Point-and-click designers take on radiology report

March 6, 2003

You've already bid farewell to films. After a few weeks of intensive speech recognition training, you're about to say good-bye to the transcriptionists. With digital image production and delivery, and automated speech recognition for creating exam

You've already bid farewell to films. After a few weeks of intensive speech recognition training, you're about to say good-bye to the transcriptionists. With digital image production and delivery, and automated speech recognition for creating exam reports, what more could anyone want for an all-digital, all-automated radiology department? How about a clearer, more standardized way of expressing radiology findings? It may be time to say hello to structured reporting.

Structured reporting has been touted as a point-and-click alternative to automatic speech recognition reporting technologies. Several companies, including eDictation, PointDx, IDX Systems, Siemens, ALI Technologies, PenRad, and McKesson, already have structured reporting products on the market. But they have to fight not only the reluctance of time-strapped radiology departments to embrace a new reporting paradigm, but also the perception that structured reporting is not yet ready for prime time.

"Structured reporting is a clinical reality today and should be considered by any practice interested in decreasing the cost of reporting and improving reporting workflow," said Dr. Curtis Langlotz, president of eDictation, a developer of structured reporting products for the radiology community.

In October, the National Institute for Biomedical Imaging and Bioengineering awarded a $750,000 Phase II Small Business Innovative Research grant to eDictation. The funds will be used to develop real-time decision support prompts for radiologists and to validate the practice management features of the company's structured reporting database, according to Langlotz.

Though structured reporting technology has been around for decades, it has made only brief forays into specific modalities, such as mammography and ultrasound, where it has encountered limited success.

"Although structured reporting systems have been recognized as desirable for many years, previous systems have not been widely adopted," said Dr. Annette J. Johnson, an assistant professor of radiology at the Indiana University School of Medicine. "This is primarily because of their limited expressiveness, static interfaces, and insufficient computing power."

Recent systems from companies like PointDx and eDictation attempt to overcome some of these limitations, said Johnson, who is conducting a study that compares structured reporting to conventional dictation. She expects results of the study to be available within 18 months.

A few institutions have had some success with structured reporting applications. The ultrasound department at Brigham and Women's Hospital uses Ultrastar for structured ultrasound reporting, according to Dr. John Carrino, an assistant professor of radiology at Harvard Medical School. The application allows technologists to do a lot of the prep work, such as checking whether a liver is normal or has a lesion.

The radiology department at Harvard Medical School has developed several prototype systems for use in thoracic, breast, and abdominal imaging, according to Dr. Robert Greenes, a professor of radiology. It seems to have worked best in limited areas, such as ultrasound, where the main findings are entered by a technologist and reviewed by a radiologist. But the structured format was not always flexible enough for more complex cases.

"The structured format did not provide the flexibility of matching what the radiologist was attempting to convey, for example when several findings support an impression, and others support another impression," Greenes said. "This required considerably more time to enter and organize, and it created frustration."

REAL STRUCTURED REPORTING

Frustration with structured reporting arises because it can mean different things to different people. The challenge to radiology departments is to first understand what structured reporting is and what it can do to streamline the reporting workflow.

"Structured reporting has more than one meaning depending upon the audience. For some, structured reporting means DICOM SR, which is a way to package a report to send it from one system to another," said Dr. David Vining, CEO of PointDx.

From a modality vendor's perspective, structured reporting is a way to encode numerical measurement information that is obtained during image acquisition, according to Dr. David Clunie, chief technology officer at Princeton Radiology Pharmaceutical Research and author of DICOM Structured Reporting. DICOM structured reporting is already a reality, and vendors are working to reencode information in DICOM SR objects to achieve interoperability.

The DICOM SR effort envisioned and championed by Dr. W. Dean Bidgood Jr., a radiologist with SR Data Solutions of Durham, NC, features a self-describing information structure that uses templates and context-dependent terminology to report clinical observations, Carrino said. DICOM SR specifications allow the interchange of expressive compound reports in which the radiologist can describe critical features of specific images in unambiguous terms. Information obtained in this way can be indexed and retrieved by additional reviewers.

