In-vitro comparison of different slice thicknesses and kernel settings for measurement of urinary stone size by computed tomography.

Non-contrast enhanced computed tomography (NCCT) is widely used measuring stone size in patients with urolithiasis. We performed an evaluation of the accuracy of stone size measuring via NCCT. In an in-vitro study, we analyzed a total of 38 uric acid and 38 phantom stones. Within NCCT, we used different slice thicknesses (1.5 mm, 2.0 mm, and 3.0 mm) and kernel settings (bone and soft-tissue window). Maximal height, maximal length, and maximal width of each stone were measured on a picture archiving and communication system workstation. Blinded to these results, a second physician measured stone size in the same way using a caliper (real stone size). We used the Bland-Altman method for the analysis of agreement between the two measuring methods. The limit of agreement that was deemed clinical insignificant was ± 1.0 mm. All measurements via NCCT correlated significantly with the real stone size (p < 0.001). This was more pronounced for bone window and smaller slice thickness. Bland-Altman plots showed limits of agreement that exceeded the a priori defined level for all types of measurement with bone window and small slice thickness (1.5 mm) being better than soft-tissue window and large slice thickness (3.0 mm). We conclude that stone size measurement by NCCT with established settings is not exact. Stone size can easily be over- or underestimated by several millimeters. Using bone window and small slice thickness leads to more accurate results.