It's a digital age in radiology, and conventional wisdom holds that a structured report is automatically a better way to convey imaging findings. But that idea took its lumps recently when a study at the at the University of Florida School of Medicine found that free text and structured reports are equally efficient and accurate in transmitting case-specific interpretive content.
It's a digital age in radiology, and conventional wisdom holds that a structured report is automatically a better way to convey imaging findings. But that idea took its lumps recently when a study at the at the University of Florida School of Medicine found that free text and structured reports are equally efficient and accurate in transmitting case-specific interpretive content.
"People make the assertion that structured reporting almost by definition is better, that because it's computerized it must be," said Dr. Christopher Lee Sistrom, an assistant professor of radiology at the university.
The study found no difference in information transfer efficiency between free text (narrative style) and structured (itemized) reports having the same content (AJR 2005;185(3):804-812). It is the first experimental evaluation of radiology reports whose primary outcomes are quantitative measures of information transfer to readers, Sistrom said.
Sistrom and colleagues designed a Web-based testing mechanism that was used to present radiology reports to 16 senior medical students and record their answers to multiple choice questions about specific medical content for each of 12 cases.
The study was delicate enough to detect even small differences in three outcomes: time to read each case, correct score, and efficiency.
"I could have detected very small differences, and the differences I found were even smaller," he said.
Researchers had expected structured reporting to be speedier and more efficient, but to have a slightly lower accuracy, Sistrom said. However, they found no significant differences for any of the three measures.
The results were somewhat surprising, Sistrom said. Physicians may need to be a little more careful before shifting all documentation and communication from narrative structures to more strictly codified formats.
Dr. David Weiss, clinical head of imaging informatics at Geisinger Medical Center in Danville, PA, and the author of several articles on structured reporting said that he was also surprised by the results of the study.
"I use itemized reports, created using speech recognition, frequently, and I find them subjectively easier to prepare, especially with the proper integrated PACS software," Weiss said.
Weiss also clarified the term "structured reporting," which is used describe a commercial product that generates a report from predefined elements. The user selects these data elements from a pulldown menu, checkbox screen icon, or, more recently, voice command. The report that is created can be in any form from conventional prose to an itemized list.
"There has been interest in what some are calling a multimedia report. This links selected images or video clips with a specific word or phrase referring to that particular finding in a report. The image can be embedded in a report as a screen capture," he said.
Prior to the University of Florida School of Medicine study there were little or no empirical data that examined the accuracy and speed with which readers extract case-specific information from free text and structured reports. Human beings organize their work around complex tasks using narrative and normal texts and should be careful about dropping that approach, Sistrom said.
"We don't know what parts of that linguistic and semantic combination really provide synergy with the way we think," Sistrom said.
Still, even though they performed no better with itemized reports, Sistrom's subjects clearly preferred a structured to free text format.
"The potential impact of a structured reporting system goes beyond report creation and report interpretation. The data elements are stored for later use, including billing and data mining for research and educational purposes," Weiss said.
MR SCAN OF RIGHT KNEE
COMPARISON: Projection radiograph of [date deleted]
HISTORY: Pain in right leg
TECHNIQUE: Routine. No gadolinium
MEDIAL MENISCUS: Radial tear at apex of posterior horn
LATERAL MENISCUS: Negative
ANTERIOR CRUCIATE LIGAMENT: Intact
POSTERIOR CRUCIATE LIGAMENT: Intact
MEDIAL COLLATERAL LIGAMENT: Intact
LATERAL COLLATERAL LIGAMENT COMPLEX: Intact
EXTERIOR TENDON MECHANISM: Intact
ARTICULAR CARTILAGE: There is narrowing and slight heterogeneous signal in the articular cartilage consistent with mild to moderate osteoarthritis. Small osteophytes are present in the distal femur and patella.
BONE MARROW: Negative
JOINT EFFUSION: Very small joint effusion. Small Baker's cyst.
OTHER: Negative
IMPRESSION:
1. Tear of posterior horn of medical meniscus
2. Small joint effusion and Baker's cyst
3. Cartilage narrowing and irregularity consistent with mild to moderate osteoarthritis
Itemized radiology reports provide information in a listed format rather than in narrative prose. Radiologists may find them easier to create, but data transfer is no better than free-style reports. (Provided by D. Weiss)
Study: AI Boosts Ultrasound AUC for Predicting Thyroid Malignancy Risk by 34 Percent Over TI-RADS
February 17th 2025In a study involving assessment of over 1,000 thyroid nodules, researchers found the machine learning model led to substantial increases in sensitivity and specificity for estimating the risk of thyroid malignancy over traditional TI-RADS and guidelines from the American Thyroid Association.
Can CT-Based AI Provide Automated Detection of Colorectal Cancer?
February 14th 2025For the assessment of contrast-enhanced abdominopelvic CT exams, an artificial intelligence model demonstrated equivalent or better sensitivity than radiologist readers, and greater than 90 percent specificity for the diagnosis of colorectal cancer.
Emerging PET/CT Agent Shows Promise in Detecting PCa Recurrence in Patients with Low PSA Levels
February 13th 202518F-DCFPyL facilitated detection of recurrent prostate cancer in 51 percent of patients with PSA levels ranging between 0.2 to 0.5 ng/ml, according to new research presented at the American Society of Clinical Oncology Genitourinary Cancers (ASCO-GU) Symposium.