Applied Radiology

Moderator: Elliott K. Fishman, MD

ELLIOT K. FISHMAN, MD: To help us to wrap up all of the presentations and discussions today, I will ask everyone a couple of questions and then we will have some time for general discussion.

Ella, in terms of CTA, the progress of CT technology has been incredible-from 4-slice to 16- to 64-slice. The software is still developing. What do you see as the biggest challenges that remain for us to do things more efficiently and better-whether the applications are pulmonary, cardiac, or kidney? What are the real challenges that you see in your practice?

ELLA A. KAZEROONI, MD: I think one of the biggest challenges is not the image quality; the quality gets better and better. The scanners work more robustly and more consistently as we increase our detector rows.

The biggest problems are the efficiency and the workflow. How do you handle these data sets? How do you look at them efficiently when you're confronted with 4000 images? You can't spend an hour reading every CTA case. It's not efficient and it's not productive. You would probably lose money and not be able to support yourself spending an hour a case. You'd be there 24/7. We need even more efficient viewing tools and different ways of looking at the data sets, forgetting the old-fashioned axial scroll-through data. We need tools that are very efficient that can drive you very quickly to the data sets, as opposed to you trying to play with the toys, trying to get it to look just right. Efficiency is a major problem.

GEOFFREY D. RUBIN, MD: There are two things I would comment on, and one follows what Ella was saying. I look at 3D workstations as having evolved into a state where we have pretty good stem cells, but they now need to differentiate. In other words, we need to see dedicated software applications that are specific to clinical problems. For example, I showed the runoff study and a tool that specifically lays a ruler along the artery. That wouldn't be useful for other types of studies, but to the extent that we can get to the point where the workstation automatically segments the anatomy of interest and lays clinically relevant quantitative data next to it, that would be a great enabler. We need to move toward developing all of those.

With respect to the heart, we still need more speed. I'm not talking about more rows, but better temporal resolution. Our dependence upon pharmacologic intervention would diminish if we had faster rotation. Actually, I am still surprised that occasionally we get motion-related artifacts on patients whose heart rates are in the 40s. So, while 333-msec rotation is a tremendous engineering and technical achievement, I hope it can get better; that will really improve those studies.

FISHMAN: What do you recommend, multiple X-ray tubes?

RUBIN: That's one way to go about it.

KAREN M. HORTON, MD: Well, in addition to being software-specific, it's a big challenge to the radiologist because you can't be an expert in the coronary arteries, and be doing CT colonography, and be doing the runoff studies. It's too much, and it's a lot of training, and these are very specific studies. In academic centers you can do that, and we have specialized people who do certain things. For the private-practice radiologist, they're going to have to do a similar thing: Have some people who are going to specialize in the heart, the colon, or in doing runoff studies--otherwise, it's impossible.

STANLEY GOLDFARB, MD: From my perspective, the challenge is communicating these possibilities to us as internists. It is a big challenge because your technology moves ahead and your abilities move ahead. When someone comes into my office with chest pain, I have to decide which technology to use, what's going to be the best outcome of this versus another test, which people are going to be able to manage the patient well, and determine how well you have integrated your activities with the surgeons and with the cardiologists.

JULIE FIELDING, MD: I'm moving toward the minimalist route. I'm really worried about the radiation dose to the patient. Every time we get a new scanner and a better toy, the dose goes up, right? Because we run it twice and we have to be extremely cautious about that. Sometimes I have very beautiful images, but I don't know if I need them to be that beautiful. Maybe I need them to be diagnostic and I don't need them to be perfect. So even though it's very attractive to me to have beautiful images, there are going to be cases in which we might be able to save time and energy by taking a more minimalist approach, and then we can put our money where our mouth is on the stuff that really takes the time. An abscess search doesn't need all that stuff. You can probably do it by 5s and take a look at it quickly. Then I would probably diminish some of my protocols, actually, make them simpler and save time and use the physician effort and the technologist effort on the tough protocols.

FISHMAN: Well, that's true. We were commenting before that we do 30 CTA studies a day, but we also do 200+ other patients. So the reality is that we're still talking about a small segment of what we do. But, in that segment, technique and technology is everything because if you ever entertain the idea that CT can replace diagnostic cath, it needs to be perfect.

FIELDING: Right. But that's why I want to spend my time making it perfect. Frankly, I think that you can do pancreatic cancer without spending as much time on it as you would a coronary case.

FISHMAN: I would definitely agree there, because the two worst things in the world now are coronaries and colons.

But the audience should note that I don't think doing a very detailed study means you have to spend a lot of time. For example, since you mentioned pancreas, we do dual-faced pancreas studies. The whole study takes 25 seconds between both phases. We process the data, in real time, in a couple of minutes, and we have 95% accuracy with what you find in surgery. To me, the worst case in the world is that I provide the surgeon with the best information I can, but then the pancreatic cancer patient has worthless surgery because they find disease that was not resectable. We're not going to be perfect in that domain, but we're 95% or so perfect, at least according to a paper published from Hopkins. So, I agree with you that we're all time constrained. But, particularly in these CT applications, the biggest error is when people don't put the time in. I think it's super-critical that if you're going to do it, do it right.

