Patient characteristics, the approach to contrast injection, and the scan itself all influence contrast enhancement.
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
Amersham Health. <
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 Assistant Professor of Radiology at the Mallinckrodt
Institute of Radiology, Washington University School of Medicine,
St. Louis, MO.
The goal in optimizing contrast delivery is to achieve the
greatest contrast enhancement with the least amount of contrast
material. No single approach to contrast delivery is effective in
all cases, however, as many factors influence contrast enhancement.
These factors can be divided into three categories: the patient,
the contrast injection, and the scan itself.
Patient factors that influence contrast enhancement include the
target organ of interest, the specific diagnostic application, the
patient's body size, cardiac output, vascular access, renal
function, age, and gender. Factors to consider in contrast
injection include the concentration and volume of the contrast
material; the rate, pattern, and duration of the injection; and
whether or not a saline flush is used. Scanning parameters are also
important, particularly the delay between contrast injection and
the initiation of scanning, scan duration, whether bolus tracking
or some other method is used to adjust for variations in the
patient's cardiac output, whether the scan must capture multiple
phases of contrast enhancement, and the degree of radiation
With 16-slice CT, the short duration of scanning makes the
timing of image acquisition critical. Faster injection and higher
concentration contrast material can be used to improve contrast
enhancement and, perhaps, reduce contrast volume. However, the
injection duration and scan delay that are appropriate for a
single- or 4-slice CT scan must be recalibrated to optimize
contrast enhancement for a 16-slice CT scan.
The effect of injection rate on arterial enhancement is shown in
Figure 1. Three time-enhancement curves are created by simulating
the injection of contrast material at rates of 1, 3, and 5 mL/sec,
while keeping the volume (150 mL) and the concentration (320
mgI/mL) constant. The simulation model is based on an adult man
with a body weight of 150 pounds, a height of 5 feet 8 inches, and
a normal cardiac output. Two trends are apparent from these
time-enhancement curves: As the rate of injection increases, the
degree of contrast enhancement increases and the duration of
contrast enhancement decreases.
The duration of contrast enhancement is prolonged by slow
injection. Prolonged enhancement is preferred with slower CT
scanners, to match the long scan duration. It is less critical with
fast scanners, however. As a result, fast injection of contrast
medium is better suited to multislice CT angiography (CTA). In
addition, fast injection achieves a higher level of arterial
enhancement without an increase in contrast volume. Conversely,
with fast injection it may be possible to reduce contrast volume
and still achieve an acceptable level of contrast enhancement.
There is a limit as to how much contrast volume can be reduced,
however. A minimum contrast volume is required for the following
reasons. First, patient safety precludes increasing the injection
rate indefinitely, as fast injections increase the risk of
extravasation and other physiologic complications. Second, contrast
enhancement must be sustained at a desirable level over the finite
duration of the CT scan. Third, as contrast medium is injected
intravenously, it is diluted by the central blood volume before
reaching and enhancing the target organ. A sufficient iodine mass
must be achieved within the circulatory system to generate a
diagnostically useful level of contrast enhancement. The critically
required minimum contrast volume for mixing with the central blood
reservoir depends on patient size, blood volume, and
High injection rates are effective for dual-phase CT scanning.
Figure 2 shows arterial and hepatic time-enhancement curves
simulated with injection rates of
2 and 5 mL/sec. Compared with slow contrast injection at 2 mL/sec,
rapid injection at 5 mL/sec magnifies the difference in the degree
of enhancement during the arterial and portal venous phases, and
increases the temporal separation between their respective
Injection bolus shaping
Bolus shaping is another way to modify contrast delivery.
Uniphasic injection is the most common way to administer a bolus of
contrast. It involves delivery of contrast material at a constant
rate throughout the injection, with a resulting peak in aortic
enhancement near the completion of the injection.
An alternative approach, biphasic injection, involves a fast
injection of contrast material followed by a slow injection. The
advantage of the biphasic injection is that it can prolong contrast
enhancement. However, neither uniphasic nor biphasic injections
achieve the ideal uniform prolonged contrast enhancement.
A third, and perhaps more effective, approach is the
exponentially decelerated injection, which involves an exponential
decline in the injection rate over time. The exponentially
decelerated injection has been tested mathematically and in
animals. In a recent human study, we compared two different
40-second injection protocols in the same patients.
The first involved uniphasic contrast injection at a constant rate
of 4 mL/sec and the other, exponentially decelerated injection
following the formula:
We found that the exponentially decelerated injection generated
uniform contrast enhancement while using less contrast material
(134 mL, versus 160 mL with uniphasic injection), although peak
en-hancement was reduced (Figure 3).
