Automated Renal Stone Volume Measurement by Noncontrast Computerized Tomography is More Reproducible Than Manual Linear Size Measurement

Abstract

We compared the reproducibility of automated volume and manual linear measurements using same study supine and prone, low dose, noncontrast computerized tomography series. The patient cohort comprised 50 consecutive adults with a mean age of 56.4 years in whom renal calculi were identified during computerized tomography colonography screening. The largest stone per patient was assessed with the supine and prone computerized tomography series serving as mutual controls. Automated stone volume was derived using a commercially available coronary artery calcium scoring tool. Supine-prone reproducibility for automated volume was compared with intra-observer supine-prone manual linear measurement. Interobserver variability was also assessed for manual linear measurements of the same supine or prone series. Mean ± SD linear size and volume of the 50 index calculi was 4.5 ± 2.7 mm (range 1.8 to 16) and 141.7 ± 456.1 mm(3), respectively. The mean supine-prone error for automated stone volume was 16.3% compared with an average 11.7% 1-dimensional intra-observer error for manual axial measurement. Only 2 of 15 cases with a volume error of greater than 20% were 5 mm or greater in linear size. The average interobserver linear error for the same computerized tomography series was 26.3% but automated volume measurement of the same series did not vary. Automated noncontrast computerized tomography renal stone volume is more reproducible than manual linear size measurement and it avoids the often large interobserver variability seen with manual assessment. Since small linear differences correspond to much larger volume changes, greater absolute volume errors are acceptable. Automated volume measurement may be an improved clinical parameter to use for following the renal stone burden.