Anthony N Price1,2, Shaihan J Malik2, Jana Hutter2, Martin Bührer3, Lucilio Cordero-Grande2, Rui Teixeira2, Emer J Hughes1, Mary A Rutherford1, and Joseph V Hajnal1,2
1Centre for the Developing Brain, King's College London, London, United Kingdom, 2Biomedical Engineering, King's College London, London, United Kingdom, 3GyroTools, Zurich, Switzerland
Synopsis
Single-shot Turbo Spin-Echo (ss-TSE)
sequences can provide excellent anatomical images of the fetal brain. However,
due to the surrounding maternal tissue, full field of view encoding leads to
long echo train lengths, impacting efficiency, causing high SAR and increased
risk of motion artefact. In this abstract we present the implementation of multiband
accelerated ss-TSE of the fetal brain, with a zoom variant that reduces both overall
scan time and the need to encode a large FOV. In addition simultaneous sampling
multiple slice locations should benefit the image registration step in subsequent
3D slice-to-volume reconstructions.Introduction
Imaging the fetal brain in utero is
challenging due to sporadic head movements, maternal breathing, and surrounding
maternal tissue, which requires large field of view (FOV) encoding. Typically,
single shot Turbo Spin-Echo (ss-TSE) sequences are employed [1-2]. However, due
to the large FOVs required, and imposed limits on SAR, PNS and acoustic noise, TSE
shot lengths can be relatively long, and may limit realisable resolution. The
time per slice may be 100s of msec, making the acquisition rate quite slow.
Multiband accelerated TSE imaging has previously been demonstrated in the adult
brain [3-4]. Here we present initial experiences of using multiband accelerated
ss-TSE imaging in the fetus, and test using zoom TSE to allow reduced PE
encoding for shorter shot length. Combining this with multiband not only offers
scan time reduction, but should also provide data with improved slice-to-volume (SVR)
reconstructions [5], compared to conventional single-band multislice
acquisitions.
Methods
Scanner code was modified to enable multiband
modulation of RF pulses in standard TSE sequences. Additional caipi gradient
blips were added around the refocusing pulses to allow FOV/2 shifting of
adjacent simultaneous slices, in order to reduce g-factor and improve unfolding.
In addition MB zoom TSE was implemented with single band orthogonal excitation
and multiband refocusing pulses (Figure 1). The excitation pulse was modified
to maximise its bandwidth to ensure precise FOV definition and minimise slab
direction water-fat shifts, which will appear in plane in the final images. Data
was processed and reconstructed offline using ReconFrame (GyroTools, Zurich,
CH.)
Parameters common to all sequences tested were:
TE=180ms (suitable for the long T2s of fetal
brain), in-plane resolution 1.5 x 1.5 mm, slice thickness 2.5 mm, refocusing
pulse angle of 130°. The max RF b1 was adjusted to balance SAR vs
minimum echo spacing (9-13 uT). Echo train length and halfscan (partial Fourier)
factor were adjusted according to reduced (PE) FOV, SENSE factor and MB pulse
duration requirements, in order to maintain fixed TE. TR was fixed at 2 seconds
for zoom TSE, while minimised for non-zoom (multislice) TSE (SB: TR=26.6s, MB2:
TR=19.4s)
All sequences operated in low SAR (<2W/Kg) and
low PNS mode, while reduced gradient slew was used to ensure acoustic noise did
not exceed 110 dBA maximum (measured in the bore), although average sound levels
was considerably lower.
All data was acquired on a 3T Philips Achieva
system. Initial tests were performed on adult volunteers using a 32-channel
head coil, and subsequently three pregnant participants (GA range 31-34 weeks)
were scanned using a 32-channel cardiac coil. Written informed consent was
obtained prior to scanning.
Results
and Discussion
Data collected on an adult brain with the fetal
protocol (TE=180ms) reveals good image quality in the conventional full FOV sequence
and MB2 accelerated acquisition, both also employed SENSE=2 in-plane (Figure 2
a-b). In the zoom ss-TSE images (2c-d) good localised excitation allowed for
reduced PE FOV. In addition, in combination with MB2 acceleration, image
quality remains.
In the fetal example (Figure 3 a-b) full FOV
TSE with SENSE2 and MB2 shows good contrast and image quality within the fetal
brain. In the zoom implementation (3c-d) some contrast is lost due to the
significant TR reduction, but image quality is again comparable. Maternal fat
is a significant issue and needs careful consideration especially for multiband
acceleration and zoom TSE. The high bandwidth excitation pulse used in zoom MB2
(Fig. 3d) reduces shift in fat excited compared to water. This avoids FOV
aliasing (3c arrow), as produced by standard pulses which are often stretched
relative to the refocusing pulse to match bandwidth. This problem would be
exaggerated with the multiband factor.
Conclusion
Multiband accelerated TSE imaging in combination
with reduced FOV zoom excitation has been demonstrated in the fetal brain. Further
work is required to assess the extent of acceleration factor supported by the
coils and geometry involved in fetal scanning. Although here anterior-posterior
FOV reduction and fold-over was used, turning this right-left in some cases
could offer advantages in reducing fat fold-over and respiratory artefacts.
Optimisation of TE and refocusing angle control with the new reduction in TR
used in the zoom TSE may improve contrast. The use of multiband TSE to excite
simultaneous slices in the fetal brain should lead to improved SVR
reconstructions in the face of fetal motion, because multiple slices locked
in fixed relative geometry can stabilise image registration.
Acknowledgements
The authors acknowledge
funding from the MRC strategic funds, GSTT BRC and the ERC funded
dHCP.References
[1]:
Rutherford, MA. Current Opinion in Obstetrics and
Gynecology. 2009; 21(2):180–186
[2]: Yamashita, Y. AJR Am J Roentgenol. 1997;
168(2):513-9.
[3]:
Norris, DG. Magnetic Resonance in Medicine.
2014;
71(1):44-9.
[4]: Gagoski, AB. Magnetic Resonance in
Medicine. 2015; 73:929–938
[5]:
Kuklisova-Murgasova, M. Med Image Anal. 2012;
16(8): 1550–1564.