Uniform contrast enhancement is desirable in CTA and cardiac CT,
as well as in brain perfusion studies.
One of the advantages of uniform enhancement is that it becomes
less critical to time image acquisition precisely to correspond to
peak contrast enhancement.
Figure 4 illustrates the effects of concentration on aortic and
hepatic enhancement. Three time-enhancement curves are simulated
for the aorta and liver by varying the concentrations of contrast
medium (300, 350, and 400 mgI/mL) but keeping the total iodine mass
constant (42 g). The volume of contrast medium corresponding to
these concentrations is 140, 120, 105 mL, respectively. When the
same injection rate of 5 mL/sec is used for all injections,
contrast medium with a higher concentration delivers more iodine
mass per second, resulting in earlier and greater peak aortic
enhancement. Since the contrast volume is reduced with
high-concentration contrast material, the duration of enhancement
is shorter, but the net effect is an increase in aortic contrast
enhancement. The effect of contrast concentration on the liver is
not as pronounced as in the aorta.
There are several potential advantages of using
high-concentration contrast material. First, we can deliver more
iodine without increasing the injection rate. Conversely, the
injection rate can be reduced while still achieving the same
desired level of enhancement. Improvement in enhancement is evident
when 370 mgI/mL contrast material is compared with 300 mgI/mL
contrast material. However, the degree of im-provement in contrast
enhancement with 370 mgI/mL contrast material as compared with 350
mgI/mL contrast material may be marginal. Cost savings are another
potential advantage of using high-concentration contrast media, as
price generally may vary by volume, not by iodine
A major disadvantage of high-concentration contrast material is
viscosity. The higher the concentration, the more viscous the
contrast agent becomes.
Injection duration may be the most important factor in
optimizing contrast delivery in CT. Injection duration critically
affects both the degree of contrast enhancement and the timing of
peak contrast enhancement and, therefore, optimal scan timing.
The contrast enhancement curves in Figure 5 show the effect on
peak aortic enhancement of contrast injections ranging from 1 to 30
seconds in duration.
As the injection duration increases, peak enhancement is delayed
and the enhancement curve becomes more asymmetric. With short
injections, the timing of peak contrast enhancement depends
predominantly on physiological variables, such as cardiac output
and circulation course. With long injections, however, the
influence of injection duration on the timing of peak contrast
The approach to determining injection duration must be adjusted
with advances in CT scanner technology. In general, injection
duration must be sufficiently long to produce good enhancement.
With a shorter scan, a reduced injection duration may be used. But
if the injection duration is too short, contrast volume may not be
adequate and contrast en-hancement may be suboptimal.
With single- and 4-slice CT scanners, a common approach to
determining the injection duration for CTA is to keep the injection
duration identical to the scan duration. This approach, however,
may no longer work with 16-slice scanners, as scans are much
shorter. An injection duration equal to the scan duration may
result in poor enhancement. We may increase enhancement by using a
higher injection rate, but there are practical restrictions on the
degree to which we can increase the injection rate to compensate
for a short injection duration or low contrast volume.
One option for estimating injection duration for a short scan is
to add a constant factor to the scan duration, for example, 10
seconds. Another ap-proach takes into consideration the patient's
physiology and the injection rate. If, for example, achieving a
peak aortic enhancement of 200 HU requires 50 mL of contrast
material to be injected at 5 mL/sec, the injection duration is 10
seconds. If the duration of the scan itself is 7 seconds and we
want to keep the enhancement above 200 HU throughout the scan, the
optimal injection duration for the scan would be 17 seconds. An
optimal injection duration for each examination can be determined
empirically or perhaps through pharmacokinetic modeling.
In peripheral run-off studies in a subject with normal cardiac
output, the transit of contrast medium or blood flow from the
abdominal aorta to the ankles typically takes 15 to 20 seconds. In
an ideal situation, the CT scan travels with or chases after the
contrast flow at the same speed and direction, so that imaging
coincides with peak enhancement throughout the course of the
vessels. Even with 16-slice scanners, however, peripheral run-off
studies typically take 20 to 30 seconds. It is therefore necessary
to increase injection duration above that required for the ideal
case, ie, the minimum required duration, by about 10 seconds.
Conversely, with faster CT scanners, scan speed may exceed contrast
flow rates and may have to be slowed to match the scan and the
propagation of the peak enhancement along the course of the
With multislice CT, it is critical to scan during peak contrast
enhancement. Three key factors in determining scan timing are
cardiac output, injection duration, and scan duration. When cardiac
output is reduced, the magnitude of peak enhancement increases
(Figure 6), but the times to both contrast arrival and the peak
enhancement are delayed.
