Meeting the challenge of structured reporting

Jun 22, 2006

Structured reporting with speech recognition is a combination of technique and technology that permits radiologists to create diagnostic reports using a formal, defined framework within a digital environment. Proponents of structured reporting point to its potential to speed workflow, reduce costs, increase referring physician satisfaction, and enable practice analysis and outcomes management.

However, achieving these benefits is not a task to be undertaken lightly. Implementing these systems involves many challenges and pitfalls, according to a pair of presentations at the Society for Computer Applications in Radiology (SCAR) conference in Austin, TX, earlier this year.

"The main disadvantages with speech recognition and structured reporting -- proven with speech recognition and inferred with structured reporting -- are a time penalty and a potential distraction from image viewing; and, unlike the advantages, these are squarely on the radiologist's shoulders," said Dr. David Weiss, clinical section head of imaging informatics and director of radiology at Geisinger Medical Center in Danville, PA.

Weiss, and Dr. Curt Langlotz, an associate professor of radiology at the University of Pennsylvania in Philadelphia, believe that structured reporting is inevitable across the radiology workspace. They noted that radiology in the U.S. has already embraced the concept with its use of structured reporting in mammography exams, the American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS).

"I would like to see a marriage between speech recognition and structured reporting using the best features of each so that we can have the advantages of speech recognition in terms of report creation and also the advantages of structured reporting in terms of data mining," Weiss said.

Langlotz identified three major challenges still facing the widespread adoption of structured reporting across all radiology modalities:

The potential for increased "look away" time
The change in reporting workflow
The lack of adequate imaging lexicons

"A good integration can eliminate distractions," Weiss said. "It gets eyes and a mind back on the images, and it makes it very easy to be paperless, at least within the reading room. The problem is it varies greatly with the vendor."

Weiss utilizes a technique he developed called "key image reporting" to maximize his reporting workflow and harness the benefits of structured reporting. The concept involves using a PACS' image annotation capabilities to create key images during a study review, then using the PACS to display those images in a key image dataset.

"If you can set your PACS to do that, every time you annotate something to create a key image, and then you take that series and tab through the key images, you can report that way," he said. "Or, even better, you may want to display all those images in a stack mode so that you don't have to go from series to series -- or better yet, in a tile mode."

Once the images are displayed in a mode satisfactory to the radiologists, they can begin their dictation. Weiss advised having the dictation screen visualized on the display as series within the tiled images, allowing the entire workspace to be seen simultaneously.

"I've separated the search process from the reporting process," he said.

The benefits to this workflow process are twofold, Weiss explained. The process creates an annotated image set for the referring clinician and diagnostic radiologist, which can be very helpful when the same patient is seen a few months later.

Macros, the creation of a report template that structures text to fill in elements within that framework, are very helpful in driving workflow in findings-only dictation.

"In a busy private practice I used about 250 macros, and I used them nearly 100% of the time," he said. "There was really not a time when I did not at least use a macro shell that got me much of the way there. I'm a big macro proponent; other people don't like them."

Weiss is testing the capability of speech recognition and structured reporting software to dictate his impressions "freestyle," with the application automatically placing the text in the proper place within the macro template. This technology, he noted, is still being developed and accuracy metrics are not yet complete.

A principal limitation in the adoption of structured reporting is the lack of lexicon availability for all imaging disciplines. The RadLex project sponsored by the RSNA is working to unify and supplement radiology terms in other lexicons, such as the ACR Index, SNOMED, the Unified Medical Language System (UMLS), the Fleischner Society Glossaries, and DICOM to develop a comprehensive radiology lexicon

Once completed, the terms will be made freely available on the Internet with cross-references to the other lexicons and standards. The project has a completed draft of an anatomy and pathology lexicon for thoracic radiology, and work is currently ongoing in five other areas of radiology: abdominal, cardiovascular, musculoskeletal, neuro, and pediatric. An overview of the project can be downloaded here.

For those practices that are interested in becoming early adopters of speech recognition and structured reporting, Weiss offered some practical advice.

"If you are considering purchasing an alternative reporting system, make sure that you give your vendor a complete list of your requirements for your integration," he said. "Work both with your PACS vendor and your speech recognition or structured reporting vendor."

By Jonathan S. Batchelor
AuntMinnie.com staff writer
June 23, 2006

Transcriptionist proofreaders ease transition to speech recognition, January 27, 2006

Cost-cutting drives European SR adoption, January 20, 2006

Speech recognition improves service to clinicians, June 5, 2005

Speech recognition drives more concise reporting, June 4, 2005