STUART MIRVIS, MD, FACR: In this closing discussion, I’d like to bring out some of the technology issues that were discussed. And then I’d like to discuss some of the medical-legal issues that have been presented.
CYNTHIA McCOLLOUGH, PHD: One of the most important technology points to make concerns automatic exposure control (AEC).Some of the technological implementations of AEC don’t treat small and obese patients quite the same. There needs to be better consistency. I know marketing people like to come up with names for this technology, but we need to be clear about nomenclature and use precise names like Z modulation, patient-size adaptation, ECG modulation, etc. Technologists will know what is affecting the exam and what isn’t affecting it across all manufacturers’ equipment.
MARK OLZEWSKI, PhD: This would be a great opportunity for vendors to collaborate with a group like the Medical Imaging and Technology Alliance (MITA) to standardize the terminology. In many centers where you have CT scanners from multiple vendors, the technologist would be able to translate smoothly from one to the other.
DR. MIRVIS: I would make an appeal to the vendors that when they do commercials on television, that they mention the radiation exposure and radiation reduction technology. We are trying to educate the public and this seems like an appropriate forum to communicate that this field is aware of the issue and that we are addressing it at all levels. It seems to me that we should be emphasizing that point. Since vendors are working so diligently on the engineering, we should tell people about that.
DR. McCOLLOUGH: It would be nice if we could do that in a relative way.We shouldn’t be saying that we have reduced the dose to 6 mSv or 5 mSv.Those are arbitrary calculations.
DOMINIK FLEISCHMAN, MD: But what else could you say? Maybe quoting a dose in mSv is not the correct way but neither is saying that we have achieved an 80% dose reduction. That would be 80% compared to what?
DR. MIRVIS: The message, should be that we are doing everything we can, in designing our equipment, to reduce X-ray exposure and maintain great image quality. It should be something that everybody can understand. Once you start saying anything about mSv you will lose the audience.
DR. McCOLLOUGH: An effective message might be that the technology recognizes the size of the patient and adjusts for large or small individuals. Then people will know that we are right-sizing the pediatric dose. Another thing that has been advocated is to put the effective dose in the DICOM header. But some organizations are pushing back on this. It seems to me like MITA is the only place where you can broker an agreement about something like that.
DR. FLEISCHMAN: I think the desire to put the effective dose into the DICOM header or on the user interface would lead us into trouble. How would we derive this number and what number should we ultimately use? It would be more reasonable to separate scanner-specific parameters, which could be used in the user interface or as documentation in the DICOM header, from patient-specific parameters. I’m not in favor of that idea of giving an effective dose, which would then be easily misused as the specific dose to that particular patient. At the moment, we cannot calculate this value.
DR. MIRVIS: As we discussed today, there are various parameters that can change that inability to compare doses across vendors. But if two people use two different K values, if two vendors use two different methods of calculation, then we have obliterated the ability to make side-by-side comparisons. It seems that the weakness of CTDI measures is the strength, because they are the one thing that is relatively consistent. Those measures could be our apples-to-apples comparison, even if it’s more like an apples-to-asterisk comparison.
RICHARD MATHER, PhD: Even if we did obliterate that ability to make comparisons, that doesn’t mean that people won’t use it that way.And you can guarantee they will. The CTDI has already been coerced from a scanner output into a patient dose, into a risk scenario, which it doesn’t have enough strength to do.
DR. McCOLLOUGH: We are the people giving the lectures at meetings and putting out the brochures that continue to promulgate this. And so there needs to be some pushback from within our community to say that this is wrong.
JAMES BRENNER, MD, JD: But if you don’t translate it into usable information (namely risk) then of what use is that information? Simply relying on machine parameters is not wrong, but it has such a limited use in terms of goals. The goals here are to have a sense of what the risk is for patients. I understand that it is not easily translatable. But simply putting the scan parameters in the DICOM header is like putting the parameters of a magnetic resonance imaging (MRI) scan on the DICOM header. It’s very interesting, but how is it going to be used?
DR. MATHER: This goes back to what Cynthia said earlier. We need 2 different parameters. We need one that can tell us what the scanner did, and then we need something else that can address the risk. Whether that information belongs on the scanner or off the scanner is a whole other topic. But we have to separate those, and they haven’t been separated yet.
DR. McCOLLOUGH: And to the point about what is reasonable: If I can show the diagnostic reference level (DRL) in an easily measurable, verifiable number, for 5 different countries, and then show how my facility performed lower on an average patient then I think I have made a reasonable demonstration of dose-reducing efforts.
Compare that with trying to figure out the effective dose for a CT scan.That calculation will always be a moving target. Since the image quality has a hand in effective dose, it would theoretically fall where it needs to be if the CT community is scanning at a reasonable limit asset by the diagnostic reference values. Then we need to determine how often we establish DRLs and if we look at them globally or nationally. They can be published multiple times a year and they become a good way to track technology. When the American College of Radiology CT Accreditation Program first launched, our DRLs were very high because they were based on values from the European Union that were based on scanner technology from the 1990s. Now we use data from 2000 and later, and we see that the DRL for a multislice head CT has gone up but so has the image quality.
