Spiral CT colonography: reader agreement and diagnostic performance with two- and three-dimensional image-display...

Area of Science:

• Radiology and Imaging
• Gastroenterology
• Medical Diagnostics

Background:

• Computed tomographic (CT) colonography is a minimally invasive method for colorectal polyp detection.
• Two-dimensional (2D) and three-dimensional (3D) display techniques are used to interpret CT colonography images.
• Assessing the diagnostic performance and reader agreement of these techniques is crucial for clinical application.

Purpose of the Study:

• To compare the diagnostic performance of 2D and 3D display techniques in detecting colorectal polyps using spiral CT colonography.
• To evaluate reader agreement among experienced radiologists using these display techniques.

Main Methods:

• A test set of 30 colonic segments (12 with polyps, 18 without) was created from spiral CT colonographic studies.
• Twenty-two lesions (11 small, 11 large polyps or cancers) were verified by colonoscopy.
• Three experienced abdominal radiologists independently analyzed the cases using three 2D and 3D display techniques, with retesting after 6 weeks.

Main Results:

• Pooled sensitivity for polyp detection ranged from 89%-92%, with specificity from 72%-83%.
• Sensitivity and positive predictive value for polyps >/=10 mm were high (91%-100% and 85%-100%, respectively).
• Good overall intraobserver agreement (kappa, 0.60-1.00) was observed, but interobserver agreement for 2D multiplanar reformation was lower (kappa, 0.53-0.80).

Conclusions:

• 2D multiplanar reformation and 3D display techniques showed similar diagnostic performance for colorectal polyp detection among experienced radiologists.
• 3D display techniques offered improved characterization of lesions in individual cases.
• Intraobserver agreement was good, while interobserver agreement varied across the display techniques.