Purpose: We assessed the degree of interobserver and intra-observer variability in the size measurement of small renal tumors with computerized tomography.
Materials and methods: A total of 29 renal masses from 21 patients were reviewed independently by 3 radiologists on computerized tomography performed with 5 mm collimation and a reconstruction interval of 2.5 mm. Measurements were made of the largest axial dimension (X), perpendicular axial dimension (Y) and slice direction (Z) with double reads performed in a subset of patients. The predicted 95% CI was calculated for each measure, assuming different readers and the same reader.
Results: The predicted error bounds for a single renal mass measurement in the X axis were +/-3.1 mm when considering multiple readers and +/-2.3 mm for a single reader. In the X axis 78 of the 87 measurements (90%) were within 2 mm of the average measure. Smaller variances in measurements were seen with single reader repeat measurements than with multiple readers. Highest variances were seen in the Z axis. Measurements of volume and cross-sectional area showed a higher degree of variance.
Conclusions: When comparing independent computerized tomography readings done with 5 mm collimation and a 2.5 mm reconstruction interval, a size change in the largest axial dimension of a renal mass of less than 3.1 mm between different radiologists and less than 2.3 mm for the same radiologist should be viewed with caution because it is within the range of measurement variability.