Intraobserver and interobserver variability of renal volume measurements in polycystic kidney disease using a semiautomated MR segmentation algorithm.

OBJECTIVE Total renal volume and changes in kidney volume are markers of disease progression in autosomal-dominant polycystic kidney disease (ADPKD) but are not used in clinical practice in part because of the complexity of manual measurements. This study aims to assess the intra- and interobserver reproducibility of a semiautomated renal volumetric algorithm using fluid-sensitive MRI pulse sequences. SUBJECTS AND METHODS Renal volumes of 17 patients with ADPKD were segmented from high-resolution coronal HASTE and true fast imaging with steady-state precession (FISP) MR acquisitions. Measurements performed independently by four readers were repeated, typically after 7 days. Intraobserver agreement indexes were calculated for total kidney volume for each patient. Interobserver agreement indexes were obtained for the six paired combinations of readers as well as for two readers after rigorous formalized training. Pearson and concordance correlation coefficients, coefficients of variation (CVs), and 95% limits of agreement were determined. RESULTS The HASTE and true FISP sequences performed similarly with a median intraobserver agreement of greater than 98.1% and a CV of less than 2.4% across all readers. The median interobserver agreement was greater than 95.2% and the CV was less than 7.1%, across all reader pairs. Reader training further lowered interobserver CV. The mean total kidney volume was 1420 mL (range, 331-3782 mL) for HASTE imaging and 1445 mL (range, 301-3714 mL) for true FISP imaging, with mean image processing times per patient of 43 and 28 minutes, respectively. CONCLUSION This semiautomated MR volumetric algorithm provided excellent intraobserver and very good interobserver reproducibility using fluid-sensitive pulse sequences that emphasize cyst conspicuity.