THOMAS FLOHR, PHD: As long as we create images using X-rays, we try to generate the images with as little photon energy as possible. Over the past few decades, we have come very far in that quest. We are using this method very responsibly. As long as the case remains that there is no other modality that gives us this type of diagnostic information without X-rays, this will remain a reasonable tool to use. All of the CT vendors strive to optimize their equipment. We have more sensitive detectors, we have new reconstruction algorithms and all of the engineering models that are built into the systems are quite remarkable.
The risk is more difficult to estimate. One potential way we can minimize risk is by having an automated exposure control that is similar to what is used on a digital camera. It’s not fair to blame everything about X-ray exposure on technologists when many physicists and radiologists do not themselves understand all the intricacies of CT scanners. Some of the calculations are simply too complicated. When they become exponential functions, it makes the task a laborious effort. I think this process needs to be automated by computers. They’re much better than our brains. Most vendors are moving toward solutions that automatically optimize the mA and the kV.
The technology is evolving rapidly. The first generation of any technology is typically not 100% optimized and I think we could save a lot of work if we had some kind of common platform that allows us to share protocols. This would help us take advantage of clinical expertise in very specific settings and having a common database for protocols would allow us to tailor protocols on new machines based off of existing and optimized technology.
DR. McCOLLOUGH: The manufacturer-developed protocols that come with the scanner are probably the biggest determinant of doses that sites use. I’m a firm believer in looking at the end of the chain. In CT it all ends with image quality. Compare this with computed radiography where there are very strict definitions of image quality. So if a vendor says we can save 10% of dose with one tool and 5% more with another tool, there is a concrete measure of image quality that can be applied to the end product.
When people are purchasing CT scanners, it can be hard to make an apples-to-apples comparison, because any vendor claim at dose reduction is still liable to change in clinical use. There is an International Commission on Radiation Protection (ICRP) Task Group on Image Quality and Dose that is proposing to measure factors like noise and power spectrum at certain doses. It would make CT more transparent and we liken it to how automobile manufacturers estimate miles per gallon.
DR. FLEISCHMAN: I think education is needed on a much simpler level. If you have an institution that has scanners from, let’s say,two different vendors, and they have scanner A and get a new scanner B from a different vendor; they take the protocols from scanner A and directly transfer them to scanner B. They use the same mAs, because theoretically, the same mAs should yield the same image quality. But this is not true, and different scanners produce different dose at the same mAs setting.
DR. MATHER: Switching gears, I think Dr. Fleischmann and Dr. McCollough brought up a very important point in terms of image quality. It’s always possible to lower the dose, but really that sort of floor of what we need for image quality is as low as you can go. So I think that is a great point that we always need to consider in this.
CHARLES WHITE, MD: Technically speaking, one of the best ways to lower dose is obviously to use prospective gating. But one of the limitations in doing that is heart rate. Are there limitations to prospective gating when you image children, young adults and adolescents, where the heart rate may be higher from the beginning? Are there any low-dose prospective approaches that can accommodate heart rates above 65 or 70 beats per minute (BPM)?
DR. FLEISCHMANN: This question pertains to the temporal resolution and the rotation time of the scanner. So I think it probably will be possible in the future to use prospective gating, even for higher heart rates. Our experience is that if the heart rate is above 60 or 65 BPM,the prospective gating is not that reliable. If the heart rate is low then you have a number of options to reduce the radiation dose. If the heart rate is high and irregular, then you do not have those options and the dose ends up relatively high. Interestingly, our own range of doses varies fairly widely. This is not necessarily because we are not using the best techniques but because those techniques do not work on every patient.
So, if you have a 25-year-old woman with Marfan’s, and you know you can’t use prospective gating, because you want to see the entire cardiac cycle and the aortic valve in detail, then you need to be aware that you will deliver a relatively high dose, and you should discuss it with your clinicians. And, ultimately, the choice is that the risk of not doing the study, and not being able to plan a valve-sparing surgery adequately, is higher than increasing the risk of breast cancer by a few percent.
DR. MIRVIS: Don, if you’re talking to a group of radiologists, mixed academics, and radiologists in general, what do you think is the most important thing that they should focus on?
DONALD FRUSH, MD: You need to start by understanding your equipment and what the indices are for that equipment. You should be familiar with what the various modifications do, in terms of image quality. You must then optimize dose by working with published protocols, vendor applications people and your own medical physicists in order to do the best possible job of dose optimization. You should be cognizant of different considerations concerning image quality and resulting dose, while generating images that you as a radiologist, your radiology group, and your clinical referrers are comfortable with. There is not going to be one standard. There’s going to be a range that has to be more clearly defined. And, within that range, you would have the flexibility to make adjustments, based on the size of the child, the indication, the region scanned, and so on.
