ELLIOT K. FISHMAN, MD:
Now that we've heard each of the presentations and had a chance to
ask specific questions about them, I'd like to take some time to
discuss some overall themes. Within the presentations, we've had a
review of standard concentrations of contrast and their use in CTA
procedures. We've discussed relevant information on the selection
of a contrast agent for CTA procedures. But I'd like to get
everyone's advice for others just getting started in CT
DAVID P. NAIDICH, MD:
Well, the issue is that to really do good-quality CTA you need to
have attained a certain level of expertise. I think the
rate-limiting step for using this technology has been the
reluctanceperhaps on the part of manufacturers, perhaps on the
part of radiologists performing routine studiesto involve
themselves with learning to do this type of imaging. Actually, this
is consistently getting simpler to do. Unfortunately, despite
progress, there is still a gap between our ability to acquire
sophisticated data, and the ability to generate the images that
would be the impetus for including these as routine clinical
procedures. Elliot, you can speak to that better than I can.
When you look at Dennis' pictures, for example, it's clear that
an average community radiologist is not out there doing this kind
of imaging. So, the question is, why aren't they? It's not because
the technology isn't available. We have to determine what the
rate-limiting steps are. We need to determine what the sound
barrier is, if you will, that we need to get past to get people
doing these procedures routinely. That's really the heart and soul
of the issue of CT at this point.
Today, we've seen that, in the right hands, CTA is an
extraordinary tool in virtually every aspect of the body.
Are there any specific contrast-related issues, for example, in the
chest or in thoracic applications? Do you have any advice on
contrast selection that you could share?
Selection of contrast is an issue in terms of contrast utilization,
cost issues, and contrast reactions. When you start talking about
pushing 5, 6, and 7 mL/sec, you will have a lot of resistance from
a general radiology audience. In fact, there would be a lot of
resistance to even going much above 3 mL/sec. That is something
that we really need to make clear. Increasing injection rates is
not a danger to patients, and it is not necessarily going to relate
to renal toxicity. To date, I have not had enough experience with
the iso-osmolar contrast agents to render a definitive judgement
concerning their use. Reflecting on what I have heard here today,
the use of these agents could be extremely important for the
general radiologist to understand. Presently, to my knoweldge,
these are not being used for general applications, such as
CHRISTOPH R. BECKER, MD:
In our department, we have a guideline to give iso-osmolar contrast
media to any patient who is coming in with a creatinine level above
1.5 mg/dL. It's incredible, but once the clinicians picked up on
the low nephrotoxicity of iso-osmolar contrast agent, they are
demanding it, since they are so strictly aware of the
nephrotoxicity of any other contrast agent.
Do you have any other specific words of advice for someone starting
off with CTA or anyone doing a minimal amount of it?
I have already mentioned that more contrast is not always better in
CTA. In particular, in atherosclerotic disease, there is some need
to dedicate a kind of contrast application. But if you keep a
simple rule of thumb in mind--1 gI/sec--with this amount of
contrast you will achieve ideal enhancement of the arteries and
vessel wall. Higher flow rates may only be necessary in vessel
territories with no calcifications such as pulmonary or visceral
I want to say one more thing about contrast. If there is a
reluctance on the part of a lot of general radiologists to perform
CTA, based on their concern about the complexity of contrast
administrationchoosing appropriate timing and doses and so
forththat should not be a reason for not doing it. There are
relatively straightforward rules, especially now with bolus
tracking, which should relieve any fear of how complex this process
But there's no doubt that 16-slice CT has made it much easier to be
successful than it was with
4-slice or single-detector CT. With 16-slice CT, a lot of the
excuses that existed with 4-slice tend to fade away. For most
groups starting with CTA, there must be an advocate within the
group who takes responsibility for the process: setting the
protocols, contrast volumes, timing, and everything else. It's not
technologists deciding, Well, let's do this and that, unless
someone in the group is very proactive. But too often, if it's
someone else's job to do it, it just never happens.
Well, the extent to which knowledge about contrast administration
is an excuse for not using CTA is something that must be clarified.
Radiologists must understand that, while there are many potential
approaches to performing CTAs depending on clinical indications
(eg, peripheral runoff studies), nonetheless, the faster scanners
now allow a more uniform approach, especially with bolus tracking.
We need to disseminate the message that we should not hesitate to
perform CTA because of the potential complexity of contrast
MARILYN SIEGEL, MD
: That also sums it up in the pediatric population. Radiologists
need more education on how to do CTA in children. CTA works well in
a young population if one understands the basics of contrast
administration: namely volumes of contrast agent, timing of scan
initiation, the use of automated bolus tracking, and technical
factors specific for children, particularly the mAs and kVp.
