Diffusion-weighted imaging maps the random motion of water molecules. Since this motion can be anisotropic, diffusion tensor imaging (DTI) can be used to detect the main direction of diffusion and to quantify its anisotropy. DTI has been applied in the kidney for tractography¹ and for the assessment of renal function after transplantation². Echo-planar imaging (EPI) is commonly used to acquire diffusion-weighted images, because it is fast in acquiring an entire image. The disadvantage of EPI is that it brings along distortion artifacts. Turbo spin echo (TSE) acquisitions require more time than EPI to acquire an entire image, but avoid distortions. The purpose of this study was to compare TSE and EPI for DTI of the human kidney.The kidneys of seven healthy volunteers were investigated at 3 T (Magnetom Prisma, Siemens Healthcare, Erlangen, Germany). Diffusion-weighted coronal images were acquired with single-shot EPI and single-shot TSE. The TSE sequence used a U-FLARE approach³ and center out k-space sampling. Respiratory triggering was applied during all acquisitions. Further protocol details for the EPI and TSE sequences are given in figure 1. Regions of interest selecting renal cortex and medulla were drawn on the unweighted (b=0 s/mm²) TSE and EPI images. Mean diffusivity (MD), fractional anisotropy (FA) and the main direction of the diffusion tensor are determined from the diffusion-weighted images. Results from EPI and TSE are compared using Wilcoxon’s signed rank test. P< 0.05 is considered statistically significant.

EPI presented well delineated, but sometimes distorted kidneys. TSE images suffered from blurring, but were virtually free of distortions in contrast to EPI. Figure 2 shows similar patterns of MD, FA and main direction of the diffusion tensor obtained from EPI and TSE images. The EPI MD map appears brighter than the TSE MD map close to the renal pelvis, suggesting higher diffusivity. This area does not contain renal cortex or medulla, but supplying vessels and major calices. Figure 3 provides an overview of MD and FA in the cohort. Statistical analysis showed no significant differences of MD between EPI and TSE, neither in the medulla [EPI: $$$(2.18 ± 0.15)×10^{-3} mm^{2}/s $$$ ; TSE: $$$(2.12 ± 0.09)×10^{-3} mm^{2}/s$$$] nor in the cortex [EPI: $$$(2.16 ± 0.10)×10^{-3} mm^{2}/s $$$ ; TSE: $$$(2.13 ± 0.05)×10^{-3} mm^{2}/s $$$].

FA was not significantly different between both methods in the cortex [EPI: $$$0.11 ± 0.01$$$ ; TSE: $$$0.10 ± 0.01$$$]. FA in the medulla was significantly smaller when acquired with TSE compared to EPI, P=0.02 [EPI: $$$0.27 ± 0.06$$$; TSE: $$$0.19 ± 0.02$$$].

Neither EPI nor TSE revealed significant corticomedullary differences for MD. Both EPI and TSE revealed significant corticomedullary difference for FA, P=0.02.

Images acquired with EPI are prone to distortions due to field inhomogeneities, e.g. at susceptibility interfaces. When imaging the kidney this may lead to distortions in vicinity of the gas-filled bowel. TSE is not affected by field inhomogeneities and therefore does not suffer from such distortions. Large first gradient moments during the TSE readout suppress the flow signal in the supplying vessels and major calices. This leads to a lower apparent MD compared to EPI close to the renal pelvis. However, TSE performs as well as EPI with respect to determining the MD in the parenchyma, i.e. in cortex and medulla. A significantly lower FA in the medulla obtained by TSE leads to slightly reduced, but still significant corticomedullary contrast in the TSE FA map. A reduction of FA when using TSE instead of EPI has already been reported by Sigmund et al⁴. They traced this reduction back to the blurring in the TSE images, which is caused by T2-decay. The similar patterns in the FA maps and in the color-coded main diffusion direction maps confirm that both methods EPI and TSE point out the same information on diffusion. TSE is a practicable alternative to EPI for DTI of the kidney. It provides the advantage of showing distortion-free images with fewer artifacts compared to EPI. Future research should focus on minimizing T2–related blurring effects in diffusion-weighted TSE.