This publication was supported by an educational grant from
Amersham Health, Princeton, NJ. The opinions expressed in this
publication are those of the authors and not necessarily those of
Dr. Bae reports relationships with Tyco Healthcare and
Mallinckrodt through patent agreements and as a consultant. Dr.
Fishman reports relationships with Siemens Medical Solutions and
Amersham Health as a consultant. Dr. Foley reports a relationship
with GE Medical Systems through an investigator agreement. Dr.
Naidich reports a relationship with Siemens Medical Solutions
through its Advisory Board and as a consultant. Dr. Saini reports a
relationship with GE Medical Systems through research support. Dr.
Becker, Dr. Sahani, Dr. Siegel, Dr. Tahktani, and Dr. Zinreich
report that no such relationships exist.
is an Associate Professor and CT Section Chief at the University
of Munich, Germany.
The approach to cardiac computed tomography (CT) differs from
that of routine CT in several key respects. First, because of the
need for very high temporal and spatial resolution, we use a
16-detector-row CT scanner for all cardiac applications. We also
reduce the table speed to 6 mm/sec in order to acquire enough
projections to adequately reconstruct images. As a result, cardiac
scanning takes longer than other applications of CT angiography
(CTA), such as aortic scanning. In addition, we obtain a digital
electrocardiographic (ECG) tracing during scanning, so we can
retrospectively reconstruct images at a dedicated time point,
typically the diastolic phase of the cardiac cycle.
Table 1 outlines the scanning protocols we use for
4-detector-row and 16-detector-row cardiac CT scanning. We use the
thinnest collimation available with the 16-row CT scanner, 0.75 mm,
and the shortest rotation time, 420 msec. As a result, image noise
is increased. To compensate, we increase the tube current to 500
mAs from the 400 mAs we use with the 4-row CT system. A clear
advantage of a 16-row CT scanner over a 4-row scanner is a halving
of scan time, to 20 seconds. With ECG pulsing--in which full X-ray
power is used only in mid-diastole--the effective radiation dose
delivered to the patient with the 4-row CT system is approximately
4 mSv, and approximately 5 mSv with the 16-row CT system. Both are
similar to the radiation dose associated with a routine chest CT or
conventional cardiac catheterization.
Figure 1 shows a comparison of three-dimensional (3D) renderings
from the same patient imaged with a 4-row CT scanner and, 1 year
later, with a 16-row CT scanner. The patient underwent the initial
CT study as part of an evaluation for acute chest pain. On 4-row CT
scanning, a stenosis in the proximal portion of the left anterior
descending coronary artery was evident. She subsequently underwent
cardiac catheterization and stent implantation. One year later, we
evaluated stent patency using a 16-row CT scanner.
Comparing the 3D renderings, it is obvious that the image
acquired on the 16-row scanner looks rougher on its
surface. This is a result of increased image noise. It is equally
obvious that the visualization of small arteries, such as the
diagonal branches, is much clearer with the 16-row scanner as a
result of its superior spatial resolution. Similarly, the right
coronary artery is too small to be visualized by the 4-row CT
system, but it can be appreciated in the image acquired on the
16-row CT system.
CT angiography of the coronary arteries often detects the
presence of coronary atherosclerosis, which is characterized by
plaques of varying density. Low-density atherosclerotic plaques can
be categorized by an enhancement of up to 50 HU,
intermediate-density plaques by an enhancement of 50 to 100 HU, and
calcified, high-density plaques by an enhancement of more than 350
Optimal vascular enhancement for coronary CTA is 200 to 300 HU.
We performed a study to determine the effect of iodine
concentration and flow rate on enhancement density.
As shown in Figure 2, we found that injecting 300 mgI/mL contrast
material at 2.5 mL/sec resulted
in a mean left ventricular (LV) contrast enhancement of
approximately 200 HU. Injecting low-concentration contrast material
(300 mgI/mL) at high flow rate (3.5 mL/sec) or higher-concentration
contrast material (400 mgI/mL) at a lower flow rate (2.5 mL/sec)
produced a similar result: LV enhancement of approximately 200 to
300 HU. Injecting high-concentration contrast material (400 mgI/mL)
at a high flow rate (3.5 mL/sec) resulted in an enhancement density
of approximately 350 HU, which overlaps the enhancement density of
coronary calcifications and may prevent their detection. Plotting
iodine delivery rates against HU shows that for a target LV
contrast enhancement of 200 to 300 HU, it is necessary to
administer approximately 1 gram of iodine per second.
In addition to the iodine delivery rate, several other factors
influence enhancement of the LV and the coronary arteries. These
include the viscosity of the contrast material, the type of
peripheral access through which contrast will be delivered and
where it has been placed, the blood volume as reflected by body
weight, cardiac output, and the use of a saline chaser.
