Due diligence leads to digital mammography success
Training, workflow process, and user-friendly workstations abet seamless transition to soft copy
By: Deborah R. Dakins
Digital mammography offers numerous technical advantages over screen-film, including faster image acquisition, image enhancement capability, and better storage and transmission for archiving and consultation. But its full potential can be met only through soft-copy display and interpretation.
That presents both challenges and opportunities for radiologists, who must let go of familiar film-based tools as they learn new software techniques to read mammograms.
While the only modifiers of film mammography are magnifying lenses and bright alternators, digital technology offers multiple ways to enhance a questionable image. Changing window widths and levels, using roam and zoom features, and applying customized algorithms to equalize tissue thickness are all techniques to interrogate digital data.
Mastering those tools has proved easier for some than others. Success is highly dependent on a range of factors that include workstation design, efficient use of personnel, data management between digital acquisition and display systems, and determining the best way to review film priors alongside soft-copy images.
Moreover, seasoned users have found that the easiest transitions are made when soft-copy displays support the same level of efficiency, quality, and speed achieved in the film-reading process.
"One of the main reasons to develop quality display systems is to allow radiologists to work quickly through the high volumes of screening mammography exams that need to be read," said Albert Xthona, product manager for digital mammography at Barco.
In fact, when it comes to soft-copy reading of digital mammograms, the challenge is less about learning new clinical interpretive skills than adjusting to new patterns of workflow.
And while definitive data on clinical equivalency of film and digital mammography continues to be evaluated by the Digital Mammographic Imaging Screening Trial (DMIST), users say that the superior image quality of digital mammography has already made them believers.
"Initially, digital doubled the time to read screens," said Dr. Maria Kallergi, an associate professor of radiology at the University of South Florida. "But people love the image quality. Once they become familiar with the medium-mouse, keyboards, and monitors-nobody wants to go back."
IS TIMING EVERYTHING?
The first stumbling block users face is recognizing that reading digital may take longer than film, at least initially.
Users' experiences vary regarding the time needed to interpret screen-film versus digital mammograms. One 2002 study found no significant differences between the two methods in the speed of interpretation, but researchers chalked up the finding to an extremely user-friendly workstation (Radiology 2002; 223:484-488).
Another study, conducted by re-searchers at Northwestern University in Chicago, found interpretation times for full-field digital mammography to be one to two minutes longer than those for screen-film, based in part on a slew of technical problems the users encountered.
The study involved four experienced breast imagers who reviewed 105 screen-film and 103 digital mammography cases. Interpretation times were 64% longer for digital than for screen-film.
Technical problems, as well as general unfamiliarity with digital mammography display workstations, contributed to the time lag, said Eric Berns, Ph.D., a research assistant professor at Northwestern who presented the study at the American Roentgen Ray Society meeting in Miami.
"The time difference is significant," Berns said. "When you apply it to a large volume of screening exams, it can add up to quite a lot of time."
In addition to actual time lost, there is the intangible factor of frustration that occurs when things don't work as they should, which also impedes radiologist workflow, he said.
Typical technical problems arose when images were incorrectly transferred from detector to display workstation-or not transferred at all. Another problem occurred when the storage drive on the review workstation reached capacity and refused to accept additional cases. Mismatches between patient name and identification number also caused confusion between the hospital information system and the radiology information system that resulted in an inability to locate images in the queue.
Most of these problems could be solved through better management of technologists and staff. Streamlining the process of managing digital images for reading, like the established protocols that exist for film mammography, will create efficiencies and eliminate frustrations.
Technical issues aren't the only reason readers took longer to review digital images. Use of image enhancement tools can add time to interpretation, as can a lack of familiarity with how mammograms are viewed on digital displays.
"Some radiologists are proficient at using software tools, while other struggle," Berns said. "Some will spend more time making sure they are comfortable with reading the image on a soft-copy display. Much of this has to do with computer savviness and confidence in using computers."
It's important to put time differences in perspective. The fact that digital takes longer than film may be outweighed by the advantages the new medium offers.
At St. Johns Hospital in Santa Monica, CA, the transition from reading films to interpreting soft-copy mammograms was seamless, said Dr. R. James Brenner, director of breast imaging at the site's Eisenberg Keefer Breast Center.
"Instead of a magnifying lens, software applications of local magnified views sufficed," he said. "Reading time was perhaps 15% longer, but in real-time that means 15 seconds, which is compensated, at least by CMS (Centers for Medicaid and Medicare Services)."
But comparing hard-copy prior films with soft-copy images continues to be the time-consuming part of the process, Brenner said. And because comparing priors is a fact of life in mammography, it's a challenge that breast centers will face for years to come. Sites have adopted a variety of ways to deal with the issue in terms of room design and interpretation protocol. (See "Reading room design," page 104.)
ENHANCEMENT TOOLS
Because of the limited luminance and reduced spatial resolution of soft-copy displays as compared with film, image enhancement tools are key to the interpretive process in soft-copy digital mammography. These range from window width and level to magnification.
