CT dose-tracking software could help hospitals benchmark imaging quality

August 12, 2011
by Brendon Nafziger, DOTmed News Associate Editor
Software that calculates personalized radiation dose from CT scans could help hospitals better track patient exposure and benchmark the quality of their imaging departments, according to a study presented at the 2011 American Association of Physicists in Medicine meeting this month.

"We're not keeping track of [radiation dose]," Dr. Ehsan Samei, study author and chief physicist and professor of radiology at Duke University Medical Center, told DOTmed News. "In most institutions, it's not monitored and it's not evaluated on a rigorous basis."

In general, the radiation dosage from CT scans captured by hospitals is based on output from the machine, and doesn't take into account patient-specific features, such as gender and age. As a result, risk estimates for patient populations at hospitals can be skewed, according to Samei.

Radiation risk is not one-size-fits-all. For instance, for reasons that are poorly understood, women are more vulnerable to radiation than men. Children are also more radiosensitive: their cells, growing rapidly, are more easily damaged, and they have longer lives ahead of them in which to develop radiation-induced cancers.

"There are huge variations from patient to patient that we're not accommodating," Samei said.

To fix this, Samei and his colleague Olav Christianson developed a computer program that would properly track radiation delivered to patients and gauge their risk by taking into account patient-specific factors. The program worked by capturing CT image data from the device that was uploaded to a dosimetry server.

The preliminary study, presented at the AAPM meeting in Vancouver, drew from 6,500 CT scans collected over a 5-week period at the Duke University Medical Center and Duke Raleigh Hospital.

A surprising result of looking at the data, Samei said, was the change in how risky the scans appeared to be to patients. By taking into account personalized dose statistics, risk estimates for patients at his hospital fell by a factor of two. That's because most of the patients showing up at the hospital for CT scans are older, Samei said, so they have a lower risk.

"If you don't take that into consideration, you'll come up with a risk factor that's significantly higher," he said. "Peak age [for scans] might be 60 years old. By saying everyone's 20-to-30 years old, you're overestimating the risk."

That said, Samei emphasized that the presumed risks from CT scanners are actually very small. The lifetime risk for getting cancer is already so high, at 42 percent, he said, it's hard to tease out the small likely additional risk contributed by relatively low-level medical radiation.

Now that he has developed the program, Samei said he wants to use it as a benchmark so the hospital can see if there are variations in quality among different rooms at the same hospital, or between institutions.

Also, if the work pans out, better tracking of dose could help hospital staff catch problems early on. For instance, in a widely reported case, in 2009, Cedars-Sinai Medical Center in Los Angeles overexposed hundreds of patients with brain perfusion CT scans. But Samei said if they had a system in place to constantly double-check dose, it possibly could have been stopped.

"Maybe they would have exposed the first two or three, but would have caught it and prevented it from going on for a long time. This would provide us an opportunity to identify cracks in our system, if the cracks exist," he said.

More immediately, Samei said his team is wrapping up a five-year study to relate radiation risk to image quality in pediatric patients, which they hope to have published next year.

"A major motivation of our work is that medical operations are complicated," he said. "They're team oriented, many people are supposed to do different parts of an overall task. And if someone fails in one element of that task, bad things can happen. That's why it is important to have a dose and risk monitoring process in place to ensure we can optimize and maintain a high level of quality of our clinical operations in spite of their inherent complexities."