Dose metrics lag behind advances in CT scanners

Paul Orlando thought he had an easy task. He would survey all the CT facilities in New Jersey to gauge the average x-ray output of each scanner. When he had that information in hand, however, he found he couldn't make sense of it. The method of calculating dose varied not only from institution to institution, but often from machine to machine within the same facility. There was no way to track the trend until he could normalize the data.

Orlando, chief of the New Jersey Bureau of Radiological Health, looked for help. He convened a meeting with New Jersey's medical physicists to gather input on how best to report dose for the state's fledgling CT Radiation Quality Assurance Program. The physicists unanimously chose the techniques used by the American College of Radiology's CT accreditation program. Introduced in 2002, the ACR program is gaining momentum to become the standard, Orlando said.

"I'm comfortable using this method because I can replicate the data," he said. "These data can then be cross-checked within institutions or regions to see if there are any anomalies. Facilities with aberrations would be asked to explain their spike in radiation dose."

CT dose differs from that of conventional x-ray in that the rotating beam produces peaks and valleys of x-ray intensity. A uniform measurement of output is difficult to obtain. Nevertheless, the FDA mandated the CT dose index (CTDI) standard in the 1970s. It was imperfect but a good start, according to Orhan Suleiman, Ph.D., senior science policy advisor in the FDA's Office of Drug Evaluation.

Each advance in CT technology, however-from four slices to 16-has made the CTDI standard more imperfect. Great swaths of irradiated tissue are left unaccounted for because the CTDI was designed for single-slice axial scanning. Scientists tweaked it repeatedly, resulting in an alphabet soup of subscript addenda: CTDI100, CTDIair, CTDIw, MSAD, CDIvol, and DTLP.

"The community has known for a while that CTDI was useful for single-slice machines but becomes increasingly wrong to use for helical and multislice scanners," said Richard Mather, Ph.D., senior manager of clinical sciences for the CT business unit at Toshiba American Medical System.

A year ago, fewer than 50 scanners in New Jersey were accredited. Today, that number tops 100, with three dozen pending. As more of the state's 321 scanners receive the ACR's imprimatur, the data on x-ray output should be easier to collect and analyze. The ACR uses CTDIvol and DLP (dose length product) to measure exposure levels.

In the meantime, Orlando makes use of an ACR spreadsheet to help normalize the data he has collected so far. When all the data are processed, Orlando will know where his state stands in relation to others regarding CT dose output. But he has run into another problem: The ACR technique, imperfect but adequate for scanners up to 16 slices, bottoms out with 32-, 40- and 64-slice scanners.

"We are in a crisis situation," said Robert L. Dixon, past president of the American Association of Physicists in Medicine and a professor of radiology at Wake Forest University School of Medicine. "We have dosimetry by decree. The FDA has certain regulations and the IEC (International Electrotechnical Commission) has certain recommendations for measuring dose. And neither of them works."

Studies have shown that CTDIvol, the most current standard developed to account for pitch, underestimates dose by 30% in the center of a body phantom and by 10% in the head, Dixon said. To arrive at the CTDI, a physicist uses a 10-cm-long ion chamber and makes one axial scan. For that to represent the dose, the ion chamber has to collect the primary beam width and the scatter tails, which may extend 10 cm or more beyond the primary beam width. The ion chamber has to be either longer or shorter than 10 cm to achieve accuracy. Dixon suggested the latter nearly two years ago (Medical Physics 2003; 30([6]:1272-1280).

"My method is better than what is currently being used. CTDI is just a prediction of dose. My technique actually measures the dose," he said.

Dixon's method uses a small-volume ion chamber and scans long enough to encompass the scatter tails. The test takes 15 seconds or less to perform, making it no more time-consuming than the long chamber method. But change comes slowly. Although Dixon's idea is receiving attention, no official bodies have endorsed it, said Richard Moran, Ph.D., chair of the ACR commission on medical physics.

At the 2004 RSNA meeting, the National Electrical Manufacturers Association hosted a gathering that included physicists and representatives from industry, government, and radiology. Their mission: Reach consensus on CT dose metrics. The group easily agreed that CT scanners should record x-ray output using CTDIvol and DLP, and it agreed on the phantom type (the same endorsed by the ACR). It could not agree, however, on the method to calculate effective dose or whether to measure CTDI in free air, said Stephen Vastagh, industry manager for NEMA.

Once the stakeholders reach total consensus, the DICOM standard writers must develop the mechanism with which the equipment can record the dose. Manufactures will then implement the standards. Each of the latter two steps should take a year, Vastagh said.

EVER LOWER DOSES

Dose is a competitive issue in scanners. Everyone wants scanners that do the most with the lowest dose, and manufacturers have responded with several automated techniques that regulate dose according to body size and width. But support is lacking for development of a method that more accurately measures the actual dose patients receive, said Hugh Morgan, Ph.D., a senior staff scientist in the CT engineering department at Philips.

That is a much more difficult measurement to make. And there is considerable resistance to forcing manufacturers to provide the actual patient dose, said Morgan, who is a member of the IEC committee that writes the safety standards on dose.

Research efforts are under way to use Monte Carlo techniques to calculate patient dose. Monte Carlo is a brute force computer method that calculates complicated problems. Using computers, researchers are generating millions of virtual x-rays and following them through to obtain better estimates of true dose to a specific model. This line of research will not be mature for several years, Mather said.

At Duke University, researchers do not bother with the CTDI. In their experience, the index miscalculates x-ray output by 20% to 30%. Instead, they use a real-time radiation detection system called MOSFET borrowed from oncology. It involves use of anthropomorphic phantoms that contain 20 high-sensitivity diagnostic detectors. The data are transferred automatically into a laptop computer after exposure.

"Each detector gives a reading of actual dose to that organ rather than an indirect estimate such as the CTDI," said Terry Yoshizumi, Ph.D., an associate professor of radiology and director of radiation safety at Duke.

The scan takes a few seconds and results are ready immediately, unlike thermoluminescent detectors, which are labor-intensive to prepare and take days to deliver the data. The technology is catching on slowly, mainly because the system can cost up to $50,000.

Yoshizumi and colleagues validated MOSFET using dosimeters and against Monte Carlo simulations. They design scanning protocols around the dose information gained from MOSFET measurements.

"It's very useful information that is not available in a textbook," Yoshizumi said.

Morgan has talked with the Duke researchers about MOSFET. Physicists want a more realistic patient dose index, but it's difficult to provide, he said.

"There are objections, and not everybody agrees on the method. As an industry, we want to do things that are common among all the players. We need a standard to follow," he said.

While CTDI might be mortally wounded, no clear winner has emerged to replace it. Manufacturers have responded in the past to the public's demand for lower dose. Perhaps they will react again when patients-or physicians-insist on knowing the actual absorbed dose.

Mr. Kaiser is news editor of Diagnostic Imaging.