Vendors of structured reporting products, in contrast, say the term refers to a method for producing a standardized report, which can then be encoded by DICOM SR.

"We view DICOM SR as an ?envelope' by which to mail a letter. We provide the means to create that letter," Vining said.

While technology exists to produce both the envelope and the letter, structured reporting has yet to be widely adopted in the clinical environment. The lag in clinical deployment comes from the need to replace the traditional dictated radiology report with something containing more structure, according to Clunie.

"It certainly is possible to encode traditional text blocks in DICOM SR objects, and some vendors already do this," Clunie said. "But it doesn't take advantage of the numerical content that could be encoded with more structure, not to mention the possibility of more precisely linking regions of images directly to findings."

Knowing what the new technology is and how it will help save money and increase efficiency is essential for a radiology department. But vendors must also do their part and develop a product that meets radiologists' needs. This includes developing a product that can be seamlessly integrated into a department's radiology information system.

Vendors must demonstrate tangible workflow benefits of their systems in clinical settings, according to Rex Jakobovits, PhD, president of Vivalog Technologies and developer of the www.structuredreporting.com Web site.

"Systems need to better leverage the structure contained in reports to offer improved diagnostic interfaces," he said.

GET WHAT YOU PAY FOR

One of the clearest advantages of structured reporting is the improved quality and speed of communication between radiologists and referring physicians. With a structured report, referring clinicians won't have to slog through vast amounts of description to get to the meat.

Vendors of structured reporting technology claim that it decreases turnaround time. The traditional reporting process consists of seven steps: image analysis, dictation, transcription, approval, coding, billing, and distribution. With structured reporting, these can be reduced to a single step at a fraction of the cost, according to Vining.

"Radiologists often think of reporting as the dictation step only. They fail to realize that there are many more human steps required to get a deposit in their checking account," he said.

Structured reporting also shifts the radiology reporting paradigm from an exam-centric to a finding-centric view. The former involves tracking the progress of a finding (such as a lesion) by retrieving multiple historical reports and reading through many paragraphs of text. A finding-centric view, in contrast, involves tracking the attributes of each finding across reports, prompting the radiologist for changes or impressions, according to Jakobovits.

Using such a form of structured reporting, radiologists could click on a nodule on an image and pull up a timeline of reported findings about the nodule. The information available on a structured report would enable the radiologists to identify trends in size, shape, and location of the nodule. Additional benefits include organization of content gathered from exams in a more precise manner that is easier to search and index automatically. The gathered data could then be easily mined for research, quality control, outcomes analysis, and teaching file generation.

"Structured reporting helps to improve the organization of medical records and reduces ambiguity. Ambiguity is the enemy," Carrino said. "It will allow a radiologist to express uncertainty about a diagnosis, but it allows the level of certainty to be quantified and expressed in a fashion that is more universally understood."

All of the benefits of structured reporting come at a price, however, and many radiology departments are already mired in increasing demand and a decreasing workforce. One of the biggest disadvantages is the need for a change in the culture and behavior patterns involved in exam reporting.

"It may be difficult to convince radiologists to switch from picking up a microphone and dictating their reports, which they've been doing successfully for a long time, to a more formalized fashion of reporting," Carrino said.

As radiologists learned with the implementation of PACS and speech recognition technology, taking time away from looking at the image can be problematic. Finally, structured reporting gains its power from providing a standardized method of presentation. In much of radiology, however, no standardized language exists.

"Radiology practice groups need to achieve consensus on terminology for imaging examination reports," Johnson said. "To optimize the quality of our information output, we must communicate about medical images using a common vocabulary."

