Report from RSNA: New tools could bolster efficacy of CTC

December 7, 2006

Computer-aided detection and colon dissection display software can improve the efficacy of CT colonography, but the usefulness of both tools depends on the radiologist’s experience and preferred method of reading.

Computer-aided detection and colon dissection display software can improve the efficacy of CT colonography, but the usefulness of both tools depends on the radiologist's experience and preferred method of reading.

Colon CAD can identify colorectal lesions of significant size, even in patients who have not undergone dietary restriction or colon preparation, as long as stool is adequately tagged, according to a study presented at the RSNA meeting.

Dr. J. G. Feltcher, a radiologist at the Mayo Clinic in Rochester, MN, used CAD to evaluate CTC data sets with fecal tagging in 30 patients who had at least one lesion equal to or greater than 1 cm in size. He found that CAD detected 94% of polyps greater than 1 cm.

But another presenter said that CAD may by complicated by fecal tagging. Zsolt Tarján of Semmelweis University in Budapest evaluated the effect of electronic colon cleansing on polyp detection with CAD, 2D, and 3D-based readings and discovered variable results.

Tarján evaluated the results from 64 patients who underwent CTC and optical colonoscopy. Of these, 43 patients had both cathartic preparation and fecal tagging, and 21 patients had tagging only. He reported that CAD combined with electronic cleansing can detect large and small polyps at a rate similar to a radiologist's. However, he also found that CAD detected more false-positive polyp candidates than a radiologist who relies on 2D views. Three-D reading of cleansed CTC data also improved polyp detection but increased false-positive rates.

Colon dissection software is another approach to CTC. Dr. Franco Iafrate of the University of Rome compared the sensitivity, specificity, and postprocessing time of a colon dissection approach with regular 3D endoluminal workup of CTC.

The software proved most effective in imaging small lesions 5 mm or less. The overall per-lesion sensitivity using colon dissection software was 54.1% for lesions smaller than 5 mm, compared with 35.3% using the endoluminal view. For larger lesions, the endoluminal view proved similar or more effective.

Colon dissection software required less time for image evaluation. The average time for CTC evaluation with the software was 11 minutes, as compared with 39 minutes when the endoluminal view was used.

Colon dissection software offers promising results but only when a full cathartic preparation is used, Iafrate said.

For more online information, visit Diagnostic Imaging's RSNA Webcast.