Given that the temporal resolution of cardiac CT studies is
>200 msec, it is necessary to lower the patient's heart rate to
<60 bpm before the initiation of scanning, primarily through the
use of beta blockers. We first ensure that the patient has no
contraindications to beta blockade, such as asthma, aortic
stenosis, atrioventricular block, or severe LV dysfunction. Then,
immediately before scanning, we inject 5 to 20 mg metoprolol
Figure 3 demonstrates the effect of beta blockade on contrast
enhancement. Without first administering a beta blocker, we
injected a 20-mL bolus of contrast material (orange curve). We then
repeated the test bolus injection after administration of a beta
blocker (yellow curve). We found that beta blockade delayed the
time to peak contrast enhancement by 4 seconds, and increased peak
enhancement by 10%.
Figure 4 shows a patient whose heart rate upon referral for CT
was 82 bpm. After injection of a beta blocker, the heart rate fell
to 65 bpm, and we were able to clearly visualize the coronary
anatomy, including the common trunk of the left main coronary
artery, ori-ginating together with the right coronary artery from
the ostium of the right coronary artery.
Accurate contrast timing is critical when imaging with a 16-row
CT scanner. If scanning begins too early, contrast material will
still be in the pulmonary arteries at the time of data acquisition,
rather than in the LV or coronary arteries.
The use of a saline chaser, which pushes the contrast bolus
forward, also influences scan timing. Figure 5 shows enhancement
over time following injection of a contrast bolus with (yellow
curve) and without (orange curve) a saline chaser. The saline
chaser hastens arrival of the contrast bolus by 4 seconds and
increases peak enhancement by 30%.
There are several advantages to using a dual injection
consisting of a contrast bolus followed immediately by a saline
chaser. First, if the contrast material is injected through a
peripheral vein, use of a saline chaser produces a time-density
curve similar to that observed with a central venous catheter,
without the use of a saline chaser.
In addition, use of a saline chaser produces much more homogenous
Less contrast media is required.
Because we are able to reduce the volume of contrast media, the
risk of nephrotoxicity is reduced as well.
In cardiac CT, the osmolality of contrast media may be important
to consider. A study by Dunkel et al,
conducted in patients undergoing cardiac catheterization, has shown
that the isosmolar nonionic dimer iodixanol induces only minor
changes in cardiac function, whereas the ionic dimer ioxaglate and
the ionic monomer diatrizoate induce pronounced effects. Whether
the osmolality of contrast media has the same effect on heart rate
and, therefore, image quality in patients undergoing CTA is
unknown. We are considering conducting a study to answer that
A study by Aspelin et al
suggests that isosmolar contrast media may also reduce the risk of
nephrotoxicity, at least in patients with chronic renal
insufficiency undergoing cardiac catheterization. We are now
testing whether that observation can be duplicated in patients
undergoing CT scanning. The study is enrolling patients with a
baseline creatinine >1.5 mg/dL. Regardless of the type of CT
examination, patients are injected with 100 mL of 270 mgI/mL
contrast material at a flow rate of 5 mL/sec. Patients are
followed-up for 7 days after scanning.
Preliminary data from the first 20 patients demonstrate no
increase in serum creatinine following exposure to isosmolar
contrast media. In fact, the mean serum creatinine level decreased
from 1.7 mg/dL at baseline to 1.4 mg/dL at 7 days.
Table 2 details our protocol for administration of contrast
media in multidetector cardiac studies. We use contrast media
composed of a nonionic isosmolar dimer at a concentration of 320
mgI/mL. We determine the circulation time by injecting a 20-mL test
bolus, then add another 5 sec to the bolus arrival time to allow
the contrast media to reach the LV system. We then inject 80 mL of
contrast material at a flow rate of 3 mL/sec, followed by a 50-mL
saline chaser and begin image acquisition after the specified
delay. The resulting dataset produces superb two-dimensional (2D)
and 3D images.
Cardiac CT is unlikely to be used to evaluate the large number
of patients with chronic stable angina, as it is unable to
accurately grade coronary artery stenoses. It shows great promise
in the evaluation of patients with acute or atypical chest pain,
however. In this group of patients, CTA can detect intracoronary
thrombus (Figure 6), an application similar to its established role
in the evaluation of pulmonary embolism.
In addition, studies comparing CTA with coronary angiography
have determined that CTA has a high negative predictive value for
significant coronary artery disease.
