Maelene Lohezic1,2, Joshua van Amerom1,2, Kelly Pegoretti2, Laura McCabe2, Christina Malamateniou2, Olivia Carney2, Matthew Fox2, Joanna Allsop2, Mary Rutherford2, and Joseph Hajnal1,2
1Department of Biomedical Engineering, King's College London, London, United Kingdom, 2Centre for the Developing Brain, King's College London, London, United Kingdom
Synopsis
Fetal MRI is a growing field but it is still far from
being a standard examination and setting up a protocol demands careful
optimization. Here we explore some factors to
consider when choosing an appropriate slice acquisition order
pattern to mitigate the effect of crosstalk between temporally consecutive
slices and the partial saturation from the excitation of the adjacent slice,
while taking moderate range of fetal motion into account.Introduction
Fetal MRI is a growing field but it is still far from being
a standard examination and setting up a protocol demands its own optimization. The
use of rapid MR sequences is essential for successful fetal imaging, where both
maternal and fetal motion are present. T
2 weighted single shot Fast Spin Echo (ss-FSE)
sequences are widely used as they provide good T
2 contrast for anatomy examination
while mostly freezing intrascan motion [1]. However, visualizing the small size of
the fetal organs requires high in plane resolution, as well as large packs of
thin contiguous slices and careless choice of imaging parameters may result in
suboptimal image quality due to slice crosstalk, partial saturation or
magnetization transfer from adjacent slices [2]. Here we explore
some factors to consider when choosing an appropriate slice acquisition order pattern by
adjusting the delay between slice samplings (T
SS) as well as the
number of slice interleaves.
Methods
All
imaging was performed on a 1.5T Siemens Aera system with
the spine array removed and using two 18-element cardiac coils
closely wrapped around the maternal abdomen.The slice interleave pattern shown in Figure 1A was explored by
adjusting TSS and the number of slice interleaves. The minimum
distance between 2 consecutive slices (δ) to avoid crosstalk between
slices in the absence of fetal motion was
determined for
the sequence under test on a spectroscopy phantom provided by the vendor by acquiring ss-FSE
stacks of 21 axial slices (Table 1), with slice separation of 1 to 6 slice
thicknesses. Then, based on T1 values relevant to fetal MRI, simulations of the
recovery of the longitudinal magnetization (Mz) after full and partial
saturation (factor α) were performed to determine
the minimum time delay between 2 adjacent slices (TAS) to allow
signal recovery.
$$Mz=1-\alpha e^{-\frac{T_{AS}}{T_{1}}}$$ with $$T_{AS}=T_{SS}\frac{N_{Slices}}{N_{Interleaves}}$$
We considered stacks of 10 to 80 slices, 2 to 6 interleaves,
and TSS of 600, 1000, 1500 and 2000 ms. Finally, one pregnant woman
(gestational age: 32 weeks) was scanned at 1.5T. Two ss-FSE stacks of 40 slices
each were acquired sagittally to the fetus brain using the
imaging parameters in Table 1. The first stacks was acquired using 2
interleaves and TSS = 1500 ms. The same acquisition was then repeated with an
increased number of interleaves as determined in the previous experiments.
Results
The signal intensity in the central slice was plotted
against the slice separation (Fig. 1B). The signal intensity in the phantom was
within 2% of its maximum for a slice separation greater than 2.75, which is
satisfied with 3 or more interleaves.
Figure 1C shows the recovery
of Mz in the amniotic fluid (T
1 = 4 s) after 50% saturation. For 10 slices 98%
recovery is only obtained for the longest T
SS and 2 interleaves, which would
not satisfy the required minimum δ. When 20 slices are acquired, 3
interleaves allows 98% for the two longest T
SS. When 40 slices are acquired
(whole brain coverage), 3 or 4 interleaves allow 98% recovery for most T
ss,
while 4 to 8 interleaves can be used for larger coverage (80 slices), thus
limiting the risk of slice contamination due to fetal motion.
Figure 2 compares the images obtained on the same volunteer with
two different acquisition schemes. The
second stack of ss-FSE images was acquired with 4 interleaves. An
improvement in SNR as well as in contrast is clearly visible in the higher interleave
factor acquisition. For instance, the higher
interleave slice order allows differentiation of
the corpus callosum and contrast between the deep grey matter and the white
matter (arrows). Moreover, smaller slice separation makes the acquisition more
susceptible to motion artefact (Fig. 2 Bottom left).
Conclusion
We demonstrated that slice acquisition order has a
great effect on image quality of ss-FSE in fetal brain MRI. Other parameters
need to be taken into consideration when setting up a ss-FSE imaging protocol
such as the choice of the refocussing flip angle, that would help control
the SAR of the acquisitions to improve the scan efficiency while preserving the
patient’s safety [3]. Another confounding factor is fetal motion. We have here
only considered artefact caused by limited range of motion, which can be
mitigated by spacing out successive slices. The effect of flowing amniotic
fluid and more complex types of motion, as encountered at earlier gestational
age will require further investigation and study of the full spin history [4].
Acknowledgements
This work was supported by the iFind Project (Wellcome
Trust IEH Award 102431) and BRC.References
[1] Y Yamashita et
al. MR imaging of the fetus by a
HASTE sequence. American Journal of Roentgenology. 1997;168: 513-519
[2] Melki PS, Mulkern
RV. Magnetization
transfer effects in multislice RARE sequences. Magnetic
Resonance in Medicine. 1992; 24:189–195.
[3]
J Mugler. Optimized Three-Dimensional Fast-Spin-Echo MRI. Journal of Magnetic Resonance
Imaging. 2014; 39:745–767
[4]
G.
Ferrazzi et al. Resting State fMRI in the moving fetus: A robust framework for
motion, bias field and spin history correction.
NeuroImage. 2014; 101:555–568