The prevalence of chronic kidney disease (CKD), a continuous reduction in renal function leading to renal failure, is steadily increasing and is being recognized as a major public health problems worldwide [1]. Common end-stage manifestations of CKD are glomerulosclerosis and tubulointerstitial fibrosis, which are strong predictors of deteriorating renal function, disease progression and potential therapeutic implications [2]. In clinical practice, percutaneous renal biopsy is the reference standard to confirm pathological changes in CKD. However, it is time-consuming and invasive with the risk of complications [1]. Moreover, it is not practical for longitudinal follow-up. Therefore, it is essential to develop a non-invasive, repeatable and accurate modality to identify the renal histological changes for tailoring treatment and evaluating renal prognosis.

Diffusion-weighted imaging (DWI) is a non-invasive magnetic resonance imaging (MRI) method based on the random Brownian motion of water molecules assuming that the movement of water molecule follows a Gaussian distribution in 3 orthogonal directions [3]. It has shown promise for assessing renal function and tissue microstructures in various kidney diseases [4,5]. However, there are always concerns that the apparent diffusion coefficient (ADC) obtained from conventional mono-exponential model may not accurately reflect the real diffusivity, because of the complexity of renal structure (i.e. abundant blood flow, anisotropy of water molecules movement especially in medulla). Several advanced DWI methods have been applied to reflect the complicated alterations related to CKD. Intravoxel incoherent motion (IVIM) imaging, which adopts biexponential fitting to the diffusion-weighted images, can distinguish both pure molecular diffusion and microcirculation or capillary perfusion. Diffusion tensor imaging (DTI) allows characterization of the orientation and magnitude of the anisotropic diffusion. Diffusion kurtosis imaging (DKI) has been developed to capture the non-Gaussian diffusion behavior with reflecting the complexity of renal structure [6]. Diffusion kurtosis tensor imaging (DKTI) is a DKI model derived from DTI, taking into account that diffusion kurtosis is a tensorial quantity [7,8], which could provide both DKI and DTI metrics.

A number of studies have compared these techniques with renal function and biopsy results in patients with CKD [[9], [10], [11], [12], [13]]. However, existing findings are still controversial, potentially due to that the majority of diffusion related parameters are assessed separately with different MR scanner vendor and criteria of groups enrolled. On the other hand, some of the studies measured in the whole kidney [9,12], or in the cortex only [13].

To the best of our knowledge, this study is the first attempt to investigate the application of conventional DWI and its extended models including IVIM, DTI and DKI in CKD patients, by comparing the quantitative diffusion metrics of both cortex and medulla with histopathological results.