DR. MIRVIS: We’re not going to box ourselves in too terribly tightly?
DR. FRUSH: Right. But you do have to understand what you’re using, and know how best to optimize that. The longer I’ve worked with CT, the more I appreciate the engineers and the medical physicists, in terms of helping with that. I think that, speaking from a radiologist’s standpoint, we have not been as receptive and listened to these experts in what we need to do, as we should have been.
Playing devil’s advocate: medical imagers have a responsibility in patient care. What if we end up in lawsuits 20 years from now because we imaged a patient 30 times, and the patient got 500 mSv and developed lymphoma? To my knowledge, that hasn’t happened, except for isolated, very-high-dose exposures. But the patient and legal counsel are not going to come after the cardiologist or the ED provider. They will come after the radiologist. To defer and to be demure, and to say that the referring clinician knows best is inappropriately dismissing our responsibility.
TONY DeFRANCE, MD: I work mainly in outpatient centers, and we don’t want to jeopardize relationships with our referring doctors, because that’s our life blood, but at the same time, many of them do not order appropriately, and do not understand what they are ordering. I routinely get two word histories like “chest pain.” I actually go in, and talk to the patients, and try and figure out why they are in my clinic. I also spend a lot of time educating referring clinicians about alternatives to CT, especially considering how much dose we are giving. Or I will call them and educate them, in a very nice way of course. There are some that will not reconsider their order, but most of my referring clinicians are willing to allow us to use a less radiation-heavy modality, especially if they are referring a 20-year-old woman.
DR. MIRVIS: A lot of radiologists would argue, they are too busy to educate referring clinicians. I think many would argue that this is a time consuming approach. Is there a more efficient way to make the clinical referrers aware of this en masse?
DR. DeFRANCE: Well, hopefully that role will be performed by guidelines from the American College of Cardiology and other national organizations. These guidelines should be supported by best-practice flow sheets indicating which patients should be sent on for CT, and which can be managed with less radiation.
DR. MIRVIS: Tony and Don raise a good point. Let’s take it a step further. Hypothetically, let’s say that a radiologist finds himself in court because 10 CT scans were performed—they were done on a complicated, 20-year-old trauma patient with multiple injuries, infections,the whole 9 yards. Years later this patient develops lymphoma and is suing because the radiologist was the director of trauma radiology at the time. The defense is that the scans were medically necessary, and that they used the best technology available at the time, and that the patient was scanned as safely as possible. What do you think the jury’s opinion would be? Please consider that in this scenario, the medical community has never established scientific proof, that all of this radiation has really made any biological difference. In fact it is very unlikely that it would have influenced this patient’s ultimate disease.
DR. BRENNER: The jury would be offered the reasonable-man instruction: what would a reasonable and prudent physician do under similar circumstances. Your actions would be examined to the extent that your institution had developed parameters for performing CT, and that you had developed documentation that your CT scan was providing a relatively low dose for the kind of examination requested. It would be necessary to establish that there was a deliberate process in the ordering of these studies, based on the clinician’s impression and the discussion with the radiologist consultant (not just the radiologist interpreting the study). Presumably, the case would be made that the equities lie in favor of performing the examination. The plaintiff’s lawyer would likely call an expert witness to talk about CT radiation dose as it compares with spending a week camping out at Chernobyl. The plaintiff’s lawyer would make a comparison between the radiation the patient received and the aftermath of Chernobyl or Hiroshima, because that is a much more frightening comparison for the jury than simply stating that one CT scan is equivalent to 200 chest X-rays, or stating a value in milliSieverts.
It would really come down in tort law to the strength of the reasonable conduct. And the more deliberative the process is, the stronger it serves on the part of the defendant. When jurors are asked why they ruled a certain way on post-trial interviews, it turns out that they want to believe that their physician is doing the right thing. We want to assure the patient that the examination is not being taken cavalierly, that it has been deliberated on, and that the physician is doing the best they can. When jurors are persuaded that this is the case, they’re more likely than not to fall into the statistics that 70% to 80% of these types of trials are actually in favor of the defendants.
DR. MIRVIS: So if a radiologist in practice read your words and heeded your advice, chances are they would be in good shape if such an accusation was made?
DR. BRENNER: Yes, assuming that those parameters were met, that you had a creditable defendant, credible witnesses, and that you had a good attorney. What we do is like a movie script. However, a bad director or bad actors can often poison a script. Conversely, good actors and a good director still cannot revive a bad script. So if you act inappropriately, it doesn’t matter who is defending you, usually, or who the experts are. But if you performed reasonable conduct, then it can be presented in a manner that usually substantiates that deliberative process and prevails.
DR. MIRVIS: We’ve covered a tremendous amount of territory today, and so I want to thank all of you for your efforts and contributions today.