Maybe not everybody should do coronary imaging. If you're not willing to put the time and effort in to learn everything, don't do it. The worst thing would be to do it badly.

Karen mentioned imaging colons. The colon is a painful study, and I don't mean for the patient, I mean for the radiologist. You're looking at thousands of images looking for an 8-mm polyp. So, virtual colonoscopy is not for everyone, because unless you're willing to look at what's most likely a negative study in the screening population, don't do it. Let the patient go for a colonoscopy.

This is like the difference between the article published in The New England Journal of Medicine by Perry Pinkhardt and the article published in JAMA . Perry dedicated himself to prepping the patient, reading the study, doing the study, and others are not as detailed. So one is 80% accurate, and one is 50% accurate. There was a good article in The Wall Street Journal that asked how is it possible that two major journals have totally different results published roughly a month apart. That was exactly the reason-the dedication to how the patient was prepped and how the study was done.

I hope one message that we send to radiologists is that if you want to do this, there's only one way to do it, and that's to do it well. If you're not going to do it well, don't do it; don't do it halfway. Radiology has a franchise of imaging. I hate to sound as if I'm trying to fight a fight here, but if you feel that you earned the franchise, you have to do it as well as possible. If you're not going to do that, then you're going to lose the franchise. That's a danger radiology has now. Unless we can take these tools that have been developed to do great studies, the person who could do it best should do it. As you said, there's no reason why the radiologist is doing it and not somebody else. The equipment is owned by the hospital.

RUBIN: If you don't do it, someone else will. I agree with you 100%. You need to do it as well as possible. I think that any radiologist who makes a conscious decision not to provide a service because it's too complicated or too hard, when it has been accepted into the mainstream and is being used by a critical mass of radiologists, should really evaluate what they're signing up for. Basically, they're saying, "I can't handle it. I can't do it." Then someone else should.

KAZEROONI: We've shown that coronary CTA is fairly challenging. You've got to learn the anatomy and have to understand the cardiol-ogist's lingo. But then you need to think about whether or not you can provide 24/7 coverage for coronary CT angiography. If chest pain centers start to ask for coronary CTA around the clock, and they're going to use the results to decide whether the patient's going to get sent home from the chest pain center or go to the cath lab, you can't just provide the service from 8 to 5. So you have to find a way to deliver this very challenging service throughout the night as well. As our first of several VCTs are coming on line, that's one of the things I'm facing as a very real challenge: How to provide coronary CTA 24/7.

RUBIN: We faced that when the pulmonary CTAs came through and it didn't take very long before the residents were basically doing it on their own. But I'm concerned that that transition will not be as simple with coronaries. What sorts of things are you thinking about in terms of being able to provide that?

KAZEROONI: We now have more thoracic radiologists than we've ever had before. But not all of the thoracic radiologists at my center do coronary CTA. We've kept it to 4 or 5 of us, to make sure that everyone can do it consistently, with accuracy and reproducibly. Moving from 5 of us doing it to 15 thoracic radiologists doing it, and being willing to do it at 3 o'clock in morning, would really change the whole dynamic of an academic radiology division.

FISHMAN: We're starting to do it from the ER, also. It is a massive challenge. You're not going to have residents reading it, not to mention the fact that our residents, like your residents, cannot do anything more because nights have become as busy as days.

Nighthawk radiology is very common now; but that is not going to work for coronary imaging. They work for looking at abscesses and giving preliminary reports. That is not going to be the way cardiac CT is practiced. So, it is a challenge, I agree. I don't have great answers because even if you have enough people, they do not want to be there or be woken up at 3 o'clock in the morning.

KAZEROONI: Putting a powerful workstation in their homes is a very expensive option. Even if you did that at their homes, they don't want to be up. The thing about the PE studies is that you can withhold anticoagulation for 6 or 8 hours, or you find out that you gave 6 or 8 hours of anticoagulation you didn't need to when the faculty reads the study the next day. That's one thing, but to send a patient home from the chest pain center who is really having an MI is a much bigger problem.

RUBIN: There are two aspects to doing the study. One is having somebody in there to acquire good data, and the other one is reading it. Maybe you can dissociate the two. You can probably train fellows to manage the acquisition side. Obviously, in private practice, they're not going to be available. But there may be a dedicated nighthawk model for reading coronary CTAs. If you could get a consortium of several major centers that are in the same time zone, then you basically put 1 person on call, once a week or once every 14 days, to cover all the reads if everybody feels confident.

FISHMAN: You need the people in the other time zones. Because in the same time zone, what good do you do?