The delay in time to peak enhancement is approximately the same as
that for contrast arrival, and therefore may be predicted by
monitoring contrast arrival with a bolus-tracking method.
In determining a scan delay that accounts for variations in
cardiac output, a common approach is to use a test bolus of
contrast material. A small test bolus is injected and repeat
sequential scans are performed at a fixed level to monitor contrast
enhancement. For CTA on a single- or 4-slice scanner, the time to
peak enhancement of the test bolus is estimated and used to
determine the scan delay.
This approach works well for single- and 4-slice scanners, but
during a 16-slice scan it may result in scanning too early and
completing the study before reaching the peak enhancement. Use of a
shorter injection duration with a 16-slice scan can reduce the time
to peak en-hancement and achieve peak enhancement in the middle of
the scan. However, the shortened injection duration may not provide
an adequate degree of enhancement during the scan.
We may, therefore, have to maintain the injection duration for a
16-slice scan at the level used for a single or 4-slice scan. In
this case, the scan delay for a 16-slice scan should be longer than
for a single or 4-slice scan, ie, estimated from the test-bolus
peak enhancement time, to ensure that scanning takes place during
the peak contrast enhancement. In an unpublished clinical CTA study
with a 16-slice CT scanner, we found that adding approximately 8
seconds to the time of peak test-bolus enhancement corresponded to
the optimal time to begin image acquisition following a full bolus
(KT Bae, unpublished data). Nonetheless, this approach is
empir-ical and arbitrary, and its appropriateness with other scan
protocols is unknown.
A better approach is to use a bolus-tracking technique, which
eliminates the need for a test bolus. Bolus tracking helps to
determine contrast arrival and is used to trigger the initiation of
scanning. Among the limitations of the bolus-tracking technique,
however, is the triggering of the scan by the detection of contrast
enhancement above an arbitrary enhancement threshold, which can
range from 30 to 150 HU, depending on the operator and CT
Another limitation of bolus tracking is a delay between the
detection of threshold enhancement and the actual initiation
of scanning. The delay varies from 2 to 9 seconds, depending on the
table position and CT scanner. One solution to compensate for a
long delay in image acquisition is to use a low enhancement
threshold. In this way, the scan is triggered before the desired
enhancement is reached, but image acquisition actually starts at
that enhancement level. With 16-slice scanners, short scan times
have reduced the impact of a delay between triggering and image
acquisition, making bolus tracking much more practical.
Table 1 illustrates the use of bolus tracking for determining
optimal scan timing during an examination for pulmonary embolus.
This example assumes a scan duration (T
) of about 10 seconds. Injection of 120 mL of contrast material at
4 mL/sec results in an injection duration (T
) of 30 seconds. The contrast arrival time (T
), defined by a threshold enhancement of 30 HU, would be about 5
seconds for a patient with a normal cardiac output and 20 seconds
for a hypothetical patient with low cardiac output. The time to
peak contrast enhancement would, therefore, be the sum of the
injection duration and the contrast arrival time, or 35 seconds for
a patient with normal cardiac output and 50 seconds for a patient
with low cardiac output. Scan timing can then be determined by the
This example demonstrates how a scan delay can be calculated
from the scan duration and injection duration, combined with the
contrast arrival time measured from the bolus tracking method, to
account for the variations of cardiac output.
The use of a saline flush immediately following contrast
injection has several potential advantages. It flushes out contrast
material that would otherwise be left behind in the injection
tubing. It eliminates the extra step of clearing the vascular
access site of residual contrast after injection. By pushing the
contrast bolus forward, it may create a more desirable bolus shape.
It increases the amount of contrast available for use in image
acquisition and may reduce artifact. Typically, a double-barrel
power injector is required for a saline flush.
In the future, optimization of contrast delivery may be
automated. We will simply swipe an identification card through a
card reader to input patient information, specify the organ to be
studied, and push a button. Automated protocols will take into
account pre-tailored input parameters such as the target organ's
ideal level of enhancement, preferred contrast bolus shape,
injection rate, concentration, and scan duration. By integrating
the patient, injection, and scan factors, the automated program
will calculate the optimal injection duration and determine the
optimal scan delay in conjunction with bolus tracking technology.
During the scan, the contrast injector and the CT scanner will
interact, automatically terminating the injection when enough
contrast has been delivered.
Effect of injection rate on arterial contrast enhancement.
