Standardizing Mammographic Workflow and Interpretation

The passage of the Mammography Quality Standards Act of 1992 (MQSA) ushered in a new era in breast imaging; with the advent of digital mammography just around the corner, the FDA set about standardizing every aspect of mammographic interpretation, from regulating facilities to developing a unified vocabulary for interpretation and reporting. Irena Tocino, MD, professor of diagnostic radiology and director of process improvement and patient information systems at Yale University School of Medicine, New Haven, Conn, says, "In mammography, a report is not finished until the radiologist has set up a path of action for the referring clinician. This is one of the core additional responsibilities that MQSA places on the radiologist. It requires the radiologist truly to be a part of a team, and that is very novel in radiology. If I read a CT of the chest, I don't have to talk to anybody, but here, you have to be sure that the clinician and patient are aware of your recommendations."

"The best way to screen patients efficiently is to set up a flow of cases and a way of reporting in a paperless environment that allow the radiologist to move from one case to the next rapidly. For that, you need all the systems integrated. Everything needs to happen within the same screen." —Irena Tocino, MD

With the additional demands placed on radiologists by MQSA, as well as the Breast Imaging Reporting and Database System (BI-RADS®) developed by the ACR as part of MQSA's requirements, it's no surprise that a standardized mammography interpretation platform is vital to optimizing workflow while maintaining compliance. "Before the early 1990s, the ACR had a program where BI-RADS was being used on a volunteer basis, but after MQSA came about, all reports had to be standardized," Tocino explains. "The main reason was so that the communication of findings and recommendations would become standardized, rather than left up to the individual radiologist's discretion." Yale uses the Synapse® PACS from FUJIFILM Medical Systems USA Inc, Stamford, Conn, but its mammographers, Tocino included, currently read on dedicated mammography workstations that came with the modality at the time of acquisition. FUJIFILM has since introduced its mammography module for Synapse PACS, and Tocino anticipates that Yale will begin reading on the Synapse PACS when the facility upgrades to the most recent version of the solution later this year. Reading on the PACS will provide many more efficiencies, as well as enhanced diagnostic tools. For the time being, however, radiologists in her department have developed their own sets of reading protocols for both mammographic interpretation and for comparing current exams with prior studies. Patients whose current studies are digital may still have some film-based prior exams; Tocino says, “Right now, we don't digitize our priors. We hang our analog priors next to our workstations and just compare them, but most of our cases are already digital, going back at least two years." Reading protocols are more individualized, Tocino says. "Everybody has a little bit of flexibility. What most of us like to do is have old films on top and recent films on the bottom to get a general picture of what has changed since last year." Because reporting is heavily standardized, interpretation is highly methodical. "I am sure there are as many protocols as there are radiologists, but we all wind up going through the same views," Tocino says. The introduction of the Synapse mammography module has the potential to refine each individual radiologist’s protocols further by enabling the mammographers to save their preferences, and even to apply them to MRI and ultrasound interpretation. One missing piece of the puzzle is anatomical standardization. "There are still some facilities where they read mammograms backward," Tocino says, referring to the outdated practice of reading the drier, reverse side of a wet film to reduce glare and increase diagnostic accuracy. She adds, "It's a source of confusion and, potentially, of misdiagnoses. I have heard stories where they performed ultrasound on the wrong breast as a result, or where they worked up the wrong breast. Perhaps 20% of mammographers in the country are still doing it that way, even after transitioning to digital. There's no good reason to do it except habit, and that's never a good reason to do anything." The Synapse upgrade will also include a bevy of new workflow-enhancement features intended to optimize mammographer efficiency, including advanced capabilities for soft-copy interpretation and computer-aided detection integration, complete with the display of computer-aided detection markers using the DICOM Structured Reporting Standard. This feature could be vital to radiologists like Tocino; she often finds computer-aided detection to be a burdensome and time-consuming addition to her workflow. "Computer-aided detection adds another step to the workflow for the radiologist," she notes. "It's used because it offers the potential to find cancers not seen by the mammographer, but it can generate false-positive markers to weed through, and it takes an experienced mammographer to handle that." The upgrade also includes support for MQSA-compliant overlays, facilitating standardized BI-RADS reporting. Tocino is an enthusiastic proponent of standardized reporting. "The BI-RADS report is structured into different sections, and pulling the data from those sections together allows facilities to understand the meaning of their findings from a statistical point of view," she says. "For instance, because we now have many years of BI-RADS experience, we can say that clusters of calcifications have a high predictive value for breast cancer." BI-RADS allows the interpreting radiologist to insert a disclaimer if the breast tissue is too dense to read using a mammogram, helping to reinforce the recent recommendation that certain patients receive breast MRI in addition to annual screening mammography. "If the patient has a dense breast, where a cancer would be easy to miss, those could be patients who are candidates for MRI screening," Tocino says. The BI-RADS system also ensures that radiologists use a common language, even when reporting specific findings at a very detailed level. "Having a standardized vocabulary was the first step," Tocino says. "The second section of a BI-RADS report is findings, and BI-RADS has very few descriptors, so we all agree on what to call what. The major categories are calcifications, masses, and asymmetries, and within those, we use subdescriptors to qualify the findings, and it all goes into the report." In the future, Tocino looks to full integration of image viewing and reporting functions as the change that will usher in a new era of optimized mammographic interpretation. "The best way to screen patients efficiently is to set up a flow of cases and a way of reporting in a paperless environment that allow the radiologist to move from one case to the next rapidly," she says. "For that, you need all the systems integrated. Everything needs to happen within the same screen." One innovation that has yet to hit its stride is voice-recognition technology, which Tocino envisions as a function used for both reporting and navigating the reporting system. "This constant clicking is a very distracting thing. I want to have earphones on and be navigating through what's happening onscreen using voice commands. If I say negative, the system should understand that I am done with the case, and the next report should come up," she says. Voice commands could be integrated with BI-RADS reporting, Tocino explains. "If I said mass, the whole modified list of what I should be saying if there is a mass would pop up on the screen. The descriptors I need would just show up,” she says. “Then, a clinician who is looking at it would only have to click on that site to see the whole description. This is how we could be our most efficient."