2D or 3D? That is the question in virtual colonoscopy

November 5, 2004

Most virtual colonoscopy experts consider primary axial views with 3D for problem solving to be the optimal method for reading exams, according to a survey taken last year at the Fourth International Symposium on Virtual Colonoscopy.

Most virtual colonoscopy experts consider primary axial views with 3D for problem solving to be the optimal method for reading exams, according to a survey taken last year at the Fourth International Symposium on Virtual Colonoscopy.

But several studies since have indicated higher sensitivity with 3D primary reads, making some experts hesitant to endorse 2D primary reads.

Last year's survey of 25 symposium participants found that 80% were comfortable with the 2D approach, said Dr. Matthew A. Barish, an assistant professor of radiology at Brigham and Women's Hospital. No consensus was reached regarding the minimum requirement for interpreting virtual colonoscopy images.

Many approaches are available to physicians for reading virtual colonoscopy exams:

  • 2D only

  • 2D plus multiplanar reconstruction (MPR)

  • primary 2D with MPR and 3D problem solving

  • primary 3D with 2D problem solving

  • advanced projection 3D techniques

In the landmark Pickhardt study presented at last year's symposium, the authors concluded that CT virtual colonoscopy with primary 3D reading was an accurate screening tool and compared favorably with optical colonoscopy (NEJM 2003;349(23):2191-2200).

Preliminary results reported at this year's symposium from a Duke University study led by Dr. Don Rockey, a professor of gastroenterology, however, indicate lower sensitivity numbers for virtual colonoscopy. The results were criticized by some participants because the study used 2D views for primary reads as opposed to the 3D primary approach used in the Pickhardt study.

Others were not so quick to jump to the same conclusion. Dr. Judy Yee, chief of radiology at the San Francisco VA Medical Center, speaking at an earlier session, said that even the Pickhardt study, with its good results, was not specifically designed to compare 3D with 2D primary reads.

This debate continues in the absence of results from studies specifically designed to address whether 2D or 3D views are better. In a preview of studies to be unveiled at the RSNA meeting this month, Barish said that two papers showed no statistically significant difference between the two image viewing methodologies, while another paper indicated improved accuracy with 3D.

A fourth paper reports that physicians spend more time with 3D than they do with 2D views. The increased time may not necessarily translate into more accuracy, according to Barish.

While experts may disagree on which read to do first, the emerging solution to improving accuracy involves a combination of using both 2D and 3D methodologies for the time being and continually exploring advances in imaging technology, Barish said.

"There is no firm accuracy evidence at this time to prefer one method over another," he said.

For more information from the online Diagnostic Imaging archives:

Duke study renews challenge of CT colonography

Radiologists enter round two of colonoscopy debate

3D virtual colonoscopy gets thumbs up

Virtual colonoscopy