Simulated aortic time-enhancement curves in which the volume of
contrast medium is held constant at 150 mL and the concentration
at 320 mgI/mL. As the rate of injection increases, the intensity
contrast enhancement increases and the duration of contrast
Dual-phase scan: (A) Fast (5 mL/sec) versus (B) slow (2 mL/sec)
injections. High injection rates are effective for dual-phase
scanning of the liver, as they magnify the difference in the
degree of enhancement during the arterial (A) and portal venous
(PV) phases, and increase the temporal separation between their
respective enhancement peaks.
Exponentially decelerated injection. Contrast medium administered
using the exponentially decelerated injection, as described by
the formula 4exp(-0.01t)mL/sec, results in uniform contrast
enhancement with less contrast material than a uniphasic
(A and B) Effect of high-concentration contrast material on
contrast enhancement. Simulated aortic and hepatic
time-enhancement curves depict the delivery of a constant iodine
mass in a 150-pound man (140 mL of 300 mgI/mL, 120 mL of 130
mgI/mL, and 105 mL of 400 mgI/mL). The use of high-concentration
contrast material is associated with earlier and greater peak
aortic enhancement. The effect of contrast concentration in the
liver (B) is not as pronounced as it is in the aorta (A).
Effect of injection duration on peak aortic enhancement: (A) in a
porcine experiment and (B) in a simulation. As injection duration
increases, peak enhancement is delayed and the enhancement curve
becomes more asymmetric. (Figure reprinted with permission from Bae
KT. Peak contrast enhancement in CT and MR angiography: When does
it occur and why? Pharmacokinetic study in a porcine model.
Copyright © 2003 Radiological Society of North America.)
. Effect of cardiac output on contrast enhancement.
Time-enhancement curves simulate the effects of decreasing
cardiac output (CO), while holding contrast volume constant at
120 mL and injection rate at 4 mL/sec. When cardiac output is
reduced, the magnitude of peak contrast enhancement increases,
and the times to contrast arrival and peak enhancement are
ELLIOT K. FISHMAN, MD:
Thank you very much for a very informative presentation. Does
anyone have any questions or comments?
W. DENNIS FOLEY, MD:
I'd like to ask a question. You initially referenced the physiology
of the body, particularly intravascular volume and cardiac output.
In reference to that, some investigators have proposed weight-based
dosing. Could you comment on weight-based dosing, and whether it is
an adequate replication of the patient's intravascular volume and
KYONGTAE T. BAE, MD, PhD:
I think so. Certainly, weight is the most important factor in
calculating blood volume or even cardiac output in physiology. To
determine adequate contrast volume, a weight-based algorithm would
be much better than the current fixed injection. You also need to
consider other factors. Going back to physiology, for estimating a
patient's blood volume, weight is a strong factor; height and
gender are also factors, but not as strong.
I have a follow-up question relating to the use of the bolus-
tracking technique, which you are basically marrying to a uniphasic
But you had mentioned the potential use of a biphasic and, then,
exponentially decelerated injection techniques. Is the use of a
biphasic or exponentially decelerated technique still going to
require a preliminary mini-bolus to model? Or can it be used with
the bolus-tracking technique?
I think bolus tracking can be used. If you are going to use an
exponentially decelerated injection, the en-hancement curve will go
up and then will stay uniform. Peak enhancement timing is not as
critical as it is for the single-phase injection, but you still
have to know when the contrast arrives. The bolus-tracking method
will help you determine that the contrast arrives at the right
You talked about a minimum volume of contrast. How do you define
your minimum volume typically?
That really depends on the patient's weight or size. Last year, at
meeting, Dr. Foley reported that he used 50 mL for CTA of the
As a rule, 50 to 60 mL is relatively minimum. But even with the
16-channel scanners, you can even go faster than a 10-second
acquisition. You can go down to a 5-second acquisition, to do a
total abdomen/pelvis. In a patient with marginal renal function,
you have two choices. You can stay at 50 to 60 mL, which is
probably a reasonable amount of contrast, or you can even go lower.
But then the question you raise is that with such a short duration
of acquisition, you may not be timing it adequately. But can you
compensate for that by, instead of injecting at 5 mL/sec, injecting
at 6 or 7 mL/sec? If you're very precise with your acquisition
interval, you can still capture a very short duration peak and use
a smaller amount of contrast.
Yes, I think that probably 50 to 60 mL can be minimum for a
normal-size patient, approximately 150 pounds, to just capture a
certain short segment. But once you scan durations longer than a
few seconds, you have to inject more contrast.
