Moritz Jörg Schneider^{1,2}, Thomas Gaass^{1,2}, Julien Dinkel^{1,2}, Jens Ricke^{1}, and Olaf Dietrich^{1}

In this study different intravoxel incoherent motion (IVIM) MRI methods were assessed and compared using phantom measurement. The phantom was constructed to mimic a capillary bed and allowed for the controlled application of fluid flow at varying rates. Advanced IVIM MRI methods beyond the biexponential pseudo-diffusion model were shown to be capable of accurately characterizing fluid flow inside a capillary network yielding intuitive parameters in a reproducible manner.

1. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology. 1986;161(2):401-407. doi:10.1148/radiology.161.2.3763909

2. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Laval-Jeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology. 1988;168(2):497-505. doi:10.1148/radiology.168.2.3393671

3. Wetscherek, A., Stieltjes, B. & Laun, F. B. Flow-compensated intravoxel incoherent motion diffusion imaging. Magn. Reson. Med. 74, 410–419 (2015).

4. Bellan LM, Singh SP, Henderson PW, Porri TJ, Craighead HG, Spector JA. Fabrication of an artificial 3-dimensional vascular network using sacrificial sugar structures. Soft Matter. 2009;5(7):1354. doi:10.1039/b819905a

5. Gaass T, Schneider MJ, Dietrich O, Ingrisch M, Dinkel J. Technical Note: Quantitative dynamic contrast-enhanced MRI of a 3-dimensional artificial capillary network. Med Phys. 2017;44(4):1462-1469. doi:10.1002/mp.12162

6. Bihan DL, Turner R. The capillary network: a link between ivim and classical perfusion. Magn Reson Med. 1992;27(1):171-178. doi:10.1002/mrm.1910270116

7. Frampton A, Cvetkovic V. Upscaling particle transport in discrete fracture networks: 1. Nonreactive tracers. Water Resour Res. 2007;43(10):W10428. doi:10.1029/2006WR005334

Figure
1: Exemplary microscopy image of the capillary network taken with a DM2500
(Leica, Wetzlar, Germany) optical microscope. A highly interconnected capillary
system is left by the dissolved sugar structures, strewn with spherically
shaped dilations presumably caused by residual air bubbles and sugar crystals.

Figure 2: Measured
signal intensities (different sequences color coded) at exemplary flow rates of
0.6ml/min and 1.8ml/min and the respective fits using the phase-distribution
model. In general, the signal model fits the measured data well as the fitted
curves closely follow the measurement points. At all applied flow rates, the signal
decay is slowed using the flow-compensated sequences FC11and FC18 indicating
that the pseudo-diffusion limit is not reached.

Figure
3: Estimated parameters and respective 95% confidence intervals. For flow rates
above 0.8ml/min the confidence intervals are appreciably small, with $$$f$$$
remaining largely constant and $$$t$$$ showing an inverse proportionality to
the applied flow rate. The estimated average
particle speed $$$v$$$ shows
a linear proportionality to the applied rate of flow with narrow confidence
intervals. The
average capillary segment length calculated via $$$l=vt$$$ remains approximately
constant at flow rates above 0.8ml/min and is close to the average segment
length of 162μm determined using optical microscopy (dashed line).

Figure
4: Regression analysis of $$$v$$$ (measurement series A) versus the applied flow yields a highly
significant linear proportionality. The estimated intercept at 0.034mm/s indicates that
there is little systematic bias.

Figure
5: Comparison of the signal fractions $$$f$$$ and the particle speeds $$$v$$$
estimated using the phase-distribution and biexponential model. For the
biexponential model, the particle speed was calculated via $$$v=6D^\ast/l$$$,
whereby a capillary length of $$$l$$$=195μm (average of $$$l$$$ estimated using
phase-distribution model) was used. The biexponential estimate for $$$f$$$ is
much lower than the estimate obtained using the phase-distribution model,
however, the estimates for $$$v$$$ show a reasonable agreement between the two
models.