Martin Soellradl1, Johannes Strasser1, Stefan Ropele1, and Christian Langkammer1
1Department of Neurology, Medical University of Graz, Graz, Austria
Synopsis
Intravoxel
dephasing due to macroscopic field variations along the slice-selective
direction z can be compensated by application of compensation gradients in
z-direction (“z-shimming”). Compensation gradients applied between echo
acquisition allow to estimate R2* also in areas with strong field gradients. However,
if equally strong compensation gradients are applied in each slice the signal dephases
in homogenous areas. We therefore propose an adaptive method where
slice-specific compensation gradients are estimated for each slice from a fast pre-scan.
With the proposed approach improved R2*-maps, compared to constant compensation
gradient strategies, with higher SNR and accuracy can be achieved.
Introduction
In the
presence of macroscopic field variations, the signal of a gradient echo is
largely influenced by intravoxel dephasing. Consequently, the faster signal decay
hampers quantification of tissue specific parameters such as R2*. Even post-processing
approaches1-3 cannot yield reliable parameter estimation in
areas with a strong macroscopic field gradient Gz. As a more
advanced approach, z-shimming techniques have been proposed which apply a
compensation gradient Gc to partly rephase the effect of Gz4–6. By applying a constant positive and negative Gc+/- between the acquisition of subsequent echoes, improved R2*-mapping
is possible7 (Figure 1B). However, this method applies the
same Gc+/- for each slice and thus
signal in homogenous areas with low Gz
is crushed.
Here, we
propose an adaptive approach where slice-specific compensation gradients Gc[n]+/- are
applied for the nth slice. In addition, instead of a single constantGc[n]+/- value we apply different
ratios of Gc[n]+/- (Figure
1C) to increase signal rephasing capacity. We demonstrate in phantom and
in-vivo measurements that R2*-maps can be significantly improved regarding
accuracy and SNR when compared to approaches with post-processing and
z-shimming with constant Gc+/-.Methods
The
proposed pattern of compensation gradients for the adaptive z-shimming is
illustrated in Figure 1C. There are two major differences to the z-shimming
method with constant Gc+/- (Figure 1B): First, slice specific compensation
gradients Gc[n]+/- are
estimated from a field gradient map (Gz-map) obtained from the field
map of a fast pre-scan and the compensation moments are then calculated for the ith
echo at echo time TEi within the sequence. Second, Gc[n]+/-
is split into three parts [1/3, 2/3, 3/3]Gc[n]+/- to increase the
coverage of compensation gradients. After a series of three compensation
gradients the applied moment is entirely rephased to obtain an echo with zero z-shim
moment.
Signal
modelling & implementation: For a train of compensation gradient
$$$ \bar G_{c,i}[n]$$$ applied with
duration tc, the total moment M[n]c,i after the ith echo is
given by:
$$M[n]_{c,i} = \sum_{i=1}^{TE_i} \bar G_{c,i}[n] t_c = G_{c,i}[n]TE_i $$
When
considering dephasing in presence of Gz, the signal S(TEi)
for a 2D spoiled gradient echo with compensation gradients can be modeled as7:
$$S(TE_i)=S_0 \exp(-R_2^*~TE_i) \int_{-\infty}^{\infty} \underline{M}_{xy}(z)\exp(i\gamma(G_z + G_{c,i})z~TE_i )dz$$
where $$$\underline{M}_{xy}(z) $$$
is the
complex transverse magnetization, and S0 the signal at
TE=0. Equation (2) was solved numerically10.
Phantom Measurements: The proposed method was compared
for R2* estimation with the constant z-shim (|Gc+/-=100µT/m|) approach and with post-processing only (Figure 1) in a homogenous
phantom. The cylindric phantom (Ø=12cm, length=20cm) consisted of agarose gel
(5g/l) doped with Magnevist® (110µmol/l). Besides z-shim gradients, sequence
parameter were identical (α=60°;21 echoes;TE1=2.8ms;
duration of z-shim gradient tc=2ms;TEnavi=103.4ms; monopolar readout
with bandwidth BW=500Hz/Px;TR=3s;resolution=1x1x4mm3;26 slices). The
z-shim values were estimated from pre-scan images (resolution=2x2x4mm3;3
echoes;TE1=2.7ms;TE3=6.9ms;PAT2). The raw data was corrected with the navigator
echo and combined as described in8,9. Afterwards, R2* maps were estimated by fitting the measured data to
Equation (2).
