Joao Periquito^{1}, Katharina Paul^{1}, Till Huelnhagen^{1}, Yiyi Ji^{1}, Min-Chi Ku^{1}, Sarah Brix^{2}, Kathleen Cantow^{2}, Erdmann Seeliger^{2}, Bert Flemming^{2}, Thomas Gladytz^{3}, Dirk Grosenick^{3}, Andreas Pohlmann^{1}, and Thoralf Niendorf^{1,4}

T_{2}*
mapping does not fully represent renal tissue oxygenation. Diffusion-weighted
imaging (DWI) can provide information about confounding factors such as tubular
volume fraction, which can be used to correct T_{2}*. By
using a three compartment IVIM model, tubular volume fraction can be mapped with
DWI. The most widely used DWI technique is spin-echo EPI which is sensitive to
magnetic field inhomogeneities and hence prone to geometric distortions. In
this work we propose a diffusion-weighted Rapid Acquisition Relaxation
Enhancement (RARE) variant for DWI of the rat kidney free of geometric
distortions to quantify tubular volume fraction at 9.4 Tesla.

The imbalance between oxygen-demand and oxygen-supply
is considered to be a common cause of several kidney diseases. Blood oxygenation
sensitized MRI (T_{2}*mapping) can provide information about
changes in renal oxygenation. Yet, experiments combining MRI and invasive physiological
measurements of the kidney under (patho)physiologically relevant conditions demonstrated
that T_{2}* does not accurately represent renal-tissue oxygenation^{1,2}.
Confounding factors such as tubular volume fraction(tvf) should be taken into
account for the interpretation of renal T_{2}* mapping and
for a reliable information about renal-tissue oxygenation. The tvf is a unique
feature of the kidney, which can go up to 45% in the medulla and can rapidly
change due to alterations in filtration or tubular-outflow^{3,11}.
Diffusion-weighted imaging(DWI) provides a
method for in-vivo evaluation of
tissue-water mobility. Renal-DWI studies commonly use a mono-exponential signal
decay model which does not differentiate between water diffusion in blood and
urine. For quantification of tvf a three-compartment IVIM
(intra-voxel-incoherent-motion) approach is conceptually appealing since it allows the
separation of the true-diffusion coefficient(*D*), the intermediate-diffusion coefficient(*D _{i}**) related
to pseudo-diffusion of flowing tubular fluid and the fast-pseudo-diffusion(D

^{ }

$$S(b)=S_0*((1-f_i-f_f)*exp(-b*D)+f_i*exp(-b*D_i^*)+f_f*exp(-b*D_f^*))$$

The contribution of
the signal(*f _{i}*) coming from the
is related to tvf

^{ }

Figure 1: a) Diffusion-weighting images with some
of the acquired b-values: 0, 24, 50, 115, 300, and 800 s/mm^{2}. Image parameters: ESP=4.6ms, TR=3.000ms,
ETL=96 Matrix size= 192x192, FOV=45x45mm^{2}, resolution=0.23x0.23x1.5mm^{3}.
b) b0 diffusion-weighting image overlapped with ROIs: cortex (red), outer-medulla
(green), inner-medulla (black).

Figure 2: Mono and
tri-exponential fit of the normalized diffusion-weighted signal for each
averaged ROI: cortex (red), outer-medulla (green), inner-medulla (black) over
the following b-values:0, 4, 8, 12, 18, 24, 34, 43, 52, 75, 115, 201, 300, 460,
600 and 800s /mm^{2}

Table 1: Comparison between pure diffusion coefficients
(*D*) calculated from the mono and
tri-exponential fit for each ROI: cortex, outer-medulla and inner-medulla. Tvf
calculated from the tri-exponential model for each ROI: cortex, outer-medulla
and inner-medulla.

Figure 3: a) f_{intermediate}
map calculated from the tri-exponential model; b) R^{2} map showing
goodness of the tri-exponential fit