If this proves to be true for symptomatic patients with only
low-to-intermediate pretest probability of coronary artery disease,
CTA may reduce the need for invasive coronary angiography in such
CTA also has a promising future in screening for coronary artery
disease in asymptomatic but intermediate- to high-risk patients. By
detecting variations in the enhancement density of coronary
lesions, CTA has the potential to identify vulnerable plaque, with
its lipid core and fibrous cap. In this way, CTA may guide the
aggressiveness of risk factor modification in asymptomatic
Finally, we often use CTA in the assessment of coronary artery
bypass grafts, as it enables us to examine with high spatial
resolution both arterial and venous grafts (Figure 7). It is a
powerful tool for studying coronary anomalies, which occur in 1% to
2% of the general population. We are increasingly using CTA to
define the anatomy of the pulmonary veins, in conjunction with
Three-dimensional renderings from the same patient imaged (A)
with a 4-row CT scanner and (B) with a 16-row CT scanner 1 year
later. (A) The 4-row CT scanner detected a stenosis in the
proximal portion of the left anterior descending coronary artery.
(B) The image acquired on the 16-row scanner shows more image
noise but better visualizes small arteries, such as the diagonal
branches and the right coronary artery.
. Injecting 300 mgI/mL contrast material at 2.5 mL/sec resulted
in a mean left ventricular (LV) contrast enhancement of
approximately 200 HU. Injecting low-concentration contrast
material (300 mgI/mL) at a high flow rate (3.5 mL/sec) or higher
concentration contrast material (400 mgI/mL) at a low flow rate
(2.5 mL/sec) produced a similar result: LV enhancement of
approximately 200 to 300 HU.
Beta blockade delays the time to peak contrast enhancement by 4
seconds and increases peak enhancement by 10%.
After intravenous beta blockers reduced the heart rate from (A)
82 bpm to (B) 65 bpm, CTA clearly depicts the coronary anat-omy,
including the common trunk of the left main coronary artery
(arrows), originating together with the right coronary artery
from the ostium of the right coronary artery.
A saline chaser hastens the arrival of the contrast bolus by 4
seconds and increases peak enhancement by 30%.
(A) In a patient with atypical chest pain, CTA detects a thrombus
(arrow) in the left anterior descending coronary artery. (B) This
application is similar to the established role of CTA in the
evaluation of pulmonary embolism (arrow).
Assessment of bypass graft patency is a common application of
ELLIOT K. FISHMAN, MD:
That's great. Any specific questions?
S. JAMES ZINREICH, MD:
I was a little curious about the contrast protocol; you're not
triggering this, you're using a test bolus?
CHRISTOPH R. BECKER, MD:
We are using a test bolus.
DAVID P. NAIDICH, MD:
How are you actually doing that? It's not clear. You're using 20
mL, but you're acquiring images at what time frame?
We are acquiring the images at every second. I know this is quite a
lot. Not only do we do it at every second, but we do this for 40
seconds. What is very important is not to let the patient hold the
breath, because if you are doing so, you are forcing people's
Valsava maneuver and you get a different time density curve that
does not allow you to conclude what will be the final aspect.
SANJAY SAINI, MD:
Are you measuring the aorta?
The ascending aorta, yes.
Why not use bolus tracking?
Bolus tracking? You are right, it works pretty much the same way,
and it's a robust technique. Our technicians liked the bolus
tracking much more than the test bolus. But for research purposes,
to learn more about this delay factor, we currently are using the
test bolus just to gain more experience in this, not in a
You are reducing the dose, in the actual acquisition parameters, as
you are doing these tests?
Yes, it's a low dose acquisition.
KYONGTAE T. BAE, MD, PhD:
If you use the test bolus out, you ask patients not to hold their
For the test bolus, you could breathe normally.
Right, but when you do real examination, you want the patients to
hold their breath.
Do you think that would be something to make a difference?
No, not really. As I said, we are adding 5 seconds to the peak
arrival time of the test bolus. If you are requesting the patient
to hold his or her breath initially when the contrast media is
injected, you have 10 seconds prior to the scan. By then, the
result of the maneuver has already finished, and he has a normal
circulation. So then this Valsalva maneuver doesn't really take any
effect anymore, and you have a rather homogenous enhancement.
What is the correlation between calcified plaques and the coronary
wall and luminal stenosis?
Well, actually there has been a lot of work done already from the
electron beam CTA without contrast. As a simple rule of thumb,
until you see the rest of the lumen in CTA, you are pretty
confident to rule out the significant stenosis in this particular
But once you have a dense clot of calcium in the coronary artery
that does not allow you to see the residual lumen, you have a
positive predictive value of approximately 50%, so you can get a
good point on whether or not there is stenosis.
W. DENNIS FOLEY, MD:
Perhaps I have an explanation as to why the saline chaser would
increase the aortic attenuation on a test bolus. Well, not so much
in the test bolus, but let's take it to the definitive bolus--would
you expect that to happen?
