have what is called a Circle CVI-42.

65:18

It does. TeraRecon does.

65:21

I think I can name more than five

65:25

or 10 softwares that you can do, but I would say start with your

65:28

PACS. Your PACS should be able to do.

65:31

GE

65:33

has AW that can do it. Siemens has syngo.via that can do it.

65:37

There are so many softwares that do

65:39

calcium scoring.

65:43

Margaret has a question. It says, "If the patient has

65:47

extensive coronary artery calcification and normal stress PET, is it still a

65:50

concern?"

65:52

I would say

65:54

concern regarding what?

65:56

One thing that it does tell me is that

66:00

if there is coronary artery calcification, then the patient has

66:04

significant disease. Normal PET tells me that the

66:08

disease is not hemodynamically significant.

66:11

So that means the patient still needs a lot of preventive

66:15

therapy,

66:16

but putting a stent in that disease is not going to help

66:20

this patient's symptoms. So,

66:23

concern in terms of future outcomes is

66:27

there, but concern in terms of acute

66:31

angina, once there is a normal PET, is not there.

66:35

That's how I would interpret these two

66:37

tests separately.

66:40

Another question is, "Does improvement in CT means that

66:44

in two to five years from now, we'll have a much more easy to do cardiac CT?"

66:48

Absolutely. I think we are going to have

66:52

easier to do cardiac CT, probably will be done without any

66:55

heart rate control to a point that we'll be doing more and more of

66:59

it.

67:01

So for example, imagine you can just put the patient in the gantry, do it like a

67:04

routine chest CT. Your throughput becomes from doing 15

67:08

cardiac CTs a day to 55 cardiac CTs a day, and many programs

67:12

are doing it. I know of one program that does 80 cardiac CTs a

67:16

dayYou can tell how fast things are happening.

67:20

And secondly, which I am actually more excited about,

67:24

our accuracy is going to increase.

67:26

Once our accuracy increase, what it tells me is that in five

67:30

years, whatever data I showed you, which is amazing for CT, it's going to

67:34

get better. This is what I'm most excited about.

67:38

Then, do you see FFR making big future improvements

67:42

or is it a mature technology? I do not know exactly,

67:46

but if I were to put my

67:50

bet on, I think it's a mature technology at this point.

67:55

The way FFR would change, though, if our CT data

67:59

improves, so for example, and it has been shown already.

68:03

If you compare accuracy of FFR

68:07

using relatively excellent

68:11

scanners, but the older generation scanner, like energy

68:14

scanners versus photon counting scanners.

68:18

If you compare photon counting scanner, FFR is more accurate.

68:21

It correlates better with CAT. So I think even though

68:25

FFR as a technology is mature, with improvement

68:29

in CT data, because FFR is data

68:33

in, data out. If you put garbage in, garbage out.

68:37

If you put better data in, it would be more accurate.

68:40

So I think that's how FFR is going to grow.

68:44

That we are going to one day show that CT FFR is

68:47

exactly same as invasive FFR.

68:51

So you only do CAT when really you have

68:55

to do an intervention. If you do not plan to do intervention, you probably do not

68:58

need to do CAT. Again, it's not true at this point,

69:02

but that's kind of future I am hoping for.

69:07

Next one. How do you mention positive remodeling in the report?

69:11

So we use CAD-RADS 2.0, and we use high-risk features.

69:15

So, that's how we mention that the plaque has high-risk feature.

69:19

We do mention which high-risk feature as well.

69:22

So

69:23

there is 50% stenosis, and there is positive

69:27

remodeling.

69:29

Makes it seem like 0% on

69:32

this.

69:35

No, so CAT is measuring only negative

69:38

remodeling. That means CAT is only measuring the stenosis of the

69:42

lumen. So CAT would still show 50% stenosis.

69:45

What CAT cannot ever tell you is plaque features.

69:49

So plaque features are only unique to

69:53

CT. So whether there is positive remodeling or spotty

69:57

calcification or

70:00

low attenuation plaque,

70:02

any of these features, or napkin ring, for the matter of fact, any,

70:06

these can never be seen on CT. So,

70:09

if you say 50% stenosis, you say 50%

70:13

stenosis because you see 50%

70:17

stenosis, or 50% of narrowing of lumen.

70:21

So which should correlate with CAT.

70:23

So CAT should not say 0% in that case.

70:26

We mention positive remodeling when the plaque goes

70:30

beyond the expected boundary of the

70:34

wall, and we sometimes even use a ratio of

70:37

1.1. So basically it has to go 10%

70:41

outside the boundary of the wall. That's when we

70:44

say positive remodeling. And

70:49

our report should still correspond to

70:53

CAT. I hope I answered that.

