Measurement and scan reproducibility of parameters of intravoxel incoherent motion in renal tumor and normal renal parenchyma: a preliminary research at 3.0 T MR

To prospectively estimate measurement and scan reproducibility of parameters of intravoxel incoherent motion (IVIM) in renal tumors, normal renal cortex, and medulla.

Twenty-four consecutive patients (twelve males and twelve females; median age 56.7 years, range 32–71 years) with 25 renal tumors (20 renal cell carcinomas, one urothelium carcinoma, three angiomyolipomas, and one oncocytoma) were examined twice using IVIM₁ and IVIM₂ with 9 and 16 b values, respectively, at 3.0 T. All the patients were re-scanned in 24–48 h. Regions of interest (ROIs) were placed in solid part of tumor, normal cortex, and medulla to derive IVIM parameters D (true diffusion coefficient), D* (pseudodiffusion coefficient), and f (perfusion fraction of pseudodiffusion). Differences in parameters between two IVIM sets and intra-observer, inter-observer, and scan–rescan differences were assessed using paired t tests. Intra-observer, inter-observer, and scan–rescan reproducibility were assessed by measuring coefficient of variation and Bland–Altman limits of agreements.

Intra-observer reproducibility of renal tumors, normal renal cortex, and medulla was excellent for apparent diffusion coefficient (ADC; CV: 3.45%–5.34%, BA-LA: −14% to 18%) and D (CV: 3.65% to 6.04%, BA-LA: −18% to 19%), good for f (CV: 11.96%–16.08%, BA-LA: −76.4% to 92.1% except f of medulla with CV of 32.59% and BA-LA of −76.4% to 92.1% in IVIM₁), and poor for D* (CV: 25.0% to 75.4%, BA-LA: −111% to 150%). The same order was in inter-observer reproducibility analysis. Scan–rescan reproducibility was the worst of the three parameters. Renal medulla showed worse reproducibility than renal tumors and the normal cortex. The metrics of IVIM₂ had better reproducibility than IVIM₁.

Excellent reproducibility evaluation for ADC and D, good for f, and poor for D* derived from IVIM was performed in renal tumors, normal renal cortex, and medulla. D* has limited reliability and scan–rescan reproducibility should be improved.