f07a.jpg (7474 bytes)Any time you hear discussions of digital X-ray image quality, you hear the term DQE — detective quantum efficiency. You know it’s important. You know it’s one of the factors you need to consider in comparing digital X-ray systems from an ever-growing number of manufacturers. And if you’ve shopped around at all, you’ve probably been told by several manufacturers they have “the highest DQE in the industry” and noticed that those numbers can vary dramatically — as much as 40 percent, depending on who you ask.

So how do you really compare system performance to system performance? How do you know if one vendor’s measurement of DQE is the same as another’s?

That question is being addressed by a group of industry representatives banding together as the International Electrotechnical Commission’s (IEC) Working Group 33 — which is working to develop an industry standard for the method of determining DQE. If you’re not a physicist, the numbers and equations can become confusing, but a standardized method of measuring DQE will at least let you know you’re comparing “apples to apples” in terms of image quality.

One bad apple…

The debate has been brewing for some time — even before the first DR systems received FDA clearance. Marketing materials tout a certain DQE for a company’s DR system, but the conditions under which those measurements are acquired are not explained. Papers presented at scientific meetings provide a little insight, but the process for measuring DQE varies from study to study, providing room for question.

The issue has become such a topic of debate that some DR manufacturers decline to provide DQE measurements or retract some earlier measurements for their own systems. An informal poll by Medical Imaging found most manufacturers reluctant to provide DQE figures for publication, especially in the context of the work being done by this group.

Taking a Slice: Measuring the Noise

f07b.jpg (6984 bytes)There’s a second standards effort underway concurrent with the IEC group focusing on the measurement of Noise Power Spectra. The American Association of Physicists in Medicine is sponsoring its AAPM Diagnostic Imaging Task Group 16: Standard for Measurement of Noise Power Spectra.

The group is chaired by Andrew Maidment, Ph.D., assistant professor of radiology and director of radiological imaging physics at Thomas Jefferson University in Philadelphia. Maidment says the group hopes to write a “cookbook” for the method used to calculate noise in digital radiography images.

“This first came up because physicists in the field were evaluating detectors and a lot of them didn’t have the background needed to do these measurements,” explains Maidment. “So the chair of the AAPM Diagnostic Imaging committee for AAPM asked me to lead this effort.”

Some of the members in the two groups are the same, but the groups are developing two separate standards. The reasons for the standards are similar — the measurements for these two factors of image quality need to be standardized to allow fair comparison of the products. Maidment says measuring noise in digital images is greatly different than analog and the group is focusing on the noise aliasing that takes place in amorphous selenium detectors.

“If you go to the medical imaging conferences, you get people presenting numbers that are not possible,” says Maidment. “What this will allow the users to do is compare the two curves for noise power spectra from two different manufacturers.”

The two groups are hoping to integrate the two standards at some point in the future, but that would require some discussion down the road. At this point, Maidment hopes to have a standard out by summer 2001, but admits the volunteer-dependent Task Group 16 has missed deadlines before.

But most manufacturers realize that establishing a standard will be in the best interest of the fledgling DR market.

“For us, it’s important to be able to give customers the data and let them compare that data to the other devices available,” says Stefanie Apeldoorn-Rassow, head of quality management and radiation physics at Swissray International Inc. (New York) and a member of Working Group 33. “That’s only possible if you use the same algorithms and measurement process.”

Raoul Bastiaens, the representative on the group from Philips Medical Systems International B.V. (Best, Netherlands), feels there has been a growing consensus among manufacturers, researchers and governmental institutes that DQE is the most important parameter for characterizing the imaging performance of an X-ray detector. Also, generally accepted is the fact that there is no standardized measurement procedure for the determination of DQE, so published data on this important parameter cannot be compared.

Bastiaens says the recent commercial introduction of several flat-panel DR systems pressed the issue and made it necessary to create a standardized measurement procedure for DQE determination.

Brian Rodricks, principal scientist at Hologic Corp.’s (Bedford, Mass.) Direct Radiography Corp. subsidiary and member of the group, agrees with that sentiment. “There are many different techniques you can use to measure DQE and all can give you slightly different results,” Rodricks says. “Very often, the conditions under which the DQE is measured are not specified. Someone may give you a very high DQE, but you don’t know what it means, because it’s just a number with no additional information.”

What matters

DQE should be one of the main factors that go into evaluating a DR device. According to an outline of the standard now in development, DQE has been established as the best method to determine the “ability of the detector device to preserve the signal-to-noise ratio from the radiation field to the resulting digital image data.”

