WASHINGTON, D.C.—The CT dose of an individual scan may be just a drop in a bucket, but at present, radiologists know neither the size of the drop nor the rate of emission, making dose tracking inaccurate and dose reduction all the more essential, according to a radiation dose safety session presented June 2 at the annual conference for the Society for Imaging Informatics in Medicine (SIIM).
A number of high-profile overexposures to radiation and the growing use of CT have spotlighted the risks of radiation from medical imaging. In response, hospitals have stepped up their efforts to track CT dose index (CTDI) volume, while states like California have gone so far as to require CT dose reporting for all patients.
The problem is, radiology does not know the volume of radiation released by most CT scanners; nor do specialists have a solid understanding of the risks of exposure to radiation.
“We have good epidemiological evidence that higher doses of radiation may result in cancer, but [the radiation administered by] medical imaging is lower,” explained Aaron Sodickson, MD, PhD, director of emergency radiology at Brigham and Women’s Hospital, Harvard University, in Boston.
The risk of cancer is typically depicted as increasing linearly with additional exposure to radiation. But this linkage is based in large part on figures from atomic bomb survivors in Japan, dose levels that dwarf the radiation emitted by CT and other ionizing procedures.
On the other hand, “almost no radiation physicist will argue that cumulative doses above 100 mSv do not increase the risk of cancer,” Sodickson pointed out. Although most CT scans are well below 100 mSv, Sodickson found that 15 percent of patients imaged at Brigham and Women's Hospital had experienced cumulative doses above this threshold, where evidence indicates a clearer danger.
And yet, dose calculations like CTDI volume themselves are based on “problematic” dose estimates performed on phantoms several decades ago, “back when Pong was developed,” said J. Anthony Seibert, PhD, from the department of radiology at the University of California, Davis School of Medicine in Sacramento.
According to Seibert, the phantom estimates are not accurate, largely because they provide effective doses for generic male and female patients (females are believed to be more sensitive to radiation exposure). Moreover, CTDI varies for children, heavier patients and between different manufacturers’ scanners, he pointed out.
Seibert argued that these factors lead CTDI volume to deliver “potentially misleading” figures: “Let the user beware,” he warned.
The solution requires more accurate, patient- and organ-specific dosimetry estimates, which may be further off. In the meantime, Seibert emphasized, these imperfect models far outweigh not having any models. “CTDI is the only thing we have right now, so we have to use it.”
Sodickson agreed, highlighting three major steps radiologists can and should take to minimize dose. The first comes prior to scanning: order more appropriate or non-ionizing imaging. Decreasing utilization, via the American College of Radiology’s (ACR) Appropriateness Criteria or clinical decision support, is a must.
At the time of scans, radiologists and staff need to optimize radiation dose protocols, to hit the “sweet spot” between lower doses and higher quality images, Sodickson said.
Finally, capturing and reporting dose estimates post-procedure, however imperfect, is essential. These figures should be converted to patient anatomy and integrated into EHRs and dose registries, to help physicians gain a better grasp of radiation dose and to drive radiology quality control programs.
Dose registries allow individual departments to understand which scanners or technologists may not be optimizing dose reduction, while also providing important benchmarks within and between radiology groups.
Radiation dosimetry is no doubt an imperfect science, Sodickson admitted, but following proper protocols before, during and after scanning can have significant impacts on lowering dose in the short run. As for the long run, Sodickson concluded, “Informatics tools are going to be absolutely critical.”