Manoj Shrestha^{1}, Ulrike Nöth^{1}, and Ralf Deichmann^{1}

In diffusion-weighted (DW) imaging with EPI readout, Nyquist ghost (NG) artifacts might be aggravated due to higher order eddy currents, especially when using monopolar DW gradients in single-refocused spin-echo EPI (srSE-EPI). Both linear and point-by-point phase corrections were tested on DW-srSE-EPI and, for comparison, also on DW twice-refocused spin-echo EPI (trSE-EPI) with intrinsic eddy-current compensation. Both phase correction methods performed equally well for DW-trSE-EPI. However, for DW-srSE-EPI with high b-values, the linear phase correction failed to fully correct NG artifacts. In contrast, point-by-point phase correction yielded considerably better results. This was confirmed in vitro and in vivo.

MRI experiments were performed on a 3T whole-body scanner (MAGNETOM Prisma, Siemens Healthineers, Erlangen, Germany), using a body TX and a 64-channel phased-array head/neck RX coil. All protocols were first tested in vitro on an agarose gel phantom, then in vivo on four healthy volunteers. Written informed consent was obtained from all participants before scanning. For each phase correction method (linear, point-by-point), DW data sets were acquired via DW SE-EPI sequences with 30 different DWG directions at a b-value of 500 s/mm^{2} (in vitro) or 1000 s/mm^{2} (in vivo). DWG directions were distributed symmetrically, using full-sphere sampling. For the phantom experiments, the diffusion-weighted twice-refocused spin-echo EPI (DW-trSE-EPI) with intrinsic eddy-current compensation (ECC) [3] and the optimized DW-srSE-EPI sequence [4, 5] were used with identical parameters except for TE: in-plane resolution=2×2 mm^{2}, FOV=192×192 mm^{2}, matrix-size=96×96, number of interleaved axial slices=72, slice-thickness=2 mm, no inter-slice gap, echo-spacing time=0.68 ms, bandwidth=1680 Hz/pixel, partial Fourier of 75%, GRAPPA acceleration factor=2. For the phantom experiments, TR=7500 ms, TE=58/70 ms (srSE/trSE) and DWG amplitude (G_{max})=38 mT/m were used. In vivo, TR/TE=6300/55 ms and G_{max}=64 mT/m were used in only DW-srSE-EPI sequence.

Online NG corrections were performed using the linear and the point-by-point approach separately. For the linear method, spatially constant and linear phase errors are obtained by a linear regression along the readout direction and subsequently removed from the EPI dataset [1]. Point-by-point phase correction does not perform a linear regression, using for each pixel along the readout direction the locally calculated correction factor [2].

In order to obtain a reference without diffusion weighting and to correct for geometrical distortions induced by static magnetic field inhomogeneities, four reference images at b=0 were acquired with positive and negative phase encoding gradients. To compute fractional anisotropy (FA) maps and principal eigenvectors, preprocessing and subsequent data analysis were performed as explained in the literature [4, 5].

- Ahn CB, Cho ZH (1987). A new phase correction method in NMR imaging based on autocorrelation and histogram analysis. IEEE Trans Med Imaging, 6(1), 32-36.
- Bruder H, Fischer H, Reinfelder HE, Schmitt F (1992). Image reconstruction for echo planar imaging with nonequidistant k‐space sampling. Magn Reson Med, 23(2), 311-323.
- Reese TG, Heid O, Weisskoff RM, Wedeen VJ (2003). Reduction of eddy-current-induced distortion in diffusion MRI using a twice-refocused spin echo. Magn Reson Med 49(1), 177-182.
- Shrestha M, Hok P, Nöth U, Deichmann R (2017). Eddy current artifact reduction in diffusion-weighted single-refocused spin-echo EPI. Proc ISMRM 25: 3353.
- Shrestha M, Hok P, Nöth U, Lienerth B, Deichmann R (2018). Optimization of diffusion-weighted single-refocused spin-echo EPI by reducing eddy-current artifacts and shortening the echo time. Magn Reson Mater Phy, 1-13.