Early, in 2010, the National Institute of Biomedical Imaging and Bioengineering (NIBIB) in Bethesda, Md., awarded the RSNA a grant to support the incorporation of RadLex®—the standardized lexicon developed by RSNA—into radiology reporting templates. The newest initiative is to engage the vendor community in the process, according to Charles E. Kahn Jr. M.D., M.S., a professor of radiology at the Medical College of Wisconsin in Milwaukee and vice-chair of the Structured Reporting Subcommittee of RSNA's Radiology Informatics Committee.
"Thirteen groups of clinical experts, convened by the RSNA, have developed more than 100 templates to create an online template library—a database that allows users to search and retrieve report templates by imaging modality, organ system or RadLex® term," Dr. Kahn said. "We are now working with vendors to define standards and workflow for these templates, to get them into the hands of radiologists."
Structured reporting templates are designed to encourage radiologists to enter detailed, clinically relevant information, ultimately leading to better patient care and making it easier to build databases, retrieve reported information and share information among enterprises, Dr. Kahn said. The templates also enable radiology reports to capture quantitative data which can help monitor the effectiveness of treatment.
To this end, Dr. Kahn and colleagues are leading an effort within DICOM (Digital Imaging and Communications in Medicine) to build upon standards that allow radiologists to incorporate measurements and annotations into a report directly from the imaging equipment.
"Radiologists should be able to incorporate measurements from an ultrasound exam, for example, directly from the DICOM data accompanying the images," Dr. Kahn said. "This approach can simplify the reporting process and reduce potential errors. Together with developers of commercial systems, we're working to ensure RSNA's reporting templates can integrate data from imaging devices and transmit RadLex-encoded reports to electronic medical record systems."
Housed at Reporting.RSNA.org, the template library is equipped with features including a tracking function to record which templates have been most frequently downloaded, Dr. Kahn said. The committee hopes to foster an interactive community in which interested users can submit best-practice templates of their own.
Working in concert with DICOM, the Integrating the Healthcare Enterprise (IHE®) effort is developing methods for the exchange of report templates. IHE radiology committees have approved a new template exchange profile for completion in the coming year, said Curtis P. Langlotz, M.D., Ph.D., chair of the RSNA Structured Reporting Subcommittee and editor of the profile. "Through IHE, RSNA will work with vendors to create several new features for radiologists who use speech recognition systems," said Dr. Langlotz, a professor and vice-chair for informatics at the University of Pennsylvania's Department of Radiology.
"The new profile will enable radiologists to download templates from the RSNA library and other sources with a single click, to exchange templates electronically with radiologists at other sites and more easily migrate templates to a newer reporting system when changing vendors," Dr. Langlotz said.
In 2010, the NIBIB extended its longtime support of RadLex with a two-year supplemental contract to integrate RadLex with structured radiology reports. The Medical College of Wisconsin, the University of Pennsylvania, the University of Utah and Stanford University are taking part in the initiative, with Dr. Kahn serving as the principal investigator.
"First, this project allows us to link the elements of structured reports with RadLex," Dr. Kahn said. "Second, it will enable us to enhance RadLex—to find those terms that radiologists use but are not currently in RadLex—so we can add these terms into the lexicon."
"Even if one radiologist uses different terms in his reports, such as 'heart' and 'cardiac silhouette' or uses terms in a different order, linking those terms to RadLex will allow the report information to be understood across systems," Dr. Kahn continued.
Ella Kazerooni, M.D., a professor of radiology and associate chair for clinical operations at the University of Michigan, joined Drs. Kahn and Langlotz to introduce RSNA's online library of reporting templates at RSNA 2010 in an informatics refresher course. Their session, "Best-practices Radiology Reporting: RSNA's Library of Radiology Report Templates," which describes Dr. Kazerooni's experience implementing structured reporting at the University of Michigan, is available for bookmarking and viewing within myRSNA at RSNA.org.
Many of the templates in the structured reporting library are already accessible as speech macros and could potentially be integrated into existing reporting systems, Dr. Kahn said.
RSNA is working toward that goal with vendors and physicians on two fronts. "First, we have engaged the IHE community to define the workflow related to the exchange and sharing of templates," Dr. Kahn said. "Second, through RSNA's involvement in DICOM, we are defining an electronic standard format for structured reports that builds on and interacts with the DICOM Structured Reporting effort and the Health Level Seven International, or HL7, standard for clinical information systems."
"Making the templates accessible through vendors is critical, because they will provide the product to the radiologist in a busy practice looking to do good quality reporting and stay productive."
Editor's note: This article launches a new, regular RSNA News section called Technology Forum, which focuses on the latest developments in information technology and their potential to change how we practice radiology. Have an idea for Technology Forum? Contact us at firstname.lastname@example.org.
Structured reports are critical to successful "data mining"—comprehensive analyses that help radiologists use data more effectively in managing everything from disease prevalence to billing errors, said presenters of an RSNA 2010 course.
Speaking during "Data Mining, Searching, and Analytics of Radiology Reports," Keith Dreyer, D.O., Ph.D., an instructor of radiology and vice-chair of radiology computing and information sciences at Massachusetts General Hospital, used an automotive analogy to describe the processes of mining structured and unstructured reports.
"It's kind of like trying to decide whether to build a racecar or a truck," said Dr. Dreyer, who serves on the RSNA Radiology Informatics Committee. Mining both types of reports has its utility, he said, "though I really do think you need a standard nomenclature like RadLex and a standard structure."
Structured reporting also reduces problems with natural language processing encountered in data mining, said course presenter Ricky Taira, Ph.D., an assistant professor of the Department of Radiological Sciences at the University of California, Los Angeles. "We can focus more on conclusions with the metatags we put into the system," he said.
The ultimate goal is to have data mining software "learn" to detect associations in various data, seeing how many patients with one diagnosis go on to have other diagnoses, said presenter Woojin Kim, M.D., an assistant professor of radiology, chief of radiography and associate director of imaging informatics at the Hospital of the University of Pennsylvania.
Said Dr. Kim, "The computer can be working while you sleep, and in the morning say, 'By the way, I found three associations. You might want to look into this.'"
Join a global community of leaders in the radiologic sciences.
Continue your education with top-quality learning resources.
With grant applications increasing, the R&E Foundation needs you.