Radiologists face many serious challenges today: increased workload, smaller workforce, and clinicians’ demand for reports minutes after a patient has been imaged. Can structured reporting (SR) help to alleviate some of these problems?

Despite the advances in imaging modalities during the last half century, the ultimate product delivered by radiologistsa reporthas remained unchanged since the specialty’s inception. Even though the American College of Radiology has developed a Standard for Communication, radiologists, like art critics, analyze images and produce narrative descriptions of what they see.1 The customary reporting process is labor-intensive, requiring seven steps (image analysis, dictation, transcription, approval, coding, billing, and distribution), is costly, and often takes hoursif not daysto complete. In addition, the process is burdened by report delays, transcription errors, report variability, and a lack of standardization (to support data mining and outcomes analyses).


A better and more efficient means of delivering diagnostic information is necessary to facilitate patient care and maintain radiology’s position as a valuable component of the health care enterprise. One approachvoice recognition (VR) technologyattempts to address the issue by eliminating transcription, which represents one of the seven reporting steps. Reports generated by VR remain narrative in nature and are subject to the same deficiencies as the conventional dictation-transcription model.

Figure 1. Disease timelines can be created to show disease progression in a structured reporting environment.

Attempts have been made since the 1960s to develop computerized SR, but these often failed due to the fact that they could not match the speed and ease of voice dictation for the radiologist. (Note: SR in this article refers to the means of creating a succinct and organized radiology report and not DICOM SR, which refers to a means of encoding clinical information in a standardized format so that it may be transferred between information systems.)

Some early examples of SR include the Missouri Automated Radiology System (MARS), Beth Israel Hospital’s Coded Language Information Processing (CLIP) system, as well as several vendor-developed systems. These early attempts, however, never established a foothold in the radiology reading room.

With changing times come changing technologies. Computerization has quickly transformed radiology in the last decade, creating innovative imaging tools that have increased expectations of clinicians, patients, and payors. Recent trends may prove to be a catalyst for the adoption of SR:

  • Transformation from conventional film to digital imaging, and from paper to electronic medical records (EMR).

  • Growth and consolidation of medical IT companies that are integrating PACS/RIS/HIS solutions.

  • Development of industry standards, including DICOM SR and HL-7, as well as the transaction profiles from the Integrating the Healthcare Enterprise initiative.

  • Public attention focused on the reduction of medical errors.

  • Threatened forfeiture of reimbursements if reports are not delivered in a timely manner.

In a 1973 article, Mani and Jones outlined several design assumptions for a SR system.2 Not surprisingly, the requirements for an effective SR solution have not changed in 30 years. In diagnostic radiology, one of the first SR successes has been in the field of mammography, with reporting systems incorporating BI-RADS nomenclature. More general radiology SR applications exist today, but widespread adoption of these systems has not yet occurred. Outside of radiology, SR has matured and grown in acceptance, especially in the fields of cardiology and gastroenterology.

The failure of many radiology SR systems may be due to the fact that they require a significant change in a radiologist’s natural work-flow pattern. The “look- away time” required to complete computerized “check boxes” or “fill-in-the-blank” forms interferes with the radiologist’s ability to stay focused on the images, and adds to his or her overall effort. As Mani and Jones noted, a change in radiology practice is required but will likely meet with resistance.2 What is truly needed is a process to “dovetail with the present modus operandi of the radiology department.”


One requirement of a functional SR system is that it fits a radiologist’s natural work-flow pattern and operates simultaneously during image analysis. A modern radiologist is now spending more time at a PACS workstation pointing and clicking on digital images, and with a few extra mouse clicks, representative images and diagnostic information can be embedded in a multimedia structured report supported by an underlying database. A radiologist can immediately edit and approve the multimedia report for final distribution, in which “a picture is worth a thousand words.”

Such a reporting process mimics what radiologists have been doing for more than a century: pointing at images and stating “where” (ie, anatomy) and “what” (ie, pathology) are seen. A key ingredient for such a system is a comprehensive, efficient, and standardized radiology lexicon. Several lexicons have emerged in the past (eg, ICD-9-CM, SNOMED, ACR Index of Radiologic Diagnoses, BI-RADS), but none have been sufficient to support a complete SR solution. A RSNA-sponsored effort is currently under way to create a standardized radiology lexicon known as RADLex (, which intends to have radiologists speaking one language.

With an SR system, any subsequent image analysis functions (eg, distance measurements, voice narratives, 3D rendered images) generated to support a diagnosis are automatically stored as secondary features in the database, and it is this database of information that can increase a radiologist’s value in the health care enterprise. With a database of radiologic information, insurers, government agencies, and researchers can use this information to perform data mining of health statistics, biosurveillance, utilization management, and outcomes analyses.? For radiologists, the database can provide a more efficient means by which to perform disease tracking (ie, following disease measurements on serial examinations) and quality assurance (QA) reviews.

“Look-away time” (ie, tendency to focus on report generation rather than image analysis) has been a hurdle to both structured reporting and voice recognition systems. Studies comparing the use of SR to conventional dictation (CD) have documented increased time of work associated with SR. Langlotz conducted a study comparing CD and SR for the reporting of knee MRI examinations, and he reported the results in three categories: (1) clerical time, 36 sec and 151 sec; (2) mean viewing only time, 4 and 109 sec; (3) reporting/viewing time, 139 and 215 sec, respectively, for CD and SR.3

To overcome look-away time, a unique approach to SR enables a radiologist to record all diagnostic findings before assigning descriptive terms, thereby enabling a radiologist to “shoot first, ask questions later”; in other words, the radiologist concentrates on image analysis first and foremost, and later inputs diagnostic information to complete the SR. In a trial of this system, four radiologists read 20 CT studies using both CD and this SR technique, and the average reporting times were as follows: CD 7.54 min, SR 9.27 min. However, the extra SR time allowed the reporting process to be completed, whereas CD still required transcription (average 8.75 min) and radiologist approval of the reports (1.21 min), and when these times are factored, SR begins to surpass CD. This study revealed that SR was faster than CD alone in 39% of cases, but when radiologist approval time was added, SR and CD were even, and when the CD time was added to transcription and approval, SR was faster in 96% of cases.