A quick way to navigate through a soft-copy screening study is to use a single button to switch among preset window widths. By adjusting contrast in this way, readers can view images at two or three different window settings to visualize tissues and see through dense portions of the breast. (See "Digital mammography: interpretation challenges, data capture advantages," April 2003, pages 61-66.)
In addition to these standard image enhancement features, vendors have also developed advanced image processing algorithms to equalize image background and breast tissue thickness or to highlight specific features such as calcifications.
Such algorithms can heighten digital mammography's technological advantage. With film images, readers can see only what has been captured on the film, Kallergi said.
"Film mammography is superb, it has evolved over the years, and we have created something exceptionally good for printing information," she said. "But it is also limited, for example, in managing different thicknesses between chest and the edge of the breast."
A study of one such proprietary algorithm was presented at the 2004 ARRS meeting. GE's Premium View, available on the Senographe 2000D digital mammography workstation, optimizes image contrast in one window, reducing the need for windowing. Similar to standard tissue equalization (TE) algorithms, Premium View reportedly improves visualization of the breast in dense regions while maintaining peripheral contrast at the skin line and chest wall.
In the study, radiologists reviewed images more rapidly and with more confidence when they clicked on the Premium View option, according to Dr. Thomas Kolb, an assistant clinical professor of radiology at Columbia University. The technique also allowed radiologists to identify more calcifications than the standard TE algorithm with which it was compared.
"Whether PV will perform as a better screening tool than TE when used in a large clinical population remains to be seen," he said. "But its ability to detect calcifications and increase diagnostic confidence suggests that it might."
The jury is still out on whether radiologists wholly trust these tools, however.
"Each vendor has different algorithms that they can apply to make images look better, and radiologists have had a favorable response," said Berns, who uses the GE system. "But there is concern among some users about whether they may be missing something on the image when they use these algorithms."
It's also unclear which image processing algorithm is best. In a 2000 study that evaluated eight image processing algorithms, Dr. Etta Pisano found little consensus (Radiology 2000; 216;820-830). Radiologists preferred different versions of processed images depending on the task (screening versus diagnostic mammogram), lesion, and device type.
"Those findings suggest that digital mammograms are best displayed on systems that allow flexibility and easy, quick access to differently processed images," said Pisano, chief of breast imaging at the University of North Carolina, Chapel Hill. "In terms of performance, however, we don't have much data on that, and we don't really know which is best."
WORKSTATION DESIGN
Digital mammography acquires more data than film does, but the difficulty of displaying those data in a way that is easy to use has been a key complaint. Soft-copy workstations have improved over the years, but they still have a long way to go, users say.
"Reading soft-copy digital mammograms is fairly easy," Pisano said. "But soft-copy display systems are clunky. Despite improvements, they are not incredibly user-friendly."
The display aspect of a digital mammography workstation consists of twin high-resolution monitors (typically 2000 x 2560 pixels) for soft-copy interpretation. With currently available monitors (5 megapixels), only a portion of a breast image acquired with a full-field digital detector can be displayed at one time at full resolution. (See "Tips and tools optimize soft-copy displays," page 102.)
As a result, roaming and zooming features are crucial to digital mammography interpretation. But intuitive use of those tools is just as critical.
Pisano, who led the 2002 study comparing interpretation times between film and digital, said that the workstation developed at UNC and used in the trial allowed radiologists to customize features to suit. This is a factor in the speed with which radiologists read soft-copy images.
Commercially available workstations that are FDA-cleared for mammography interpretation offer varying levels of flexibility and customization. Most have programs that allow users to tailor image display protocols to their preference and to specify the order and layout of views displayed on soft-copy monitors.
How users interact with workstations is a key variable in efficient interpretation of digital mammography images, said Bradley Hemminger, an adjunct assistant professor of radiology at UNC.
In a 2003 study, Hemminger found that the more interactions required while reading a soft-copy mammographic image, the less accurate the result tends to be-and the more time it takes to arrive at it. Keeping interfaces simple is important (J Digital Imaging 2003;16;3;292-305).
"If I hand you a hammer and a nail, it's very clear that you hit the nail with the hammer," he said. "But if I give you a workstation that requires a lot of buttons to be pushed on the hammer in order to make it work, you are going to be focusing on how to do that and not on simply hammering in that nail."
In the case of digital mammography, workstations should allow radiologists to look for microcalcifications and masses, rather than focusing on what needs to be done in order to best see the image, he said.
Radiologist training, digital mammography process management, and user-friendly workstations are strategies and tools that can help sites make the transition from film-based to soft-copy mammography. But just as important as becoming familiar with digital tools is ensuring that digital technology is as good as film, Pisano said.
"Learning to interpret soft-copy images is not the challenge in digital mammography," she said. "It's showing that digital is as good as film in a large screening trial. And DMIST will show us that."
|