Hanna Frantz1, Thomas Huefken1, Patrick Metze1, Kilian Stumpf1, Tobias Speidel1, and Volker Rasche1
1Department of Internal Medicine II, Ulm University Medical Center, Ulm, Germany
Synopsis
Zero echo
time (ZTE) imaging is an established approach for three-dimensional imaging of
tissues and materials with ultrashort T2* relaxation times. An intrinsic
disadvantage of this approach are missing data points within the center region
of k-space due to hardware limitations. These missing points are often
retrospectively interpolated or subsequently acquired using single-point
imaging approaches or additional ZTE acquisitions with decreased resolutions.
This abstract presents an approach that relies on the interleaved combination
of ZTE read-outs with different resolutions combined with a Compressed Sensing
reconstruction, to acquire more information around the k-space center, for high-quality
ZTE without prolonged acquisition times.
Introduction
Magnetic resonance imaging (MRI) of tissues or samples
with ultrashort T2* relaxation times such as e.g. lung1 or bone2
requires imaging approaches with short echo times (TE). Zero echo time (ZTE) enables
data acquisition with almost negligible TE by application of the excitation
pulse during the presence of a static read-out gradient. Since data cannot be
acquired during playing out the RF pulse, missing data points in the center of
k-space are one of the major limitations of ZTE, especially on clinical
scanners.
Several approaches have been introduced to acquire or
estimate missing data points3, including algebraic reconstruction of oversampled data, or additional acquisitions based on low-resolution ZTE or
Cartesian single-point imaging sequences. While algebraic reconstructions are
limited to dead time durations of a few Nyquist dwells4, additional
acquisitions lead to prolonged scan durations and therefore to an increased
patient discomfort.
We have introduced a concept for interfacing an
independent receive unit to a clinical scanner, thus minimizing front-end
switching times down to some µs5. However, the rather long
excitation pulses on clinical systems still limit the application of ZTE even
with T/R switch time in the µs range. The proposed method exploits the concept
of the ZTE sequence. But instead of performing additional acquisitions, every n-th
gradient is played out with decreased amplitude (lower resolution), to ensure
coverage of the k-space center. A final acquisition without gradients measures
the DC component (k0) in the steady-state. Non-linear reconstruction techniques
are used to compensate aliasing artefacts.Methods
All images were acquired with a 3 T whole-body clinical
imaging system (Ingenia 3.0T CX, Philips Healthcare, Best, The Netherlands) with
an independent spectrometer (MRI Console Cameleon, RS2D, Mundolsheim, France),
allowing for faster T/R switching, as well as for higher readout bandwidths. A
custom-built hexagonal shaped single-loop coil with a side length of 40 mm was
used for signal reception.
The proposed sequence is based on a kooshball
trajectory6 and FID sampling but modified such that each n-th read-out
gradient is executed with ¼ of its initial amplitude. The lower resolution
k-space data is used to fill the sphere around the k-space center, consisting of 2 missing k-space points in read-out direction for the given setup, which would
remain unsampled in the case of a standard ZTE acquisition. Since no additional
read-outs are added, the low-resolution gradients introduce minor undersampling
to the outer regions of k-space. The occurring artefacts can easily be
compensated by a Compressed Sensing (CS) reconstruction7.
For image reconstruction, an in-house built
reconstruction framework, implemented in MATLAB (The MathWorks, Natick,
Massachusetts, USA) was used. The k-space density functions were calculated
using a Voronoi tessellation8.
The suggested approach was evaluated on a rubber
phantom (60 x 53 mm) immersed in saline
solution and initially in-vivo, by measuring the wrists of three healthy volunteers
after obtaining written informed consent. All relevant scan parameters are
listed in table 1.Results
Figure 1 shows images of the rubber phantom, acquired
with a gradient strength of 15 mT/m, (A, B) and 24 mT/m (C, D), with (A, C) and
without (B, C) lower resolution gradients.
Figure 2 shows coronal (A,B) and sagittal (C,D) slices
of the phantom, acquired with a gradient strength of 24 mT/m, with lower
resolution gradients replacing every 25-th gradient. Images A and C were reconstructed
using gridding and a direct FFT, whereas images B and D were reconstructed
using a Compressed Sensing reconstruction combined with TV sparsity transform.
The bright area above the phantom is caused by the coil itself.
Figure 3 shows the in vivo images of a wrist, acquired
using a standard ZTE trajectory without dead time correction for a gradient
amplitude of 15 mT/m (A) and 24 mT/m (C) and with the modified approach
including the low-strength gradients for a gradient amplitude of 15 mT/m (B)
and 24 mT/m (D).
Figure 4 shows coronal (A,B) and transversal (C,D)
slices of a wrist, acquired with a gradient strength of 24 mT/m, with lower
resolution gradients replacing every 25-th gradient. Images A and C were
reconstructed using gridding and a direct FFT, whereas images B and D were
reconstructed using a Compressed Sensing reconstruction combined with TV
sparsity transform. The bright area above the wrist is caused by the coil
itself.
Adverse effects of more missing points around the
k-space center can clearly be appreciated in case of the 24 mT/m gradient
amplitude.Discussion and Conclusion
Replacing ZTE read-outs with lower resolution
gradients yields useful additional information within the center region of
k-space without any increase in total scan duration. The resulting
undersampling of the outer region of k-space does not introduce adverse
aliasing artefacts, but rather artefacts which can easily be compensated by a
CS reconstruction. The introduction of the lower resolution read-outs becomes increasingly
important with increasing gradient strength, since more k-space center points
remain unsampled. The presented approach yields qualitative ZTE images within
the same acquisition duration as a standard ZTE acquisition since no additional
acquisitions are required. The concept of lower resolution gradient can
furthermore be used to additionally acquire k0 at every m-th interleave to
derive a gating signal for e.g. lung imaging, which is a dedicated application
for ZTE imaging.Acknowledgements
The authors thank the Ulm University Center for
Translational Imaging MoMAN for its support. This work was supported by DFG
RA1660/9-1. Technical support from Philips Healthcare is gratefully acknowledged.References
-
Wild JM,
Marshall H, Bock M, et al. MRI of the lung (1/3): methods. Insights Imaging.
2012;3(4):345-53.
- Wehrli FW.
Magnetic resonance of calcified tissues. J Magn Reson. 2013;229:35-48.
- Froidevaux R, Weiger M, Brunner DO, et al. Filling the dead-time gap in zero
echt time MRI: Principles compared. Magn Reson Med. 2018;79(4):2036-2045.
- Weiger M, Brunner DO, Tabbert M, et al. Exploring the bandwidth limits of
ZTE imaging: Spatial response, out-of-band signals, and noise propagation. Magn
Reson Med. 2015;74(5):1236-47.
- Eder M,
Horneff A, Paul J, et al. A Signal Acquisition Setup for Ultrashort Echo Time
Imaging Operating in Parallel on Unmodified Clinical MRI Scanners Achieving an
Acquisition Delay of µs. IEEE Trans Med Imaging. 2020;39(1):218-225.
- Wong ST,
Roos MS. A strategy for sampling on a sphere applied to 3D selective RF pule
design. Magn Reson Med. 1994;32(6):776-84.
- Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of
compressed sensing for rapid MR imaging. Magn Reson Med. 2007;58(6):1182-95.
- Rasche V, Proksa R, Sinkus R, et al. Resampling of data between arbitrary
grids using convolution interpolation. IEEE Trans Med Imaging.
1999;18(5):385-92