Training could ease perception errors with CTC virtual dissection
CHICAGO—Although new 3D software tools, such as virtual dissection, are providing diagnostic imaging clinicians new visualization possibilities for interpreting CT colonography results, specific training regarding the appearance of polyps on folds and fold distortion is important to decrease perceptual errors when using the applications, according research presented this week at the 94th scientific assembly and annual meeting of the Radiological Society of North America (RSNA).

“Optimal CTC interpretation incorporates both 2D and 3D search techniques as some polyps are more conspicuous using one display method,” said Kevin Christensen, MD. “3D virtual dissection allows a 360? view of the flattened colonic lumen, providing a rapid 3D visualization method, but is associated with anatomic distortion.”

Christensen, from the department of radiology at the Mayo Clinic in Rochester, Minn., discussed the results of his group’s study to determine causes for discrepant interpretations between primary 2D and 3D virtual dissection review techniques at CTC.

The research team conducted a study of 452 consecutive patients, 40 years of age or older, who had a screening colonoscopy exam performed as well as an untagged CTC. According to Christensen, each study was reviewed using either primary 2D or virtual dissection 3D search.

Cases with discrepant interpretations were retrospectively reviewed and false-negative interpretations were classified on both supine and prone acquisitions as: technical, exam quality, occult, perceptual (fold or fold distortion), and characterization.

He reported that there were a total of 121 polyp 5 mm in size, or greater, in 74 of the patients, with 35 discrepant interpretations that served as the cohort for the team’s discrepancy study.

There were 16 false-negative interpretations (32 positions) using a primary 3D virtual dissection search in which polyps were detected at 2D review. The false-negative interpretations consisted of 11 adenomas, three hyperplastic polyps, one carcinoma in-situ and one carcinoma, Christensen said.

“Perceptual errors accounted for the large majority of false negative exams, with over half of these secondary to polyps on folds and one-fourth arising from fold distortion,” he said. “Other reasons for false-negative interpretations were technical, quality and occult.”

There were 19 false-negative interpretations (38 positions) at primary 2D search that were seen with 3D virtual dissection. These were broken down to 13 adenomas, five hyperplastic polyps and one carcinoma. Christensen said the reasons for false negative interpretations included perceptual, characterization and quality.

He noted that limitations of the study were that the CTC exams did not include tagging; distension was not uniform throughout the cohort; and other display options were not tested.

Christensen observed that the most common cause for false-negative interpretations at both primary 2D and 3D virtual dissection search is perceptual error; and the most commonly missed finding was an adenoma of less than 1 cm on both visualizations. However, the researchers believe that false-negatives on 3D virtual dissection can be reduced with proper interpretative training on the application.

“Readers unfamiliar with idiosyncratic distortion caused by flattened colon views should be specifically trained to recognize polyps on folds and fold distortion,” he said.