Gaining acceptance of CTA in children means providing protocols and
training radiologists how to do it.
It seems to me that you see a reluctance, at least in the
literature, of those doing pediatric imaging to really get involved
in the 3D components and CTA. There seems to be very little in the
literature about it.
True, there is reluctance about performing CTA in children and, in
large part, this reflects the paucity of literature available on
the topic. It also reflects the concern about radiation. However,
radiologists need to recognize that the technique for CTA in
children is the same as in adults, but with minor modifications. If
you choose the correct parameters and understand the clinical
indications for CTA and 3D imaging in children, you get great
results and minimize radiation exposure. We need to provide this
information to the radiology public.
SANJAY SAINI, MD:
The main thing I worry about is that CT used to be goof-proof.
Everybody got the same scan and it worked every time. Now CT has
become more complicated. There are a few specific things that I
want to mention. First, on every abdominal CT study, whether it is
a routine study or a dual-phase study, we use bolus-detection
techniques, as I described. Once the technologists get used to it,
they will use it in every case.
Second, we have not done enough about modulating the contrast
dose with respect to patient weight/size or clinical application.
For example, when we scan the kidney for renal mass
characterization, we give less contrast than we do for a liver
lesion detection study. With the multidose delivery systems now
available, this will eliminate waste of contrast and provide cost
savings. Third, for CTA studies, we have lowered the tube-potential
to 120 to 100 kVp. At this point, however, this change is based on
anecdotal evidence rather than scientifically validated data.
Finally, we have become comfortable with the automated exposure
technique and use it on all abdominal studies with noise index of
15 to 20 HU. This can lower radiation dose by >30% and is
especially important in nononcologic patients.
KYONGTAE T. BAE, MD, PhD:
I would like to echo David's comments. It looks as though we are
still learning, and there is a time lag, perhaps 3 or 4 years,
before community radiologists practice and follow the guidelines
drafted and experimented with by leading academic institutions with
state-of-the-art technology. Dennis' protocol doesn't even include
single-slice CT, but there are many places that only have
single-slice CT scanners for CTA. Until we figure out and provide
some consistent guidelines, it is natural for community
radiologists to be reluctant to practice CTA. Perhaps, an automated
injection protocol and scan system may relieve some of the stress
of trying to figure out an optimal imaging protocol. I think it
will definitely help general radiologists accept CTA as a routine
You had a very detailed analysis of contrast administration and
choosing agents. Do you have any short words for people getting
started, especially on choosing agents?
In our institution, we use 350 mgI/mL, that's the highest
concentration contrast we have. For all CTA, we would like to use
the highest concentration we can get. For scan timing, we advocate
a bolus-tracking method instead of a test-bolus method. But then,
the next issues become: what enhancement triggering threshold value
do we need to select? What additional delay should we add for a
fast scan? We are still in the learning curve. I don't think we
totally understand the most consistent and effective way to perform
CTA. I hope, after this meeting, we'll come up with more conclusive
suggestions. But I expect there will be a lot of trial and error
until we can determine a definitive technique that everyone can use
with all scanners.
I want to make an important point. I don't think people need to
wait until all the answers are in on CTA, because by the time all
the answers are in, there will be another detector starting the
questions all over again. So, for all radiologists, if you are
going to do CTA, now is as great a time as any. Will things change?
Will we learn more? Of course. It's like a Don Henley quote, The
more I know, the less I understand. All the things I thought I
knew, I'm learning again. That's very true; it is a constant
learning process. But there is no sense putting it off. As we've
shown this morning, the capabilities are incredible.
S. JAMES ZINREICH, MD:
If you take a look at neuroradiology, it is a little different.
That is, it's not the technique or the protocol that really makes a
difference; performing an examination is done with relative ease.
The question is, are you going to use CTA versus MRA? You have to
evaluate the patient population that you are dealing with--for
example, the stroke patient. In this case, you are dealing with an
elderly person who had specific symptomology and who has already
had a CT examination. The patient is on that table now; you are
dealing with a critical time period. We want to make a diagnosis
within 3 or 6 hours. You can't afford to play around by going to MR
and angiography. You need to be able to make a diagnosis as soon as
possible. Therefore, doing a CTA examination really makes sense and
ought to be the very next step. So if I have a stroke patient on a
CT table, I'm contemplating or diagnosing stroke. Am I dealing with
vascular stenosis? The next step ought to be automaticI'm doing a
CTA examination, and whether the patient is renally stable or not
has to be taken into consideration. It's the same thing with
subarachnoid hemorrhage. You are dealing with a patient who isn't
healthy, who is not going to be able to cooperate with an MRI
examination and hold still for an extended period of time. You know
there will be problems. So, when I'm dealing with a CT, the next
step ought to be a CTA. If I'm having problems with the CTA, I
ought to go to the next steps, and contemplate whether I am going
to do a DSA or an MRA in order to improve my diagnostic capability.
