1. Home>
2. Specialties>
3. Gastroenterology>
4. Summary and Comment

Progress in Shortening Reading Times for CT Colonography

CAD programs can detect lesions reliably.

Who should read screening computed tomography colonography studies? The possibilities include radiologists, gastroenterologists, or both, or even other technical personnel. However, the prospect of reading CT colonography images is unappealing to many, because reading times often exceed 20 minutes and reimbursement is low. Computer-aided detection (CAD) has been investigated widely as a means of highlighting potential polyps in CT colonography images and thereby shortening reading times.

In a study of 41 patients with one or more polyps, Mayo Clinic investigators tested a CAD program (written by one of the investigators) that employed 360° virtual dissection to display the colon as if it had been cut lengthwise and opened fully. Two experienced radiologists reviewed only those polyps that were marked by the CAD software and categorized them as true or false positives. Each radiologist’s findings were compared with those from optical colonoscopy (the reference standard).

For mid-size polyps (6–9 mm), the CAD software’s sensitivity was 78.3% as evaluated by reviewer 1 and 91.3% as evaluated by reviewer 2; for large polyps (1 cm), sensitivities were 94.9% and 97.4%, respectively. The average number of false-positive detections by CAD was 4.28 per case, with many of these being the ileocecal valve or the enema tip. The mean reading times for the images were 4 minutes and 26 seconds (reviewer 1) and 5 minutes and 38 seconds (reviewer 2).

Comment: This investigation shows that CAD software is continuing to progress and that it might be able to reduce interobserver variation and substantially shorten CT colonography reading times. This use of CAD as a "first reader" would make interpretation of the intraluminal portion of the studies more rewarding, which suggests that gastroenterologists should continue to be involved in CT colonography interpretation.

— Douglas K. Rex, MD

Published in Journal Watch Gastroenterology November 2, 2007

Citation(s):

Johnson KT et al. Computer-aided detection (CAD) using a 360° virtual dissection: Can CAD in a first reviewer paradigm be a reliable substitute for primary 2D or 3D search? AJR Am J Roentgenol 2007 Oct; 189:W172. (http://dx.doi.org/10.2214/AJR.06.1378)

• Original article (Subscription may be required)
• Medline abstract (Free)