Dose management (including dose-reduction strategies) is a dominant topic of conversation throughout the imaging world. Cross-disciplinary efforts to resolve the issue are moving to the forefront of both vendor and provider dockets, spurred on not least by quality metrics that tie reimbursement rates to patient outcomes.
At UC Davis Medical Center in Sacramento, California, radiology professor J. Anthony Seibert, PhD, and his colleagues are working to lower the radiation dose that their patients receive by some 20%. It’s an ambitious goal, but one that they believe is reasonable, with the implementation of statistical iterative reconstruction techniques, dose modulation, and geometry-based modeling.
The process goes well beyond CT, Seibert says, and is needed much more in interventional radiology, where per-patient radiation doses can exceed those of CT and even PET exams, in many cases. He cites the National Council on Radiation Protection & Measurements as having raised awareness of the overall impact of increased use of CT and nuclear-medicine procedures.
To gauge the scope of that impact, Seibert points out that in the United States, the natural background radiation, at sea level, is about 3 mSv per year (including exposure to cosmic rays and radon gas), but that tally rises to about 6 mSv per year—double the normal background radiation—when it includes the impact of medical-imaging procedures.
Imprecise Measures
Clouding the issue is a lack of understanding, at the provider level, of just how much radiation patients receive as a result of different imaging procedures. “We also have dose indices that may not be indicative of what the true dose to the patient is,” Seibert says.
Parameters such as the volume CT dose index and dose–length product (DLP) do not measure the dose as delivered to the patient, Seibert says; rather, they indicate the dose delivered to a calibration phantom. To the extent that these phantoms fail to match the patient’s body habitus, the inferred dose “might be significantly underestimated or overestimated,” he says.
“It gets very complicated,” Seibert says. “It’s something that’s being actively investigated, but currently, we don’t have a good handle on things as much as we would like. We can be off by a large amount—I’d speculate perhaps two to five times over or under, depending on the situation.” On most modern equipment, CT scanner manufacturers calculate the volume CT dose index based on acquisition-technique factors and measurements of a calibration phantom that is 16 or 32 centimeters in diameter. In situations that require an accurate measurement of the dose to a specific patient, medical physicists must evaluate the size differences between the calibration phantom and the patient.
A recently published task-group
report from the American Association of Physicists in Medicine1 describes a methodology for improving the estimate of the reported volume CT dose index. A large discrepancy often occurs in the imaging of pediatric patients, who can span a wide range of body sizes, complicated by the fact that some manufacturers use the 32-centimeter phantom to estimate the volume CT dose index, and some use the 16-centimeter phantom.
If medical physicists can be confused by discrepancies in the technological information that’s available to them, imagine what one concerned parent could do with some data on radiation dose from Internet sources. Seibert recalls that the mother of one pediatric patient was extremely concerned that her child was overirradiated by a postsurgical CT exam by a factor of four (according to the volume CT dose index), relative to a presurgical planning CT exam.
On further review, it was discovered that the presurgical CT scanner used the 32-centimeter phantom for a pediatric body exam (regardless of the patient’s size), and that the postsurgical CT scanner used the 16-centimeter phantom.
“We went back and performed a size-specific dose estimate to correct the volume CT dose index that’s posted on the scanner, with the patient’ imaging records,” Seibert says. “When we actually did the size-specific dose estimate, it ended up that the patient was not over-irradiated at all.”
He adds, “That just gives you an inkling” of the amount of uncorrected dose information circulating, even where best practices are in use and within the databases of dose registries. There is no doubt that estimates will be improved with time, he says; the first step is to recognize that the dose estimates do have wide variations, many of which are easily explained.
“Understand that there might be some wide variation in the data reported with volume CT dose index,” Seibert says; part of this could be attributed to the mistaken mapping of particular procedures’ definitions. For instance, a head CT scan, as registered in the PACS and the RIS, might actually represent an interventional CT scan that includes a combination head–neck study that includes CT angiography, he says.
“When you investigate, many of these outliers can be explained, but who’s going to have the time to go back and do that?” Seibert asks. “Unless somebody really needs to investigate a situation, we don’t have the manpower or the time.”
The Davis Project
UC Davis Medical Center is working to improve the situation by developing automatic algorithms that correct for body habitus for improved dose estimates, not only for the reasons mentioned previously, but also to account for changes in California law. In July 2012, volume CT dose index and DLP must be included in all radiologist reports; this, Seibert says, is adding to the anxiety of radiology administrators and radiologists because they don’t want to have to deal with yet another detail.
“We’re trying to deliver that information automatically, right in the radiologist’s report (prior to dictation), and that’s not easy,” Seibert says. Some vendors are implementing measures to automate the recognition of DICOM Radiation Dose Structured Reports (RDSR), which contain a lot more information than legacy secondary-capture objects have historically provided, Seibert says. As an example, information on how CT tube currents change with patient attenuation characteristics and many more details about the acquisition parameters and calibration phantom are included in the structured report.
“Imaging informatics is a key player in this whole issue of extracting dose information,” Seibert says. “The understanding of the DICOM standard is intrinsic to being able to retrieve the RDSR and parse the relevant content in an effective way. Cooperation among informaticists, algorithm developers, and medical physicists is extremely important. We are just at the beginning of how we can make this a lot better.”
Seibert describes the effects of the radiation-dose question on DR, which (as a field) is in transition from totally proprietary methods of calculating detector exposure, as an indirect measure of patient dose. Unlike screen-film detectors that readily indicate underexposure or overexposure conditions, DR devices compensate for these variations and do not provide a direct link from image appearance to radiation dose.
Different DR systems report exposure indices in different ways; some go up when the detector is overexposed, and some go down—and, as Seibert points out, while radiographic studies are relatively low-dose procedures, they account for 60% to 70% of diagnostic imaging.
Standardization efforts like that put forth by the International Electrotechnical Commission (IEC 62494-1) are helping to motivate manufacturers to create a uniform exposure index, he says, which can help fine-tune individual exams with a specific target index value and provide a consistent indication of underexposure or overexposure as feedback to the technologist. With consistent feedback, the likelihood of lower doses is enhanced and the need to repeat exams should be reduced.
The ultimate goal of dose estimation—whether for CT, radiography, interventional, cardiology, or nuclear-medicine procedures—is to be able to determine individual organ doses for each procedure, with the intent of accumulating information on specific organ doses (so that, at any point in a patient’s history, a reasonable risk estimate for radiation dose can be assessed accurately), Seibert says. That quest is not just a personal health issue, but a population health indicator that would be a significant improvement over the effective-dose methods currently (and often inappropriately) employed to determine risk.
“The awareness of the public of radiation dose is a good thing, but it can be mismanaged or misleading,” Seibert says. “What we really need to do is make sure that people aren’t afraid of radiation when exposures are justified. We all take risks every day; we just have to match the benefits, relative to the risks that we are taking.”
Matthew Skoufalos is a contributing writer for Radinformatics.com.