In-vivo Measurements: The proposed method was
compared with the constant z-shim (|Gc+/-|=230µT/m) and the post-processing only approach. Two subjects (age=29/32yrs)
were measured with a multi gradient echo (mGRE) sequence (α=60°;17 echoes;TE1=3ms;tc=2ms;TEnavi=64.82ms;
bipolar readout with BW=500Hz/Px;TR=2.5s;resolution=1.7x1.7x3mm3;35
slices; pre-scan: resolution=2x2x3mm³, TE1=2.7ms;TE3=6.9ms;PAT2).
Results
Figure 2 illustrates R2*-maps
estimated with post-processing only (B), constant z-shim (C) and the proposed
adaptive z-shimming (D). It clearly shows that R2* values in areas with
increasing Gz are underestimated when using only the post-processing
method. By application of constant z-shim gradients results are improved,
especially in areas where |Gz| is around |Gc|=100µT/m (e.g.
slice 5/20). In contrast, with the proposed method (D) homogenous R2* maps are
obtained in the whole phantom.
In Figure 3 images of the
signal decays for the three different sequences are illustrated (slice 11;TE14-TE16).
It shows that with the proposed approach the signal is successive rephased (arrows)
in areas with strong Gz. Whereas without z-shim gradients the signal
is already dephased in the same areas.
Comparable in-vivo R2*-maps
are shown in Figure 4. Post-processing corrected maps underestimate R2* values
in areas with strong Gz (frontal cortex, arrow in slice 11) which can
be improved by an application of Gc+/-. With the proposed
adaptive slice-specific approach the contrast and SNR improves in areas with
strong Gz (arrow in slice 11/13) and in areas with small Gz
(circles in slice 17/26).Discussion
We have developed a new method for adaptive slice-specific
z-shimming for 2D mGRE sequences. Compared to a fixed compensation gradient scheme,
the proposed approach reduces signal loss. Further, by splitting up Gc[n]+/-
into three parts (1/3, 2/3, 3/3) a broader range of field gradients Gz
is rephased (Figure 3). The proposed approach needs just a short (<25s) low-resolution pre-scan for estimating Gc[n]+/-. Interestingly, even in areas with relatively small
|Gz|<50µT/m the adaptive approach outperforms the post-processing
approach regarding SNR (Figure 3, slice 26/17).Acknowledgements
This study was funded
by the Austrian Science Fund (FWF grant numbers: KLI523, P30134)
.References
1.Fernández‐Seara MA, Wehrli FW.
Postprocessing technique to correct for background gradients in image‐based R*2
measurements. Magn Reson Med. 2000;44(3):358-366.
doi:10.1002/1522-2594(200009)44:3<358::AID-MRM3>3.0.CO;2-I
2. Preibisch C, Volz S,
Anti S, Deichmann R. Exponential excitation pulses for improved water content
mapping in the presence of background gradients. Magn Reson Med. 2008;60(4):908-916.
doi:10.1002/mrm.21730
3. Hernando D, Vigen KK, Shimakawa A,
Reeder SB. R2* mapping in the presence of macroscopic B0 field variations. Magn
Reson Med. 2012;68(3):830-840. doi:10.1002/mrm.23306
4. Frahm J, Merboldt K-D, Hänicke W.
Direct FLASH MR imaging of magnetic field inhomogeneities by gradient
compensation. Magn Reson Med. 1988;6(4):474-480.
doi:10.1002/mrm.1910060412
5.Yang QX, Williams GD, Demure RJ,
Dardzinski BJ, Mosher TJ, Smith MB. Removal of local field gradient artifacts
in GE images by slice excitation profile imaging (GESEPI) in high field. Magn
Reson Med. 1998;39(3):402-409. doi:10.1002/mrm.1910390310
6. Truong T-K, Chakeres DW, Scharre DW,
Beversdorf DQ, Schmalbrock P. Blipped multi gradient-echo slice excitation
profile imaging (bmGESEPI) for fast T2* measurements with macroscopic B0
inhomogeneity compensation. Magn Reson Med. 2006;55(6):1390-1395.
doi:10.1002/mrm.20916
7.Nam Y, Han D, Kim D-H. Single-scan R2⁎
measurement with macroscopic field inhomogeneity correction. Neuroimage.
2012;63(4):1790-1799. doi:10.1016/J.NEUROIMAGE.2012.08.062
8.Wen J, Cross AH, Yablonskiy DA. On the
role of physiological fluctuations in quantitative gradient echo MRI:
implications for GEPCI, QSM, and SWI. Magn Reson Med.
2015;73(1):195-203. doi:10.1002/mrm.25114
10.
Soellradl M, Ropele R, and Langkammer C. Gradient echo modelling with macroscopic
field variations and large flip angles. In Proceedings of the 27th Annual
Meeting of ISMRM, Montreal, 2019