Potential challenges for application of intravoxel incoherent motion (IVIM) imaging for kidneys includes differentiation of blood perfusion-related component from glomerular filtrate. Also there is a need to consider proton relaxation which alters signal intensity in estimating local blood volume. We propose fluid-attenuated inversion recovery diffusion-weighted imaging (FLAIR DWI) to acquire IVIM imaging where the component from glomerular filtrate is eliminated. Moreover, combination of DWI and FLAIR DWI can provide renal tissue T₁. Acquisition with two echo times may provide blood T₂ which can be converted to blood oxygen saturation fraction. The purpose of this study is the use of IVIM imaging to estimate blood oxygen saturation in kidneys with exclusion of glomerular filtrate (water) fraction.

Three healthy volunteers were scanned with a 3.0T MR system (Achieva dStream, Philips Healthcare, Best, the Netherlands) with 32ch ds-Torso and 32ch ds-posterior coils. FLAIR-DWI and spin-echo (SE)-based DWI with short and long TE were imaged in coronal plane with free breathing. Basic imaging parameters were: slice thickness = 8 [mm], number of slice = 1 , acquisition matrix = 67x112 , FOV = 230 [mm], number of signal averaged=15, b=0, 250, 500 [s⋅mm^-2]. Parameters of FLAIR-DWI were: TR/ TE_short/TE_long = 5500/59/84 [ms] and TI =2000[ms]. Parameters for SE-based DWI were: TR/TE_short/TE_long = 3000/59/84 [ms]. A 3-compartment (renal tissue, blood and glomerular filtrates) model was considered¹.

Definition of parameters

Longitudinal relaxation rates: R_1t for renal tissue, R_1b = 0.58 [s^-1] for blood², R_1w = 0.234 [s^-1] for water³. Transverse relaxation rates: R_2t for renal tissue, R_2b for blood, R_2w for water. Apparent diffusion coefficients: D_t for renal tissue, D* for blood. Constants proportional to proton density: m_0t for renal tissue, m_0b for blood, m_0w for water. Signal obtained with DWI can be expressed as: S_DWI(m₀, R₂, R₁, D, K, b) = m₀ E_r2 E_b (1 + (1 – 2e^{R1 TE/2}) E_r1), where E_r2 = e^{–R2 TE}, E_b = e^{–bD + K bbDD/6}, E_r1 = e^{–R1 TR}. Signal for IRDWI: S_IRDWI(m₀, R₂, R₁, D, K, b) = m₀ E_r2 E_b (1 – 2αe^{–R1 TI} + (1 – 2α– 2 (1 – 2α) e^{R1 TE/2}) E_r1), where α is the efficiency of inversion (0.95 was assumed). Because m₀ E_r2 always appears, we define m₂ = m₀ E_r2.

We adopt a 3-compartment model so that the signal obtained with DWI can be expressed as: S_DWIb= S_DWI(m_2w, R_1w, D_w, 0, b) + S_DWI(m_2b, R_1b, D*, 0, b) + S_DWI(m_2t, R_1t, D_t, K_t, b), and that with IRDWI: S_IRDWIb = S_IRDWI(m_2w, R_1w, D_w, 0, b) + S_IRDWI(m_2b, R_1b, D*, 0, b) + S_IRDWI(m_2t, R_1t, D_t, K_t, b).

Data processing

FLAIR DW images and DW images with long TE were registered to DW images with short TE using affine transformation. Then, the following processing was performed.

1.Eight image data sets (DW images and IRDW images with short TE and long TE, b = 250[s⋅mm^-2] and 500 [s⋅mm^-2] each image) were used to calculate m₀, R_1t, R_2t and D_t; K was set as 0; signals from moving blood and glomerular filtrates were ignored at b≥250[s⋅mm^-2].

2.With parameters obtained from the above 1, 4 image data set (DW images and IRDW images with short TE and long TE, b = 0 for each image) were used to calculate m_0w, m₀b and R_2b.

3.Blood oxygen saturation was measured using a formula for τ_CPMG = 20 ms by Lu, et al⁴, where R_2b was as a function of oxygen saturation (Y) and hematocrit (Hct) at four different τ_CPMG (5, 10, 15, 20ms).

Estimation of glomerular filtrate (water) fraction. Figure 1 shows. a map and histogram of water fraction in bilateral kidneys.

Estimation of R_2b. Figure 2 shows a map and histogram of R_2b in bilateral kidneys.

Estimation of oxygen saturation (Y). Figure 3 shows a map and histogram of oxygen saturation (Y) in bilateral kidneys. Distribution of oxygen saturation indicates that renal cortex is more saturated than medulla.

Transverse relaxation time was estimated with two-dimensional DWI of two echo times (59/84 ms) although a CPMG sequence is desirable. R₂ values maybe overestimated due to the effect of diffusion on transvers relaxation especially on long-TE acquisition. Consequently, estimated oxygen saturation may be underestimated. Further study is warranted to modify protocols for better R₂ estimation. Nonetheless, the images show heterogeneous distribution of oxygen saturation between cortex and medulla, indicating its potential use for evaluating local oxygenation in normal and diseased kidneys.FLAIR DWI has a potential to estimate blood oxygen saturation with extracting glomerular filtrates fractions in kidneys.