Missed lesions and false-positive findings on computed-tomographic colonography: a controlled prospective analysis.

BACKGROUND AND STUDY AIMS: The aim of the present study was to analyze the reasons for false findings on computed-tomographic (CT) colonography. PATIENTS AND METHODS: A total of 100 consecutive CT colonography examinations were carried out before conventional colonoscopies scheduled on the same day. Before the study, an experienced radiologist received training in analyzing CT colonographies. The radiologists and endoscopists were blinded to each others' findings. The patients received standard polyethylene glycol bowel preparation and were scanned in the prone and supine positions using a helical CT scanner and commercially available software for image analysis. Each pair of examinations was later followed by an unblinded analysis, comparing the CT colonographies with video recordings of the conventional colonographies in order to determine the reasons for tumors being missed or false-positive diagnoses arising on CT colonography.
RESULTS: Ninety polyps were detected in 41 patients. For patients with tumors > or = 5 mm and > or = 10 mm, the sensitivity was 67 % and 75 %, respectively, and the specificity was 84 % and 95 %, respectively. The most important reasons for the 38 false findings of tumors > or = 5 mm were perception errors (21 of 38) and misinterpretation of flat lesions in particular, including a high-grade dysplasia and a flat elevated Dukes A carcinoma. Residual stool was frequently the reason for misinterpreting lesions > or = 10 mm (four of 10).
CONCLUSIONS: Perception errors were the main reason for false findings of lesions > or = 5 mm, including one flat malignant lesion. Residual stool caused four of 10 false findings for lesions > or = 10 mm. Reading CT colonographies requires a high level of expertise, and conventional colonography is still regarded as the gold standard for detecting colorectal lesions.