Structured reporting can also overcome a common weakness of many radiology reports: because a traditional report is dictated as the images are viewed, the most important findings are frequently buried within a series of unremarkable notations. As a radiologist randomly identifies diagnostic findings, a SR system can organize the findings in a report that identifies the most important findings first. This feature requires that a radiologist assign a priority code to specific diagnostic findings to indicate those that are thought to be life-threatening or significant.? With labels indicating the importance of diagnostic findings, radiology reports can now be tracked to ensure that critical information is acknowledged by referring physicians, hence shifting malpractice liability away from radiologists.


SR solutions provide several distinct advantages over CD methods:

  • Faster report turnaround. Excessive report turnaround time (typically 24 hours or more) negatively impacts radiology’s quality of service and a patient’s quality of care. Many clinicians raise this issue as the greatest shortcoming of radiology.

  • Data mining is key. With CD, the entire narrative report is a unit of information, whereas each diagnostic finding constitutes a unit of information in SR. The ability to process diagnostic findings in a database supports many unique and valuable clinical applications.

    For example, radiologists spend inordinate amounts of time comparing diagnostic findings to prior reports and images during serial monitoring of diseases. Structured reporting’s database of diagnostic findings allows a radiologist to create disease timelines, showing progression of disease with both images and plotting of metrics (Figure 1, page 56). Radiologic findings may also be correlated with other diagnostic tests, such as surgery and pathology reports, by mapping radiology lexicons with those lexicons used in other fields (eg, mapping radiology’s RADLex to pathology’s SNOMED). In this manner, more efficient radiology-pathology correlation can be performed, hence providing radiologists with a QA measure.

  • Doorway to PACS. A referring physician often wants to review the actual images described by a radiology report. Even with a PACS system, this task is often difficult or even impossible due to poor correlation of narrative descriptions with actual image data, or limited (even nonexistent) access to image data. An SR solution can eliminate this barrier to care by creating links between image coordinates of specific diagnostic findings and the SR; hence, a referring physician can easily pinpoint a radiologist’s findings in the images when the SR is reviewed.

  • Automatic notification of important findings. Radiologists frequently assume that their responsibility ends when a report is approved and sent to a referring physician. However, radiologists must ensure that critical information is effectively communicated; otherwise medical malpractice may result. The ability to prioritize significant image findings in SR and automatically notify physicians of critical findings by telephone, fax, and emaileven pagerssubstantially reduces the risks associated with poor communication.

  • Physician profiling. Structured reporting methods may be used to measure how efficiently and accurately radiologists review examinations. During the process of computerized reporting, the total review time, time spent per finding, and number of findings can be recorded for utilization management studies. The specific findings in SR can be correlated during overread sessions, or with computer-assisted diagnosis yield a QA metric.

    Structured reporting can also be used to monitor the ordering practices of referring physicians. These features create a utilization management and QA metric that would be appealing to hospital administrators and insurers. SR could also have applications in resolving turf battles with nonradiologists who want to read and bill for radiologic procedures. The key to winning these turf battles may be for radiologists to adopt and promote SR by which radiologists image interpretation skills can be compared to those of nonradiologists, similarly to what has been done for mammography with MQSA-mandated reporting using the BI-RADS lexicon. A question for the radiology community to ask the government, insurers, and the public may be, “Can a surgeon’s scribble in a chart regarding an office imaging study (eg, vascular ultrasound) be sufficient to justify an operation?” Unless radiologists can find a way to regulate the final imaging product (ie, the report), then imaging is an open field for any physician to practice.

  • Medical information portals. Diagnostic findings contained in an electronic SR can be linked to information portals, thereby putting a world of information at the fingertips of radiologists, clinicians, and patients. For example, the use of “anatomy-pathology” codes can establish portals to libraries of online medical information (eg, textbooks, atlases, commercial web sites) that can assist the radiologist during image analysis, enlighten a referring physician after receipt of a report, and facilitate communication of diagnostic information to patients.


The demand for SR’s functionality and value-added proposition may finally be the impetus for widespread adoption. For clinicians wanting to improve patient care, features like disease tracking and data mining will be appealing. For hospital administrators focused on decreasing costs, SR’s reduction in labor-intensive reporting processes and its support of electronic billing will be attractive. Finally, for the radiologist who will be asked to read more studies, yet provide better patient care, SR offers the ability to perform previously difficult tasks (such as disease tracking) in less time, produce a more uniform “product” to share with clinicians, dramatically reduce report turnaround time, and possibly even reclaim market share lost to other specialties reading imaging studies.

Continued refinement of SR, including integration with voice recognition and natural language processing, will be necessary to ensure that needs of radiologists, clinicians, and patients are met with unparalleled success.

David J. Vining, MD, is clinical associate professor, Department of Radiology, Wake Forest University School of Medicine, Winston-Salem, NC, and principal of a structured reporting software company.


  1. ACR Standard for Communication: Diagnostic Radiology. Accessed May 13, 2003.
  2. Mani RL, Jones MD. MSF: a computer-assisted radiologic reporting system. I. Conceptual framework. Radiology. 1973;108:587-596.
  3. Langlotz CP, et al. Pilot comparison of report creation times for structured reporting and conventional dictation. Presented at: RSNA; November 7, 2001; Chicago. Abstract 604.