CT colonography: false-negative interpretations.

PURPOSE To retrospectively evaluate if false-negative interpretations at computed tomographic (CT) colonography are due to observer error. MATERIALS AND METHODS This study was HIPAA compliant and had institutional review board approval, with waiver of informed consent. An initial unblinded review of CT colonographic image data was used to generate reconciliation reports for all false-negative polyp candidates 6.0 mm or larger. These findings were then verified by two experienced readers. After reports from the original study and reconciliation reports were reviewed, errors were classified as observer (measurement or perceptual) errors, technical errors (eg, those caused by insufficient distention, fluid), or not reconcilable. Per-polyp and per-patient sensitivity values were calculated for adenomas 6.0 mm or larger in the original data set and again by assuming elimination of technical and observer errors. RESULTS Of the original data set of 228 available polyps, 147 were adenomas; for this subgroup, the per-patient sensitivity was 70% and 68% at 10.0- and 6.0-mm thresholds, respectively. When all histologic types were considered, 114 polyps were false-negative findings. Of these, 53% (60 of 114) were attributed to observer-related errors, and 26% were attributed to errors classified as technical. After detailed retrospective reconciliation of individual polyps (so as to exclude any potentially correctable observer error), the per-polyp sensitivity of CT colonography for adenomas 10.0 mm or larger increased to 93%, and the per-patient sensitivity increased to 91%. When observer and technical errors were accounted for, eight (5.4%) of 147 adenomas 6.0 mm or larger could not be detected. If all technical errors and observer errors were scored as true-positive findings, the sensitivity for adenomas 6.0 mm or larger would have been 95% on both a per-polyp and a per-patient basis. CONCLUSION The major contributor to error at CT colonography was observer perceptual error, while observer measurement error played a smaller role.