Yuxi Pang^{1}, Dariya I Malyarenko^{1}, Michael Schar^{2}, Lisa J Wilmes^{3}, David Newitt^{3}, Michael A Jacobs^{2}, and Thomas L Chenevert^{1}

To eliminate technical variability in quantitative diffusion imaging applications, the systematic bias in diffusion weighting gradients should be corrected. The major source of this bias is the system-specific spatial gradient non-linearity (GNL) that can be rectified using gradient design information independent of the scanned object. This study characterizes the residual sources of nonuniform diffusion weighting introduced by imperfect object-dependent B0 shimming. Using controlled de-shim gradients, we show that an imperfect shim leads to systematic offsets of the otherwise symmetric GNL profile relative to the isocenter. The empiric strategies are proposed to mitigate the shim-induced errors in ADC measurements.

**Introduction**

**Methods**

**Results and Discussions**

**Conclusion**

1. Newitt DC, Tan ET, Wilmes LJ, et al, Gradient Nonlinearity Correction to Improve Apparent Diffusion Coefficient Accuracy and Standardization in the American College of Radiology Imaging Network 6698 Breast Cancer Trial, J Magn Reson Imaging. 2015 Oct; 42(4): 908–919.

2. Bammer R, Markl M, Barnett A, et.al. Analysis and generalized correction of the effect of spatial gradient field distortions in diffusion-weighted imaging. Magn Reson Med. 2003;50(3):560-9.

3. Malyarenko DI, Newitt D, Wilmes LJ, et.al. Demonstration of nonlinearity bias in the measurement of the apparent diffusion coefficient in multicenter trials. Magn Reson Med. 2016;75(3):1312-23.

FIGURE1:
B_{0}-gradients SI-profiles for Z-channel of (a) System 1 and 2
with a constant gradient slope and an offset, respectively and (b) System 3
where level of incremental de-shim gradient was controlled as indicated in the
legend. Image inserts show *b*=0 DWI that illustrate corresponding spatial
distortions of ice-water phantom images due to B_{0} inhomogeneity.

FIGURE2: The de-shimming
effect on fractional ADC bias along SI-direction of Z-gradient channel (a)
measurements at the extreme de-shims (±0.1mT/m) and reference (blue) with the vendor-provided
GNL over-plotted (note vertical and horizontal offsets for de-shimmed profiles);
(b) Residual (shim-induced) error for three de-shim settings in (a) relative to
the system GNL profile (dashed).

FIGURE3: Dependence of constant, C0, and
linear, C1, fraction-ADC shim-bias terms on de-shim gradients for 7 controlled
measurements and 2 *b*-values (correlation and P-values listed in table). The
error-bars represent ±SD for the linear least squares fit coefficients. The
linear term, C1, and its fit-errors (SD) were scaled by 10^{2} mm to
C0-scale (units of dimensionless fraction-ADC bias).

FIGURE4: ADC ROI measurements (mean ± SD) with over-plotted
GNL model fit (dashed) for Z-gradient DWI of System 1(a) SSE versus (b) DSE
acquisition and of System 2 (c) pure GNL versus (d) offset GNL model. Magenta lines
represent the level of B0 shim-induced residual bias errors (relative to system
GNL) that are evidently reduced using DSE acquisition in (b) or empiric offsets
(by -10mm and +0.03ADC units) of GNL model in (d), guided by the measured shim-gradient
profile (Fig. 1a).