Use of structured reports and quantitative data is growing as providers look to offer effective, personalized medicine. While these efforts are improving communication, they still require plenty of manual operations by the radiologist during interpretation. However, new technologies are offering a way to efficiently generate standardized reports and easily capture quantitative data.
The benefits are touted in literature, at conferences and around hospital hallways. Structured reporting can improve communication with clinicians, streamline radiology interpretations and reduce errors in coding, protecting reimbursement. The movement toward structured reporting has blossomed in tandem with a push to capture and incorporate standardized quantitative measurements and a common lexicon.
In the wake of the success of the American College of Radiology’s Breast Imaging-Reporting and Data System (BI-RADS), the college has developed terminologies and classification systems for hepatocellular carcinoma (HCC) imaging findings called LI-RADS, as well as PI-RADS for the prostate and HI-RADS for head injuries.
All this structure and standardization should be a boon to reporting practice, but radiologists must buy in, and the transition can slow down workflow. Even with the LI-RADS framework, for instance, reporting suspected HCC is challenging. Radiologists have to apply the LI-RADS flowchart to each lesion, and dictate findings into a specific reporting template that they have to select and bring into the dictation software interface.
LI-RADS might get everybody speaking the same language, but the workflow still has many steps in which human error can appear. Because of this, Toshimasa J. Clark, MD, and colleagues from the University of Washington in Seattle, developed a pair of software tools to improve consistency and timeliness of LI-RADS reporting, an innovation that could aid the institution’s large liver transplant center that serves five states.
“Almost all of our liver studies are reported in a structured manner despite the lack of a mandate, but most of these structured reports are painstakingly created by copying a template and editing each field by hand,” says Clark.
The first of the two complementary applications is a mobile app for iOS devices that allows for rapid characterization of hepatic observations in various situations, including during multidisciplinary conferences. This app is available through Apple’s iTunes App store.
The second application is desktop software written in Java that enables semi-automated, real-time application of the LI-RADS algorithm while also generating structured text that can be copied and pasted into reporting software.
Radiologists still have much to do, including identifying, measuring and characterizing lesions, so the software doesn’t fully automate the process. Still, generating the report text should make workflow more efficient. Even better, says Clark, would be a tool that is integrated with computer-aided detection and reporting software to directly report lesion characteristics with minimal user interaction.
“Closer integration with reporting software and PACS would certainly foster increased use,” he says. “Mandated structured reporting, as happened with the FDA and BI-RADS, also would encourage users to find workflow solutions that produced structured reports.”
Clark doesn’t believe vendors have done enough to this point to support the use of structured reporting by creating applications such as those from the University of Washington. These homemade solutions can be freely downloaded, with the open-source desktop software available at www.liradsapp.com. With the increased attention to structured reporting, however, the dearth of structured reporting tools could be remedied soon.
“Standardized reporting is part of the natural progression of radiology,” says Clark. “While some might resist it as part of the process of making radiology reports a commodity, from a patient care perspective it leads to more consistent inclusion of all relevant information, easier parsing of reports for ordering clinicians, and, in the case of LI-RADS, consistent characterization of hepatic observations as probably or definitely HCC. If I were a patient being assessed for possible HCC, I’d want my radiologist to report it in a structured fashion using LI-RADS terminology and categorization.”
A tool that streamlines structured reporting is one thing, but an application that can aid in a complicated process like liver volumetry is a different beast. Before a transplant, the volume of a living-donor’s liver must be precisely measured and most current tools, even semi-automated ones, require extensive user interaction. Radiologists may have to outline the organ on a stack of images to acquire the quantitative data needed.
Manual volumetry of hepatic CT scans can take nearly 40 minutes per case, while interactive tools might be able to cut that time in half, but full automation is the goal to strive for.
”There’s a huge potential to improve efficiency of radiology operation,” says Kenji Suzuki, MS, PhD, of the University of Chicago. “There will be huge savings of the radiologist’s time with automated software.”
Suzuki and colleagues demonstrated just how huge the time savings could be with a study of automated CT liver volumetry software they created, the results of which were published in the October 2011 edition of the American Journal of Roentgenology.
Since there was no commercially available, fully automatic software, Suzuki and colleagues developed automated liver volumetry on the basis of 3D active-contour segmentation. They used a prospective study to compare their software with manual and software-aided interactive volumetry in 18 liver donors imaged under a liver transplant protocol.
Results showed both the fully automated and interactive software had excellent agreement with manual volumetry. The difference came in time-saved. User time for manual volumetry was 39.4 minutes while the interactive software took 27.3 minutes. The automated software required less than a minute.
The team followed up their CT volumetry software with another tool for automated MRI liver volumetry, and published a study of its effectiveness in the American Journal of Roentgenology this past January.
This software first applies preprocessing to a T1-weighted MRI of the liver in the portal venous phase to reduce noise and enhance boundaries. Then a 3D fast-marching algorithm generates a rough approximation of the surface and shape of the liver, before a 3D geodesic-active-contour segmentation algorithm more precisely refines the liver boundaries.
The automated MRI liver volumetry tool once again showed excellent agreement with gold-standard manual volumetry while slashing user time. Suzuki and colleagues reported that in a study of 23 patients, the automatic software required about a minute of processing time compared with the 24 minutes required for manual volumetry.
Suzuki says their automated software is not FDA approved, but the researchers are hoping to commercialize it to make it available to all users as an efficient way to conduct volumetry.
Rethinking the calculator
Efforts to make efficient use of structured reporting have even led to innovative uses of the simple calculator. Computerized, equation-based calculators have been used in medicine for years, but a group at Cincinnati Children’s Hospital has developed a calculator that is coupled with a structured report generator to efficiently create standardized reports.
First built by Neil Johnson, MD, and showcased as a poster at last year’s Radiological Society of North America (RSNA) annual meeting, Cincinnati Children’s applications, consisting of three separate calculators, automates certain equations and uses specifically devised language to provide information relayed in standard language that can be copied and pasted into a report.
“It generates the calculation and creates the entire report for you,” says Alexander J. Towbin, MD, of Cincinnati Children’s and lead author on the RSNA presentation.
Most commonly used to determine bone age, the calculators also can be used for leg length and femoral/tibial torsion. They’ve seen regular use at Cincinnati Children’s for more than eight years and are used in more than 2,500 interpretations each year.
The calculators are not the only way his institution is improving reporting through automated systems, says Towbin. The department’s RIS is tied to the reporting system and allows for clinical histories to be automatically pulled from clinician comments and entered directly into an imaging report’s clinical history section. A similar process automatically imports histories obtained from patients by technologists on the day of the scan.
Looking ahead, Towbin says some of the biggest advances in the area of reporting tools will come from DICOM structure reporting. “DICOM structured reporting can take information straight from a modality—like an ultrasound machine or a CT scanner—and put it directly in a report, so true quantitative information can go into a report.”
Towbin says such technology has been used in obstetrics and echocardiography for years, but is not yet common throughout radiology, though this could change in the near future as new tools are developed.
“The concept would be then that the technologist makes all these measurements, and instead of me dictating what our technologists have already measured and done a good job with, that it just goes straight into our report, and I have the chance to edit it if I think it’s incorrect,” says Towbin.
Whether it’s a fully automated liver volumetry tool or a comparatively simple calculator, tools that improve efficiency and cut down on the more tedious or clerical work of radiologists let them get back to what they do best: advancing care with high-quality reports.