When we're dealing with difficult, elderly, sick patients, in whom
CTA makes sense, it ought to be the second thing that you think of
immediately after performing a CT examination. You also deal with
another set of patients--head/neck patients. CT is very important
in that patient population. When you do CT in those patients,
contrast has to be given. Without contrast, that examination is
We've had a new development in our department. Recent literature
has shown that, in patients with early strokes (up to 3 hours of
onset), you can see the stroke on the source images of the CTA,
pretty much in the same fashion as we see it on diffusion-weighted
images. Indeed, we have pretty nice cases on this, in particular
during the night, we are doing this in many cases. So I think this
is a boost for CTA in these particular patients as well.
It's a kind of perfusion, but to be more accurate, it's
blood-volume imaging. It's really showing you the core infarction
area, even in the absence of any early stroke sign on the other
Right, estimating the original blood volume, absolutely.
The difference of enhanced density is in the range of 20 HU. It is
unbelievable to me that this is even coming up now with
multidetector CT, because we've had this kind of technique
available since the start of CT, 30 years ago. But, as we are now
using the high flow rates with multidetector CT, it becomes so
obvious that this enhancement is in the healthy tissue and we are
taking advantage of this for stroke assessment.
The topic at the moment is CTA, and I didn't want to get into the
perfusion CT studies. But, obviously, that is the very next step to
follow. Therefore, you would have the entire information about this
stroke patient, and, most importantly, you would have it very
I also want to make one point between the neuroimaging applications
and, eventually, the cardiac applications. Medical centers are
moving CTA to the ER, particularly in a place like NYU, where they
have the 16-slice scanner in the ER. In the past, typically the
worst scanner was in the ER, one of the older scanners. Now you may
actually need your best scanner in the ER, because so much of our
work really comes from that direction these days. I think that
roughly 60% of the admissions to Johns Hopkins come through the ER.
You can do a lot of the work before the patient comes in. But it is
a big source of most hospitals' patients.
W. DENNIS FOLEY, MD:
I want to pick up on what David said about how to get CTA accepted.
I agree that you need a champion within the radiology group.
Another thing that is probably going to happen is that the vascular
surgeons and other surgeons will come back to the radiology
department, and ask the radiologist to do what they have seen
presented at their vascular surgery meetings. So that will have
some impetus in increasing the use of CTA. In terms of IV access,
which is just a simple thing, we have a portable ultrasound machine
in CT that we use for getting IV access when the technologists
can't find it. Another simple thing about injection is that we
always have the patients put their faces up, with their hands up
against the face of the gantry. If the arm is back, you get an
inadequate injection in many cases.
About the 3D renderings, we do have the remote 3D lab, and I
know we're the champions of it. I agree that radiologists should be
much more involved in making 3D images. So we are going to force
ourselves to do it by putting a Windows workstation right next to
our PACS diagnostic workstations. Ultimately, the manufacturer will
make 3D data sets at the console, which they will basically send
out to radiologists, automated.
Dennis, who has the time to make all those 3D images? I know
Elliott takes 60 seconds to do it, but if you are reading 50 CT
studies a day....
Right now, the problem is how things are structured: you have PACS
systems, so we do PACS; you have 3D systems, and we can do 3D. It
is incredibly inefficient. If you have technology where everything
comes up on one system, we would be in the 3D world immediately.
Basically, everything we are doing now is volume acquisitions. So
you are reconstructing by ones and reconstructing by fives, and
sending this here and this there. But the reality is, if you just
get a thin section and send it to one place, you could do
everything from the axial, the multiplanar, and the 3D images
interactively, at a rate of 15 or 30 frames a second.
But is this really needed for diagnosis?
It depends on the situation. If I'm staging lymphomas and looking
for nodes, honestly, the multiplanar images may be nice to look at,
but the axial images will give me all the answers. But more and
more, our practice, I think, is like yours--we do so much oncology
But there is also the issue of our referring physicians. For
example, John Cameron is a surgeon at Hopkins who has done a
thousand Whipple procedures, and now he will not do a Whipple
without a CT angiogram. It becomes very cost-effective since there
are no other studies done, and the clinicians really rely on
One of the things I noticed is that no one ever questioned us
doing 3Ds per se; it was never an issue. Now it's gotten better,
faster, and easier. Now the first question clinicians have is, How
can I get one of those machines in my office? They really don't
want to see the axial images. Right now, we create a thousand
images, but who wants to see them? They just want to see the images
that show their pathology and 3D does that incredibly well.
