A Clearer View: Enriching Radiologist Workflow

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon

Eliot Siegel, MDWoojin Kim, MDThe next frontier in radiologist workflow won’t be another enhancement to hanging protocols, a faster processor, or an improved graphical user interface, Eliot Siegel, MD, and Woojin Kim, MD, predict. Instead, these two experts in radiology informatics are looking outside the PACS and its workstations to offer radiologists a clearer view of patients’ enriched clinical histories. Siegel, who is professor of diagnostic radiology and associate vice chair for informatics at the University of Maryland Medical Center (UMMC) in Baltimore, says, “We believe PACS will be significantly different in the near future from what it was in the first 20 years. We want a PACS where the workflow for the radiologist is much more tightly integrated with information from the electronic medical record (EMR).”

Kim, who is assistant professor of radiology and associate director of imaging informatics at the Hospital of the University of Pennsylvania in Philadelphia, points out that as the radiologist’s role has evolved, the amount of clinical feedback that he or she receives has decreased. “A lot of radiologists will look at an image and say it’s a carcinoma and needs a biopsy, but a lot of times they’d love to find out if it was a malignant tumor, in the end, and they never get that feedback,” he says.

Siegel adds that in the analog days, a radiologist might receive some kind of informal clinical information from a referring physician—information that is absent in today’s electronic workflow. “We’re interpreting studies in a vacuum,” he says, “so the question becomes, ‘How can we, in a context-specific way, give the radiologist what he or she needs?’”

Deeper Needs

For years, radiologists have noted their growing lack of interaction with clinicians—and one casualty of radiologists’ mounting isolation is their understanding of patients’ clinical histories. Of course, this information is increasingly housed in facilities’ EMRs, but as Kim points out, “It takes a lot of discipline to go into the EMR to get that clinical information.” It is amazing, Siegel adds, that “there really isn’t something to provide me with relevant clinical information while being integrated into my workflow in the PACS.”

Radiologists aren’t the only ones seeking more from their information systems. Kim notes that the product created by radiology is, of course, the report, and the accuracy and usefulness of that report will be increasingly important to health-care administrators seeking to augment quality while controlling costs.

“The ability to mine data is very important—not just for radiologists, but also for administrators,” he says, “but once you’re doing that, you can expend to include other specialties, such as pathology, cardiology, and gastroenterology. You can give a cross-section between radiology and pathology to see which patients live in both worlds in your institution, and then see how many radiology reports were concordant or discordant with the pathology reports. If there are outliers, you can educate them.”

The potential of such access to other information systems in the enterprise, which Kim and colleagues have modeled, is high; he says, “It helps refine clinical decision support, improves quality, and reduces overall medical imaging costs.” Siegel adds that this kind of technology will not only provide additional clinical insight, but will also enhance radiologist productivity.

“The next realm will be taking information from all those different systems—including dashboarding for business analytics, departmental management, and performance—and putting them together,” he says. “It won’t slow down radiologists, if we do it in an intelligent way. You can show a lot of complex information graphically in a way that can be very rapidly appreciated.”

Looking Ahead

Kim observes, however, that leveraging enriched clinical histories in radiology will require the intelligence to preselect information that will be useful to the diagnostic process. “You might be impressed with the amount of information you can access, but you can’t look through it all and maintain productivity,” he notes. “The next question is how you know what’s important. Depending on the condition, some results are more important than others, and you have to build in intelligence to rank that for you. Otherwise, you’ll be faced with information overload.”

For this reason, Siegel envisions radiologists in the future working with an automated fellow—a computer program that presents