Diffusion-weighted imaging (DWI) has been investigated for more than two decades to detect, grade, and stage tumors across the body, e.g. [[1], [2], [3]], as a complementary or even an alternative method to invasive biopsies [4]. For example, since biopsy of the renal medulla is quite rare due to safety concerns [5], imaging might play a strong alternative role. Recently, the Intravoxel-incoherent motion (IVIM)-DWI technique [6] has been of more interest because of its potential to separate relatively substantial kidney flow effects from microstructural parameter estimates from diffusion and give additional insight into the tubular and vascular flow. Several recent reviews [[7], [8], [9]] have summarized works exploring IVIM-DWI in recent years to assess renal kidney function, acute kidney injury, chronic kidney diseases, and kidney tumors.

Most DWI acquisitions use an Echo-planar imaging (EPI) readout which, while rapid and insensitive to intershot motional phase errors, is prone to artifacts caused by eddy currents, and static field inhomogeneity [10], as well as requiring some form of respiratory motion mitigation. There are multiple solutions to treat each of these (as reviewed recently [11]), such as using bipolar gradients to minimize eddy current effects [7], using FSL TOPUP [12] to correct for magnetic field inhomogeneity using images acquired at forward and reverse phase encoding directions, and prospective triggering or retrospective registration for motion mitigation. Kidney motion quantification has also been a subject of interest in radiotherapy in pursuit of optimizing dose delivery, e.g. [13,14], where maximal craniocaudal kidney movements of around 10–16 mm have been measured with regard to the breathing phase.

Several recent studies demonstrate motion dependence of field inhomogeneity in different body organs and correct it either retrospectively or prospectively (e.g. [[15], [16], [17], [18]]). Vannesjo et al. [15] have profiled field maps of the cervical spine during breathing. For this purpose, they considered different scenarios of breath-hold (expired and inspired) and free-breathing (deep and normal) and tracked respiratory motion using respiratory sensors during the acquisition while acquiring field maps. Since their work has focused on the cervical spine, they have reduced the dimension of the profiles to only one dimension, i.e. the craniocaudal direction. Additionally, they used online motion shimming as a prospective method to correct this artifact. Hutter et al. [19] allowed more arbitrary ordering of diffusion encoding in successively acquired slices, in combination with a second reversed phase encoded image, to improve motion and distortion correction of brain DWI in the setting of fetal imaging. Andersson et al. [20] took subject rotation into account when applying distortion correction using a field map for brain DWI.

In the case of renal imaging, Coll-Font et al. [16] have retrospectively corrected magnetic field inhomogeneity distortion of renal DW-MRI acquiring near-simultaneous forward and reverse images of the same diffusion encoding with time lags of 30–50 ms allowing for correction of each image with its respective reversed phase encoding at approximately the same breathing phase and directional diffusion encoding assuming there is insignificant breathing motion within this time gap. Similar methods have been used to correct for motion dependence of magnetic inhomogeneity in the brain [17,18]. Reversed phase encoding has also been employed for distortion correction of kidney DWI in other recent studies [21,22]. Coll-Font et al. [16] discuss minor out-of-phase movement of the two kidneys which is probably caused by cardiac motion. The out-of-phase movement of the right and left kidneys has also been considered in [23] where their kidney MR processing flowchart recommends separate motion correction for the right and left kidneys. More recently, it has been shown that diffusion-weighted image quality is dependent on the kidney laterality [24]. This could be caused by a number of confounding factors, one of which might be this out-of-phase movement.

In the present work, we first corrected for motion in each kidney separately similar to the approach of [23]. Second, we introduced an alternative method to [16] for correction of magnetic field inhomogeneity for DTI sets which only required the acquisition of 32 forward and 32 reversed b-value = 0 images to sample breathing-phase-dependent motion in the kidneys and determine B0 field distortion correction separately for different breathing phases. Additionally, similar to the approach of [15] for the cervical spine, we demonstrated field inhomogeneity variations in the kidneys with regard to the breathing phase and kidney region.