Different strategies for MRI measurements of renal cortical volume

Abstract

Purpose: First, to measure renal cortical volume (Vc) using different MRI methods combined with a semiautomatic segmentation method. Second, to compare MRI measurements of Vc using the semiautomatic segmentation method with a manual segmentation method. Third, to evaluate the reproducibility of Vc measurements. Materials and Methods: A total of 22 pigs, 13 with healthy kidneys and nine with diseased kidneys, underwent MRI with two non-contrast-agent methods (MRI100x900 and MRI300x1100) and one contrast-agent method (MRIGd). For each of these methods, Vc was estimated using semiautomatic (Vc100x900. Vc300x1100, and Vc Gd), as well as manual (VcMANUAL) segmentation. Reference volumes were estimated by stereological means (VcSTEREOLOGY). MRI-derived volumes were compared with reference volumes using a two-tailed paired t-test, and variances between methods were visualized with Bland Alrman plots. Reproducibilities were analyzed using covariance analysis. Results: Neither measurements of Vc100x900. Vc300x1100, nor Vc Gd differed markedly from VcSTEREOLOGY using semiautomatic segmentation. Vc was comparable when semiautomatic and manual segmentations were performed at the same scan. Reproducibility of MRI estimated Vc values offered intra- and interindividual differences <5%. Conclusion: Vc can in both healthy and diseased kidneys be measured accurately with MRI using both semiautomatic and manual segmentation, and measurements can be obtained with a high reproducibility. Measurements of Vc100x900 may in the future become suitable in patients with chronic renal failure. © 2007 Wiley-Liss, Inc.