Yes, that is a pretty good point. As a rule of thumb, of course,
the smaller the amount of contrast media you are injecting, the
higher the benefit of using a dual injection. Well, this 30% is
certainly true for the test bolus, and I would expect the benefit
of a dual injection and a scan that lasts 20 seconds in the range
of 10% or so, but we haven't finally determined this. But this will
be also part of our further studies.
If you see this curve, and then your next slide showing the Hoppers
paper, you get more homogeneous enhancement with saline chase. But
if you go back to your slides, then it looks like saline chase is
not more homogeneous than flush.
I need to say that the saline chase can be somewhat cumbersome.
Because we have seen, if you are pushing the contrast media
further, you have a rather long scan range and you get a kind of
If you are doing a runoff study, you get a kind of washout in
the distal part of the arteries as well. It happens pretty much the
same way, so particularly if you have a long scan range, it is not
an advantage to use a saline chaser. It's only an advantage if you
have a small bolus and a small scan time.
I think you have to make sure that if you inject a certain amount
of iodine, then you can get the saline flush benefit. But if you
try to cut that out, and replace with a saline, at the tail end you
may have a washout.
Exactly, exactly. We have seen in CTAs of the abdomen, for
in-stance, that we have a complete washout of the femoral arteries
at the end of the scan.
When you're doing an aorta, say someone has a question on
thoracoabdominal aneurism or it is a question of dissection and how
far it goes, are you altering your protocoling when you're starting
from the top of the arch and going down to the bifurcation? Are you
changing the method by which you give contrast, and acquisition
protocol, and so forth, to account for the fact that even on a
16-detector, you are really talking about a huge volume of data
Actually, amazingly, with the 16-row CT scanner, the scan time is
not getting as short as one may expect, because the tendency is
also to get the advantage of highest spatial resolution. So, you
are using thinner concentration than you have used before. So the
range of the entire scan toward abdomen would last 25 to 30
seconds, even with the 16-row scanner.
So, in this respect, there is not so much difference between
these two scanners, in particular in dissection. The same holds
true for congential heart disease. We are using the test bolus all
the time because it's hard to really predict when the arrival is in
the different lumen, and in the different heart chambers in
congenital heart disease. So in this respect, we choose a more
Are you using 1-mm col-limation to cover that entire distance?
Well, in every CTA we use the thinnest collimation available.
Why are you doing that? Are you changing the rate at which you are
giving contrast? It's basically the same across the board?
Well, actually, between the 4- and 16-row CT scanner, the reduction
of scan time is roughly 15%. So we are also cutting down the amount
of contrast injected by 15%. Or, what we are doing is keeping the
amount of contrast media the same, but injecting it at the higher
You're not changing the duration?
I want to go back to emphasize two points, specifically for cardiac
imaging. I know you mentioned that everyone gets beta blockers,
which is what we are doing as well. But you did mention isosmolar
contrast and heart rate change. How significant a factor is
Well, I need to refer to one study that has been performed with
cardiac catheters. Indeed, I don't know the effect this will have
on the CTA studies. This will be one of our future projects.
Just to interject, we've looked at the same issue: you give beta
blockers, and you are seeing heart rates at about 60. First of all,
just putting the machine on, I think, raises people's heart rate
immediately, so that is one thing we've seen. But you get the heart
rate down to
60, you start injecting the contrast, and you are watching it, and
it is 120 right away. Again, typically for 16-slice CT, you like it
to be under 90. But we found the same thing, that going to
iso-osmolar contrast does have less of an effect.
The other thing to re-emphasize is the issue of different
contrast concentrations. Is there a point at which there's too much
iodine? In the cardiac imaging, your results have been that just
going to higher concentrations isn't going to work out. Can you
just expand on that a bit?
Well, actually, that's true. The higher the iodine concentration,
the better its detection of the small vessels, and this is true for
the pulmonary arteries, and I have hardly seen any calcifications
in the mesenteric or visceral arteries.
So, we are using the high injection rates for these particular
areas. But for the coronary arteries, there has been some work done
indicating that you are able to quantify the calcifications in the
CTA images. But this can only be done once you are really able to
delineate these classifications, and therefore the enhancement has
to be no higher than 300 HU.
You said that high concentration contrast will result in less
homogenous management. So with the high concentration of contrast,
you have more variation?
Right, with the high concentration (400 mgI/mL) injected at 3.5
mL/sec, you see a lot of scatter, most likely due to the viscosity.
I need to re-emphasize that the contrast media had not been warmed
up. We are now warming up all the contrast media. This is, of
course, of concern here.
Okay, thank you very much.