70:57

I love the next QA. Deep learning is prone to

71:01

hallucinations. I love that, and I am also a

71:05

little skeptic of how many deep learning algorithms are coming.

71:10

But is there not a risk when using deep learning to

71:13

enhance image quality, artifacts will be introduced and

71:17

mistaken for improved image quality?

71:19

I think you have a very good point here,

71:23

and

71:26

I have the same worry on this, but at the same time, this

71:30

puts a little bit of onus on us. Before we introduce

71:34

any deep learning algorithm, we always should do

71:38

image quality assessment and understand how it is

71:42

changing the data that we see.

71:44

Now, every time doing a full multi-center,

71:48

nationwide or multi-nation study would be difficult.

71:51

So that's why at least when we incorporate any

71:56

deep learning, whether we do it in cardiac MR or we do it in cardiac CT,

72:00

before we introduce it, we do a in-house study of at

72:04

least 25 to 50 patients in which we do

72:08

what is the current standard of care and what is the so-called new, and we

72:12

compare these things objectively

72:15

and as well as subjectively. Like do I

72:19

think there is an

72:21

artifact introduction? And that is

72:24

basically the role what FDA also does.

72:27

I do believe, at least in

72:30

US, that any AI algorithm that goes

72:34

through FDA approval process, it has to go

72:38

through the same benchmarking. We still, at

72:42

least I can tell you of our system, we still do not

72:45

change any new deep learning algorithm without doing our

72:49

own internal assessment. Because that's my biggest worry,

72:53

that we are going to introduce some artifacts without knowing.

72:56

And even though experienced readers can easily figure out that this is an artifact

73:00

versus this is an image enhancement,

73:03

some young readers, especially our trainees, we are an academic institution,

73:07

they would really struggle to have that.

73:09

So we are very cautious of what you just mentioned, and I think that's a

73:13

very good question.

73:16

The other question isCan deep learning mislead about clean or

73:20

calcified?

73:22

Yeah. No, I think it's just a continuation of the same question.

73:25

But I agree to you.

73:27

But

73:29

my bigger concern actually is not that DL will introduce

73:32

calcium. I hope it does not do it, and if

73:37

there is such a poor deep learning algorithm, I hope it never gets FDA

73:41

approval. But

73:43

my bigger worry, which FDA or nobody is actually

73:47

looking at, is we are making guidelines based on

73:53

total plaque quantification, and that depends on

73:57

quantifying the plaque on a coronary CT.

74:00

Now,

74:01

as coronary CT is going to improve, I bet you six

74:05

months from now, there will be a new deep learning algorithm that would enhance the

74:09

image quality further.

74:11

And the question is, FDA is looking at

74:14

just, did it introduce artifacts?

74:17

Did it

74:19

lead to any big issues? Nobody is actually measuring plaque and

74:23

saying that if on the previous algorithm,

74:27

the plaque was quantified as 250,

74:30

is it still quantified as 250 on this newer algorithm?

74:34

So I do think this is a very big topic in the CT

74:38

community that we need to standardize or find a

74:42

way to use phantoms to

74:46

basically

74:47

get some correction factors for every improvement

74:51

that we are getting, because nobody's going to stop improving CT.

74:54

This CT technology as well as deep learning is going

74:58

to keep coming. Now, can we introduce some way,

75:03

like a phantom or something, that we can use it for correction factors that

75:07

every time a new technology comes, we

75:10

basically link it to what was baseline previously so

75:14

that we do not change our quantitative numbers?

75:18

Because I bet you in two years, there will be guidelines saying,

75:22

if the total plaque volume number is this, your patient

75:26

should get this therapy. And if the patient's total plaque volume is this,

75:30

should get that therapy. And that is what I am most concerned about.

75:34

But there is enough discussion happening on this that I

75:38

am very confident that

75:40

this hopefully will not happen in our field.

75:47

I think you got all the questions.

75:50

Thank you.

75:51

Thank you so much for your lecture and for staying on to answer all those extra

75:54

questions. We really appreciate it.

75:56

No, absolutely. Thank you. Thank you for the discussion at the end, and I

76:00

really appreciate the opportunity.

76:02

Of course. And thank you to our audience for staying on and participating in

76:06

this NOOM conference and asking such great questions.

76:09

You can access a recording of today's conference and all our previous NOOM

76:12

conferences by creating a free account.

76:14

We will also email out a link to the replay later today.

76:18

Be sure to join us next week on Thursday, April 9th at 12:00 PM Eastern, where Dr.

76:22

Alka Singhal will deliver a lecture entitled Ultrasound Physics.

76:26

You can register for that at modality.com, and follow us on social media for

76:30

updates on future NOOM conferences.

76:32

Thanks again, and have a great day.