Quantification of renal diffusion-weighted images using a bi-exponential model

Introduction Although well-studied in applications in the brain (1), the perfusion contribution to diffusion signal decay has not been quantified in renal applications. This effect, sometimes called “intravoxel incoherent motion (IVIM)” contributes substantially to overall signal decay in kidney, and it has been attributed to renal capillary perfusion (2). Since in kidney a large fraction (~20%) of water is filtered into tubules, we hypothesize that tubular fluid flow also contributes to IVIM. If confirmed, quantification of IVIM could lead to non-invasive assessment of renal vascular and tubular functions. In this study, we acquired volunteer DWI data with multiple diffusion weightings and multiple repetitions, and used a bi-exponential model for data analysis, with the aim of accurately quantifying the diffusion and IVIM effect in various tissues of kidney. Methods Two healthy volunteers (39 and 43 years) were scanned on a 3T MRI (Tim Trio, Siemens) with a spine coil and body array coil combination. Coronal single-shot echo planar DWI sequence were acquired using: 5 slices; slice thickness 6 mm; FOV 345×410 mm; matrix 162×192; TR/TE 2000/78 ms, 3-Scan Trace mode, parallel imaging (iPat) = 2. Diffusion imaging using 27 b values from 0 to 1300 s/mm with a 50s/mm step was performed with nine (Case 1) and two (Case 2) repetitions. Each repetition took about 2.5 minutes. The volunteers were asked to breathe shallowly. Images were co-registered using cross correlation method to correct for respiratory motion. Registered repetitions were then averaged prior to analysis. On the unweighted images (S0) the kidney was partitioned into renal cortex and medulla. On one kidney an incidental cyst was also analyzed as a separate region. For each tissue regional signals were fitted by )] exp( ) exp( ) 1 [( 0 P P D P ADC b F ADC b F S S ⋅ − ⋅ + ⋅ − ⋅ − ⋅ = [1]