Background: Computed tomographic (CT) colography (virtual colonoscopy) is a new imaging method for detection of colon polyps and cancer.
Objective: To evaluate the sensitivity of CT colography for polyp detection in a population without symptoms that included persons without colon neoplasia and with radiologists blinded to colonoscopic findings.
Methods: Forty-six persons without symptoms underwent spiral CT followed by same-day colonoscopy with subsequent inspection of two-dimensional axial CT images, interactive multiplanar images, and surfaced and volume-rendered images of the colon (three-dimensional CT colography).
Results: Three-dimensional CT colography was superior to two-dimensional axial imaging for detection of colon polyps. Three-dimensional CT colography depicted 1 of 4 (25%) adenomas 2 cm in diameter or larger, 6 of 10 (60%) adenomas 1 to 1.9 cm, 6 of 14 (43%) 6 to 9 mm, and 7 of 65 (11%) 5 mm in diameter or smaller. Three-dimensional CT colography showed a polyp that might have led to colonoscopy in 3 of 4 (75%) patients whose largest adenoma was 2 cm or larger, 5 of 6 (83%) patients with largest adenoma 1 to 1.9 cm, 3 of 7 (43%) patients with largest adenoma 6 to 9 mm, and 4 of 16 patients (25%) with largest adenoma 5 mm or smaller. Large, flat adenomas of the right colon were difficult to identify with three-dimensional CT colography. The specificity of three-dimensional CT colography for patients with adenomas 1 cm in diameter or larger was 89%. Examination of patients with missed adenomas after unblinding indicated that meticulous bowel preparation and adequate distention are critical to accurate interpretation. Perceptual errors were common.
Conclusions: CT colography as performed in this study is not adequate as a colorectal cancer screening test. Several technical factors that appear critical to accurate performance of CT colography are defined.