But if that is the case, then the 3D images are really used to
capture the information, not make the diagnosis.
Well, it's both. We have a paper coming out in
about this. I read 3D images for all pancreatic patients before
they went in for potential surgery. Typically, if you do imaging of
the pancreas, it's approximately 70% accurate in determining who is
resectable. We read them all preoperatively, and then they went to
surgery. We found then that the preoperative 3D imaging was roughly
93% accurate. So there was a 23% increase in preoperative imaging
diagnosis with 3D imaging. That's a significant increase in
patients going to surgery.
One of the issues has been that the quality of 3D imaging for
the last 15 years has been variable in its implementation. There
was an article on pancreatic cancer in
a couple years ago from a group at Duke. They looked at axial
images and then filmed 12 or 16 3D images. They found that the 3D
images didn't add anything and they couldn't read them as well. But
you have to look at what they did: they were shaded-surface images.
The data sets were 3 years old, so they were all from
single-detector scanners and the radiologists had <1 month of
experience. So the conclusion was that if you don't know what
you're doing, you get bad results.
In neuroradiology, 3D imaging is critical; it's like the essay.
That is the only way the referring physicians will accept that
particular data, because that is the only data that they can truly
feel familiar with. We also have to realize that we have to do 3D
images, because if we don't do them, somebody else will. So we'd
better learn to do it, and do it fast.
I think one of the problems with asking whether CTA is clinically
effective or not is that we're looking at this issue too closely.
In fact, we will need to depend on the whole armamentarium of
imaging tools to assess these huge volumes. CTA is only going to be
one component of a lot of approaches that will be intermixed. So it
may be that an individual case needs to have CTA performed; perhaps
it requires volume rendering; or perhaps a CAD application will
need to be utilized initially. Imagine an environment, ultimately,
in which you will have access to all these different tools
simultaneously. CTA will only be one component of the whole
approach of how to investigate data sets that are up to a 1000
images now, or even potentially much greater. Even though this
forum is focused on CTA, I don't think we should look at it as a
totally separate phenomenon. It should be part of a more global
approach to how we will think about data management and data
interpretation. It's just one tool amongst others. Ultimately, it
will have to be integrated into a PACS. If all necessary tools are
not somehow in one place at one time, we will never break the CTA
I believe that at the 2003 RSNA meeting, both GE and Siemens will
have 3D on the PACS. Basically, it's impossible to have all these
separate imaging systems: 1) we can't afford them; 2) even if you
can afford them, you don't have enough desk space; and 3) even if
you had enough desk space, you don't have the time to keep going
back and forth. It's kind of silly; it's poor management. Part of
the problem is that we've done CT for 25 years, and we always did
it in a slice-based environment. If CT started at 16-slice scanners
and the rest didn't exist, and you started today, there never would
be a question. No one would ever think of looking at axial images,
that would be the dumbest idea you ever heard.
That's akin to the idea that if someone initially started with CT
as the first imaging modality, and only subsequently developed a
chest X-ray, the chest X-ray would be considered an improved
solution to the problem of a global assessment of lung disease.
That also applies to the fact that if PACS was always there, and
someone invented film, it would be incredible.
There is no better time than now to do CTA. Take some time and
make some effort to plan, in choosing the contrast agent and
concentration, and the timing. Take the time to actually learn how
to do the tests, the bolus tracking, and the postprocessing. You
have to learn each of those steps; if any of them fails, you are
not going to be very successful. But I think for good patient care,
it's not an option.
This is the important thing that people may forget: as
radiologists, we always promise that we control radiology, because
we can deliver the best care. But I think if you are not doing CTA,
you are not doing the best care. At the SIR meeting last year, the
president of SIR noted in his address that 80% to 90% of all
diagnostic angiograms are history because of CTA and MRA. He said,
If you're still doing it, just get over it. You can't be doing it.
And if you are doing it, you are doing bad patient care.
You've asked about selling CTA to the clinicians, and you made a
point about CTA and pancreatic cancer. I think one of the important
things is to prove that it works. You've shown how it does today,
but the first time we've heard anything about numbers is this 70%
If you look at the proceedings today, we've had 4 hours of
discussion, primarily technical. Some very critical points were
made. But the next point, the bread and butter for the general
radiologist, is going to be: now how am I going to do this in
stroke? How am I going to do this in aneurysms? How am I going to
do this in the tibial artery? How ever am I going to approach it?
What am I going to do with the liver? They need specific points on
how to do it.