FINDING A COMMON LANGUAGE

The RSNA RadLex project was designed to study the viability of developing a radiological lexicon by creating and testing a lexicon for one radiology subspecialty, thoracic imaging. A multiyear project, RadLex has aims to develop an updated lexicon for the uniform indexing and retrieval of radiology information resources.

The RadLex project will be useful for indexing teaching files, organizing research databases, and creating clinical structured reporting systems, according to Langlotz.

"One of the missing components needed to deploy structured reporting is a set of good codes to describe radiological procedures and findings," Clunie said. "Though there are many code sets for describing billing-related issues and pathological findings, there is a void in radiology. One of the goals of RadLex is to fill this void."

The reason that a standardized lexicon has not been applied to radiology so far is obvious: Getting radiologists to agree on one standard is not easy.

"Some controlled terminologies already exist, and the goal is not to replace those but to adopt them and fill in the needed gaps for imaging," said Carrino, a member of the RadLex steering committee. "The plan is to divide and conquer in the sense that this will be an iterative process of progressing through specific organ-based domains."

An additional obstacle to the RadLex project is the fact that different modalities each have their own specific nomenclatures, which need to be incorporated into the lexicon. If the nomenclatures aren't included, the risk of misuse and misunderstanding of terms increases.

"Connectivity between information systems requires sharing a protocol, but communication requires a shared semantic context. Facilitating communication is the largest promise of RadLex," Carrino said.

At the 2002 RSNA conference, the RadLex steering committee decided to develop its first lexicon in thoracic imaging by collaborating with the Fleischner Society and the Society of Thoracic Radiology, according to Langlotz.

"A thoracic lexicon development committee will meet in the spring, with the goal of producing a thoracic imaging lexicon. Once finalized, RadLex will be freely available as a searchable Web-based resource," he said.

TAKING THE PLUNGE

Radiology departments that take the plunge into structured reporting should set priorities, provide the appropriate electronic infrastructure, and work closely with their radiologists on accepting the new technology.

"I think departments should consider carefully their priorities regarding reporting," Johnson said. "If the group considers the ability to produce concise, useful, and timely reports to be a high priority, then a structured reporting system should be considered along with speech recognition systems."

Radiology departments also need to have the correct electronic environment to support structured reporting and should make sure that any technology they look into can be integrated with their HIS or RIS. Radiologists need to find the right balance between check list, speech recognition, and dictation, according to Carrino. While several structured reporting technologies are available, radiologists may want to make sure they find the technology that supports the type of reading they do.

"Practices should assure themselves that the system they intend to purchase can be used for all the imaging modalities they perform," Langlotz said. "The ability to seamlessly integrate with their existing information systems should be a strong consideration. Reporting systems that partner with vendors of scheduling and billing systems can provide additional cost savings to many practices."

Radiology departments may want to look into systems that provide a backup reporting method, such as speech recognition or digital dictation, for those radiologists who prefer to dictate the report, rather than point and click, he said. Radiologist acceptance may be the main sticking point for implementation.

"Radiologists will resist adopting a system that imposes restrictions on their reporting," Jakobovits said. "Structured reporting user interfaces need to be flexible and introduced gently into the workflow. Free text should still be an option. The menus and controls should be voice activated. The interface should be integrated with existing dictation systems."

In addition to introducing such a system slowly into a radiology department, more drastic changes may be called for. The reward system may have to be jiggered to place value on structured reports, with adequate time and compensation provided for the effort, according to Greenes.

"This is at odds with the volume-oriented reward system we have," he said.

Despite the hurdles involved in implementation, increased economic pressure, as well as pressure from referring physicians for faster, more concise radiology reports, may make the decision to move to structured reporting a critical one.

"Radiologists need to embrace structured reporting as a necessary overhaul to the practice of radiology," Vining said. "Radiologists face many challenges: a shrinking workforce, increasing numbers of exams, and turf battles. The solution lies in changing the practice of radiology reporting?something that has not occurred since Roentgen discovered x-rays in 1898."

Related Content:

News