Purpose: To determine (1) the sensitivity for detection of small polyps with varying MR slice thicknesses using a resolution phantom; (2) reader confidence in polyp detection; and (3) image acquisition time.
Methods: A resolution phantom was created using a 3D printer. Polyp morphologies were sessile (height = diameter), flat (height = 1/2 diameter of the base), and pedunculated (stalk length = polyp diameter). Polyp diameters were 5, 7, 10, and 12 mm. Images were acquired with section thicknesses of 5, 3, and 1 mm. Images were independently reviewed by 4 board-certified radiologists who were blinded to phantom design and sequences parameters. Readers recorded maximal polyp diameter and confidence level that a polyp was present on a 1-100 point scale. Image acquisition time was also recorded.
Results: All polyps were detected by all 4 readers in the 5-mm-section thickness series. All polyps were detected by 3 readers in the 3- and 1-mm-section thickness series. The fourth reader identified 11/12 polyps in the 3- and 1-mm-section thickness series. Confidence levels were not statistically significantly different for the different section thicknesses (p = 0.28). Increasing the section thickness from 1 to 5 mm decreased image acquisition time from 3 min 54 s to 41 s.
Conclusions: Five-millimeter-section thickness was adequate for identification of 5-12 mm polyps regardless of shape. Pending further reduction in acquisition time, this prototype sequence holds promise for segmental imaging of the colon with MR colonography.