And when to do it.
We talked about CTA today, and we didn't address the kidney or the
pancreas. There are other topics we obviously didn't do. One thing
that people will be interested in, after they get the core
information we presented today, is: How can they do it in practice?
How exactly do you do it? How do you get trained on it? People sit
at lectures and hear us all speak about CTA. Everyone will listen
to it, but then they'll go home and do the same thing they did
before, as always. I think there is a gap. People are now able to
say CTA is really good, and those images are really cool. But they
will have 12 excuses why they can't do it themselves. They're too
busy, they don't have time to learn, they don't know where to
learn, they haven't been trained, etc.
I always ask a question of audiences: how many people feel they
are comfortable with their workstation and with using it. Nearly
always, 1 person in 100 raises their hand, so 99 are uncomfortable.
So, can we tell people how to go from this point? We can provide
protocols, but that doesn't make anyone do the process. So maybe we
need to address workflow; how people can really do it. I think that
would be a very important point to people: how to manage workflow
But, we have avoided another issue, another crucial step toward
gaining acceptance of CTA. At some point, we must have the data
that shows that CTA is really clinically effective. To do this will
require that radiogists relearn both basic anatomy and newer
clinical concepts relating to vascular disease. If you went into
the general radiographic community and asked them to interpret some
of the images that we've seen today, they would be somewhat at a
loss. One of the great things about CT when it was first introduced
was the decreased dependance on learning angiographic anatomy. Now,
suddenly, you have to relearn anatomy again! One of the most
important steps toward general acceptance of CTA will be the need
to retrain radiologists about radiology.
I think that is a very good point. With all the vascular studies we
are doing, many of us were not really trained in angiography. Now,
they train radiologists in angiography early; I do the lectures and
have kept that angiographic focus. But general radiologists may be
uncomfortable discussing the issues related to aortic surgery or
aortic intervention. But they can relate to the image, and discuss
that one-on-one with the vascular surgeons. His angiographer would
not have a problem doing it, but his angiographer is not doing the
diagnostic angiography nowthe general radiologist is. So that's
where the retraining issue is.
So, if you were going to do specific topics for another meeting, it
might a good idea to have a clinician there. What does the vascular
surgeon want to know from CTA? Even things like vessels, as David
mentioned. I haven't looked at vessels beneath the knee. I know
there's a bunch of vessels that go there, but who knows what they
are? Our RSNA exhibits this year are on vascular anatomy. We have a
medical artist drawing the pictures, and it's interactive, and I
think that might not be a bad thing. When we were all residents
1000 years ago, Kodak used to give out materials showing all those
vessels. They were very popular, and I still have them.
I still have them, too.
That was the only material you ever had like that. Things like that
might not be very hard to reproduce for current radiologists and
Well, I also think the manufacturers of contrast agents should be
very much involved in popularizing and teaching these issues.
I want to bring up one point we haven't discussed yet. We realize
that money makes the world go around. So, one thing radiologists
might think about is how they get paid for the CTA studies. If you
bill a CTA examination as an abdomen/pelvic CT, then not only is it
a noncompliant code, but it is less reimbursement on the
professional side. So, radiologists also need some education
regarding how to appropriately bill for these studies.
Radiology is a very big job, including getting reimbursements for
CTA. Because some people didn't properly organize, they did it
really poorly. Gordon Harris, PhD, at Mass General went to HCFA and
everything else, and we got reimbursement straightened out a bit
better; but it's still lousy, I think. The trick is, you've got to
know how to bill, so you can get paid well. It often doesn't pay to
bill the CTA codes--if you use an abdomen and a postprocessing
code, you do a lot better. But that's probably not an unreasonable
thing. This is all part of the workflow issue. How do you build CTA
into your practice and make it successful?
Contrast media injectors and CTA equipment don't even talk to each
other. I now understand, from the GE side, that they are beginning
to think about integration. I suspect it is the same with the other
companies. From the point-of-view of the technologist who is
implementing the procedure, that would make it easier for them.
I think both GE and Siemens have worked on that. The issue that
always slows it down is the liability issue; it's not a technical
issue, quite frankly, it's purely liability. If they want to
implement it, it's pretty easy to implement within reason. It's
just that it's been that way for 10 years already.
I'd like to thank everybody for your excellent participation. I
can say that I probably learned a lot more than I contributed. For
the general population who will read these presentations and
discussions, I think the one conclusion we can make is that, for
all of us who are doing a lot of CTA, the process is still
continually growing and changing. We can also thank Amersham Health
for putting this meeting together so we could have this discussion.
I believe we'll have some good materials and really important
information for the radiology public.