“The measurement of DQE can be tricky,” says Paul Granfors, a senior scientist with GE Medical Systems (Waukesha, Wis.) and representative on the IEC group. “It can result in erroneous values, if a careful procedure is not followed. Even when a correct procedure is followed, results can vary widely depending on the exact measurement conditions.” As an example, Granfors says, the X-ray spectrum used for the measurement can have a big effect on the resulting DQE measurement.

The standard will specify the method, X-ray spectra and exposure levels used for DQE measurement. Outlined areas include operating conditions, radiation quality, test object, system geometry and exposure conditions.

A draft of the standard developed by the group says, “DQE is already used by manufacturers to describe the performance of their equipment” and is required by regulatory agencies like the FDA for admission policies. However, there is presently no standard on the measurement conditions nor on the measurement procedure, which may have the consequence that values from different sources are not comparable.

Getting the apple rolling

The seedling for Working Group 33 was born as early as June 1999 when a workshop titled “New Detector Technology Diagnostics and Impact on Radiation Protection Strategies” was held in Finland. At this meeting, a need for a DQE measurement standard was first discussed and consensus was reached that a standard was needed.

From there, a small group convened in Hamburg, Germany, in October 1999 and agreed that a standardization effort was necessary and that it should come under the umbrella of the IEC. A new working item proposal was prepared at that meeting, including a rough draft of the proposal.

According to Hartmut Duschka, secretary of the group and a scientist working with Siemens AG (Erlangen, Germany), the working group was created when a “call to experts” was sent out to the 24 countries represented in the IEC Subcommittee SC62B, which specifically is aimed at developing standards in medical imaging equipment. In May, there was a positive vote from the IEC to establish the Working Group 33, which is formally titled, “Characteristics of digital X-ray imaging devices — Determination of the detective quantum efficiency.”

Currently, there are representatives from most of the major DR vendors on the working group and organizers say there are more companies coming. There also are scientists from European academic institutions working with the group. With members in place, the first meeting was held in Frankfurt, Germany, over two days in July. By all accounts, that meeting was very successful and produced positive feedback on the draft proposal.

“We had a very successful two-day meeting in Frankfurt where the initial draft was received by members,” said Philips’ Bastiaens. “Based upon those discussions, the convenor will issue a new draft for comments that will be discussed at the next meeting.”

Based on that input, there will be a revised proposal that will be reviewed at the next scheduled meeting in February in San Diego, prior to the annual meeting of the International Society for Optical Engineering (SPIE). Bastiaens says the group hopes to have a final standard by the middle of 2002.

Getting to the core

The standard as written will be applicable to flat-panel and CCD-based DR systems, as well as CR devices. According to the minutes of the first meeting, it was decided that the standard should exclude fluoroscopy and dental applications. Mammography will be included in a later, separate standard.

The first draft of the standard sets forth a number of requirements that DR vendors will have to meet to satisfy the standard. For example, the draft standard states that for all tests, equipment must use a 12-pulse or high-frequency generator, a geometrical arrangement is provided in a figure which must be met dictating the distance between tube and detector and other issues. The operating conditions for testing of DQE also are spelled out in the draft as is the radiation quality measurements. Other information includes determination of the noise power spectrum and the output amplitude.

Much of the standard revolves around complicated advanced physics equations that the average radiology administrator need not be concerned with.

Hitting home

So the question you’re asking is “How does this affect me and my radiology department?” Well directly, it doesn’t. The standard is being developed for use by manufacturers and testing houses. Members of the committee are quick to point out that the standard covers the process by which DQE is measured and not a standard on the quality of the images alone.

But the use of such a standard will provide the security of knowing the DQE measurements provided by those manufacturers are captured by the same methods.

“The standard itself will not improve the quality of the images,” Duschka says, “but the actions taken by the manufacturers to comply with this standard will.”

“The future IEC standard will help in the selection of X-ray equipment with the adequate properties for the intended applications,” says Duschka. “The DQE figure will be based on definitions and measuring methods which are identical for all manufacturers claiming compliance with the future IEC standard. This way, comparable figures will be presented to the users.”

“This standard outlines a measurement procedure, so it is not comparable to a standard that products have to comply with,” says Bastiaens. “Products will not have to comply with a standard, but have to present DQE figures measured in an approved way.”

While it may be more than 18 months away from publication, the work of the IEC will benefit the vast majority of digital X-ray purchasers. Members of the working group say users can expect a rush of new DQE measurements for most digital X-ray systems.

Andrew Maidment, Ph.D., assistant professor of radiology and director of radiological imaging physics at Thomas Jefferson University (Philadelphia), said his group is working on how to inter-compare noise power spectra from two machines on the same graph if one is measured in terms of X-rays and in terms of voltage. end.gif (810 bytes)