I've seen Dennis' work also, but that was really in select
instances where you had questions to answer in small areas. But if
I'm studying an abdominal aorta in a patient with reasonable renal
function, what would be the minimum contrast you would suggest? In
cases of normal BUN creatinine, are there some routine normals that
Again, it depends on patient size, contrast concentration, and
injection rate. While I haven't done a systematic study, in an
average-size (150-pound) patient, I would think 70 or 80 mL would
be sufficient, if you inject at 5 to 6 mL/sec. But I have little
practical experience with injecting low volumes.
I'd like to come back to minimum volume. You suggested that
recirculation plays a very important part in increasing aortic
enhancement. It requires about 7 seconds into the circulation
before you start to see recirculation, and that's fairly
Yes, it could be in some fast circulatory systems. But, in general,
it takes much longer than 7 seconds to see the circulation effect
In certain parts of the anatomy, in fact, you want to avoid that
recirculation effect, particularly in cerebral work. If you wanted
to image the kidneys before you saw the renal vein, which may or
may not be an issue, do you want to avoid recirculation, to avoid
that second hump of increase of aortic attenuation?
So in other words, for cerebral and for renal studies, instead
of trying to delay your acquisition, you may be better off scanning
in the early phase, after aortic arrival.
S. JAMES ZINREICH, MD:
I have a question: you mentioned saline. In which studies would you
apply this technique?
I cannot speak for the neuro studies, but for cardiac and all other
body parts, a saline flush will be very helpful. I think Chris
Becker has some slides on that, and he has more experience with
that than I do.
DAVID P. NAIDICH, MD:
Why would you not use it?
In the intracranial compartment, obviously, you have this 6-second
transit time, and that's what we want to administer the contrast
toward. So, in the head, especially the intracranial compartment,
it's a shortened time.
SANJAY SAINI, MD:
With the flush, though, you grab the tail of the injection;
otherwise the tail sort of sits there. Presumably, you use it in
We decided to use the dual injector also. Of course, we are doing
dual-phase imaging, because otherwise you are leaving 8 to 15 mL
behind that you are not using, between the line and the peripheral
vein. So you're right, there is really no downside for using the
I guess in the neuro applications, when you are doing carotids,
you are doing a single acquisition; for arterial, it's probably not
going to have the impact that it has on the body. So you can
potentially lower the volume, even in cranial applications. You
could lower the volume of the contrast, probably by 10% or 15%, by
making more efficient use of the contrast.
Let's say you have two individuals with the same cardiac output,
and one has twice the weight of the other individual. Due to the
same cardiac output, you would assume that the peak arrival
enhancement would be at the same time. But because individual B
weighs twice as much, we would want to deliver twice the contrast.
So it seems to me you would have to change the injection rate when
you give more volume, so that your timing is exactly the same.
Something I had not thought about until today was that it seems
that if you change the volume, assuming cardiac output is the same,
you have to adjust the injection rate.
I have not thought about putting the weight and the cardiac output
together, either. But cardiac output increases with a patient's
There have been several articles published, and our experience is
very similar, that for most patients, you can do preset delays. If
I have a 16-slice scanner and I'm doing routine aortic work, why
don't I just acquire it 25 seconds arterial, 55 to 60 seconds
venous, and not worry about everything. Then I'll go by Mike
McCarey--they looked at a range of patients with aortic aneurisms,
and it worked fine. We do 30 patients a day, for the most part,
using preset delays. In some select patients, you'd be more
careful. But is it possible that in the majority of patients you
don't need to do that?
Yes, the majority of patients have close to normal cardiac output,
so fixed delays would work. Some people like to use fixed delays,
others like to use optimized settings in each individual.
Certainly, from the throughput aspect, fixed delay will work fine.
But there will be some patients in whom you are going to miss the
For people first getting into CT angiography, I think that this
contrast delay always concerns them significantly. So I wonder if
we could have rules that say that for this group of patients, you
are going to do preset, but in this group of patients you need to
do an optimized delay. For instance, in coronary angiography, I
would agree that all those patients need some sort of monitoring to
determine when to acquire CTA. Or in older patients, in whom it's
But if you are doing something as simple as renal donors, or
patients under age 50, do you feel comfortable in the simple preset
delay? Is there some way we could come up with those rules?
I think it is a philosophical difference. Really, in every case we
do, we use bolus tracking to time it. But I know the other school
of thought is to use a fixed delay. Then there will be a few cases
for which you miss optimization of contrast. But in the majority of
cases, it will be fine with the fixed delay. I don't think that
there's a wrong and right answer. But that's the approach I would