Imaging societies grapple with consistency of digital images

September 26, 2005

Just as beauty is in the eye of the beholder, image quality -- how good or how bad -- depends on the diagnostician. Some may like an image processed one way, and others may like it another. But there may be reasons beyond personal likes and dislikes that determine how images are judged.

Just as beauty is in the eye of the beholder, image quality - how good or how bad - depends on the diagnostician. Some may like an image processed one way, and others may like it another. But there may be reasons beyond personal likes and dislikes that determine how images are judged.

Digital systems, with their inherent differences in acquisition techniques and algorithms, generate images differently. They should, however, all deliver the same diagnostic data.

The development of standards for delivering those data is the subject of collaboration among the American College of Radiology, American Association of Physicists in Medicine, Radiological Society of North America, and Society for Computer Applications in Radiology.

Working with manufacturers, regulators, and users, the groups will assemble a panel of experts to discuss the topic. They plan to publish a white paper and request feedback on it within the next three to five months, said Ehsan Samei, Ph.D., director of the advanced imaging laboratories at Duke University.

"Technology overcomes previous obstacles, making it possible to have improved image quality," he said. "However, a poorly designed or operated system can render images suboptimal, impacting their clinical utility. This flexibility necessitates clear guidelines."

The goal of this effort is to pave the way for consistent, high-quality digital image presentations. Problems persist in this area despite efforts such as the DICOM gray-scale standard display function.

Inconsistent image presentation can particularly affect emerging areas, including digital mammography, said Dr. Margarita Zuley, a breast imager at the Elizabeth Wende Breast Imaging Clinic in Rochester, NY.

Digital mammography involves many new variables, including processing algorithms, soft-copy workstations, and display monitors, which can affect image quality in different ways.

"We still don't know how all of these variables interact in terms of image quality," Zuley said.

While a screen-film mammogram produced at one facility can be transferred easily and with full image integrity to another, digital mammograms can be site- and vendor-specific, she said. They may not be transferable to another facility with a different vendor's soft-copy workstation and PACS.

"The effect that these technical parameters and processes have on the final image quality is not completely known from a clinical perspective," Zuley said. "We need further refinement and standards for many aspects of the imaging chain."

Each of the components of an imaging system, from acquisition to display, should provide the best possible performance, Samei said.

"The main challenge remains the ability to relate various physics- and engineering-based characteristics to diagnostic quality, a goal that is an active area of research," he said.