I'd like comments on where you see this going in the future. Where do you see CTA going in terms of workstations and applications? What changes do you see looking forward over the next couple of years? Application-wise, where do you see the progress going?

KAZEROONI: Right now we're talking about the 64-slice scanner being the cutting edge, with newly evolving scanners that are rolling out. But what comes after the 64-slice? How will that help us do anything differently than what we think we're about to embark on? Yes, we're getting reproducible coronary CTA as we get into the 64-slice scanners, but we haven't really gotten into the idea of robust plaque characterization and quantification. It may take the next generation of scanners or two beyond 64, whether it's more detector rows or smaller detector rows combined with faster gantry rotations, to get the resolution we need to do accurate plaque quantification and characterization. That's something I'd like to see happen in coronary CTA, but I think it's going to take another evolution, or more, to get to.

FISHMAN: If you could do coronary CTA at night, how many studies would you do at night if it were available?

KAZEROONI: I imagine it could be between 5 and 10 a night.

RUBIN: In our center, it would probably be 4 to 7 or so.

FISHMAN: That was kind of the PE model. At Hopkins, for example, we know that roughly 10 patients a night are kept over to get nuclear scans in the morning: 92% of them end up being negative. It's like PE studies; they used to do those with nuclear studies and there were 3 to 5 a night. As soon as we did CT, it became 15 to 20 a night.

RUBIN: It depends on the size of your ED.

One area that I'd really like to see evolve is perfusion imaging and, specifically, looking at tissue perfusion in a spectrum of diseases; one would be arterial occlusive disease, and trying to ascertain end organ impact. It's already being done in the brain, of course. The other is oncology and looking at the impact of drugs targeting angiogenesis. The fundamental difference is while we've gone from 4- to 16- to 64-row scanners, the basic size of our detector collector is about the same: It's 2 cm for most; there's one 4-cm detector out there. If you want to do perfusion imaging, we need to see wider detectors so that if we wanted to look at the heart and look at myocardial perfusion, for example, you'd like to be able to encompass the heart in one set of rotations.

So, I think, the big technologic challenge, in addition to temporal resolution I mentioned before, is being able to widen the detector so that we have a bigger swath. While it sounds easy to just stack up a bunch of detectors, the challenge is, of course, cone angle and having the algorithms to deal with it because out at the far ends of the cone, there's much more dissociation as you work your way around the gantry. So it's not trivial, but it would certainly really open up CT to a whole new range of assessments.

HORTON: In addition to more technologic advances, it has to be incorporated into your daily practice. It is like when you're going to look at every case in 3D, not just certain cases. When you start to do that, you find a lot of things that you were missing. You don't know what you're missing when you're only looking at the axials, so I don't think people will be requesting 3D CT; that's just how you're going to look at the information. You will see a lot more in patients that you weren't expecting to see.

FISHMAN: Basically, you're saying that the whole workflow is really changing.

HORTON: Right. Yeah. You'll be reading 3D primarily.

FISHMAN: Stan, you commented on this before. What do you think the radiologists need to do to work with the other physicians in terms of bringing them up to speed on what we can do? What is it that we should be doing that we're not doing?

GOLDFARB: I support the recommendation about the medical grand rounds coming to individual conferences. In our hospital, for example, our radiologist comes to the renal conference on a regular basis and talks about how to handle vascular access. That's enormously helpful, and it has generated a very collegial sense about this difficult issue between the surgeons, the radiologists, and the nephrologists. So reaching out, particularly to the medical specialties in these conferences, would be very useful. I also agree with this issue about getting functional data; it's a little bit beyond perfusion data. If you can correlate these fabulous images with some sense of metabolism, that linkage would be spectacular.

FIELDING: It's a workflow thing, but I have to teach entirely differently. It's a 24/7 environment, and if you want to bring new radiologists up to speed, you have to think about how to teach them in a volume atmosphere, and you have to teach them appropriate communication skills. We can no longer sit and go over individual anatomic pieces quite as well as we used to anymore. So, it's rethinking the entire workflow. I'm looking at 24-hour-a-day coverage very soon, and that's going to be a challenge for everybody to get used to.

FISHMAN: It's been terrific for the past 8 hours looking at the challenges of imaging. It's definitely the best of times and the worst of times. Technology has never been better and the opportunities have never been better. But, also the challenges have never been greater.

We're all busier. We have higher volumes, more demands. Yet, to develop it to the next level really takes close interaction. Also, it's very clear from everyone's comments that training and education is paramount; how we train residents and fellows, and how we train our associates is all very critical. I hope the material produced from our discussion will help our colleagues get some feel for where things are and help them learn new applications.

I want to thank GE Healthcare for sponsoring this forum and the resulting supplement, and thank Applied Radiology for putting it all together. This was our fourth meeting on CTA applications and I think it's a very good opportunity for all of us to get together. I know it's a great learning experience for me. I hope the material from our discussion will be useful to all of the readers. Thank you all very much.