Pedro Lima Cardoso1, Gregor Körzdörfer2, Eva Hečková1, Mathias Nittka2, Siegfried Trattnig1,3, and Wolfgang Bogner1,3
1High Field MR Centre, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria, 2Siemens Healthcare GmbH, Erlangen, Germany, 3Christian Doppler Laboratory for Clinical Molecular MR Imaging, MOLIMA, Vienna, Austria
Synopsis
Motion-induced
artifacts in quantitative MRI may not be visually detectable. MRF has shown to
be robust to in-plane movement, and methods to mitigate it were made available.
However, the impact of clinically prevalent motion patterns has not yet been
investigated. The effect of realistic motion data from patients with
neurodegenerative disorders and young/elderly controls on 2D FISP-MRF is
therefore investigated in this simulation study. Results are validated with in-vivo
motion-tracked measurements. Although through-plane motions were shown to significantly
impact MRF parametric maps (particularly T2), MRF may preserve its sensitivity
in clinical cases in which expected lesion-related differences are large
enough.
Introduction
Motion artifacts in MRI are often easily discernable as blurring
or ghosting. In parametric maps derived with quantitative MRI, however, these
may manifest as an unrecognizable bias1.
Even though resilience to in-plane motion has been shown in Magnetic
Resonance Fingerprinting (MRF)2, studies have suggested that 2D-MRF
is rather sensitive to significant through-plane motion3.4.
This study investigates the impact of clinically representative
motions patterns in 2D FISP-MRF5 in a simulation setting. Results
are validated with volunteer measurements accompanied by real-time motion tracking.Methods
1D-through-plane motion was implemented in MATLAB code simulating
the behavior of a prototype 2D FISP-MRF5 sequence with a
prescan-based6 B1+ correction and spiral sampling with a spiral
angle increment of 82.5°.7 This was attained by assuming a linearly
distributed ensemble of spins aligned along the z-axis, which was displaced in
slice direction, proportional to the magnitude of the input motion. A constant
B1+ field equal to 1 was assumed for simplicity. Input T1 and T2 values for
gray (GM) and white matter (WM) were extracted from ROIs (in the frontal lobe) of
previous healthy volunteers MRF measurements (same sequence). MRF signals were
calculated using a Bloch simulation. These (motion-corrupted) complex signals
were matched to a dictionary derived using the same simulation assuming no
motion, and differences between resulting relaxation parameters and those not
affected by motion were calculated.
The set of motion data included in the 1D simulation was derived
as follows: Motions from four groups of participants (n=147) were collected in
previous studies using a sequence containing 3D EPI-based navigators for
prospective motion correction8. Participants comprised patients with
Parkinson’s Disease (PD; n=44) and Mild Cognitive Impairment (MCI; n=38) as
well as young (YOUNGCON; n=31) and old (OLDCON; n=34) controls scanned using
head restraints9. Motion logs spanned a period of ~20 minutes
(TR=1.6s) and were split into segments of 20s (i.e., the acquisition time of one
MRF slice) resulting in 22,638 motions of 20s duration. Each motion segment was
interpolated to the MRF temporal resolution using cubic splines to resemble a
smooth (realistic) transitional movement. These motions were input into the 1D
simulations (Figure 1).
For validation of the simulation results, a healthy volunteer was
scanned on a 3T whole-body scanner (MAGNETOM Vida, Siemens Healthcare,
Erlangen, Germany) using a prototype implementation of the 2D FISP-MRF5
sequence described above, with undersampling factor 48, 1500 TRs per slice, FOV
256mm, resolution 1.0x1.0x5.0mm3. The volunteer was instructed to
execute a continuous slow nodding motion around the left-right axis throughout
the acquisition of a single brain slice (~20s). Motion was recorded with an in-bore
camera (KinetiCor Inc., Honolulu, HI, USA). For comparison purposes, a scan
with no intentional motion was acquired in addition. Pure through-plane motion
was derived at the position of the ROIs (Figure 3(c)) from the motion log by
combining translation and rotation uniquely along z and input into the
simulation pipeline to determine theoretical changes in relaxation parameters. Difference
maps were calculated with reference to no motion. ROIs in GM and WM were
manually drawn to assess motion-induced differences in in-vivo parametric
measures. Changes in relaxation parameters were compared between simulation and
in-vivo measures.Results
Figure 2 displays frequency line histograms of simulated motion-induced
T1 and T2 relaxation times changes calculated for GM and WM. A low percentage
of patient motions (up to 3%) resulted in T1 changes in both GM and WM by more
than 5%. T2 relaxation times, however, showed higher susceptibility to
through-plane motion. 7% of the motions in YOUNGCON and up to 34% in the PD and
MCI groups, as well as 22-23% of the motions in OLDCON, resulted in a >10% T2
change.
As illustrated in Figure 3(b), T1 values were largely insensitive to a
through-plane nodding motion. The opposite is observed for T2 values, where
differences of >40% may be apparent with a <2mm motion amplitude. Even
though substantial large differences were observed in GM, changes in WM relaxation
times derived by simulation were in excellent agreement with ROI changes (Figure
3(d)).Discussion/Outlook
Substantial motion-induced differences in MRF parametric maps could
potentially impair detection of lesion-related alterations in clinical
populations. However, studies employing MRF e.g. in brain tumor patients have
shown that lesion-related alterations in relaxation values are greater than the
changes that could be induced by the investigated motions here10,11.
A slice thickness of 5mm was used throughout this study. However, investigation
of smaller lesions (e.g. MS) may require thinner slices (~3mm), where an even
larger susceptibility to motion is expected. In such studies, head restraints are
mandatory, but cannot prohibit through-plane motion completely.
Simulations containing synthetically derived typical motion
patterns occurring at different time points of the MRF acquisition are planned
for investigating the sequence’s temporal sensitivity to motion. Validation of
these will be attempted on a phantom with realistic brain relaxation times and
in-vivo with motion tracking.Conclusion
Clinically prevalent through-plane motions were
shown to have a significant impact in 2D FISP-MRF parametric maps (particularly
T2). However, depending on the extent of expected lesion-related changes,
MRF may preserve its sensitivity in most of the cases. Nevertheless, particularly
for studies investigating subtle quantitative changes, the development of
motion detection and correction strategies will be essential.Acknowledgements
This study was funded by the Austrian Science Fund (FWF) project KLI679.
We are grateful to Drs. Dan Ma, Yun Jiang, and Mark Griswold for sharing the Bloch simulation code.References
[1] Callaghan M. F. et al., An evaluation of prospective motion
correction (PMC) for high resolution quantitative MRI, Front Neurosci. 2015; 9: 97
[2] Ma D. et al., Magnetic resonance fingerprinting, Nature. 2013; 495: 187–192
[3] Mehta G. et al., Image reconstruction algorithm for motion
insensitive MR Fingerprinting (MRF): MORF, MRM. 2018; 80(6): 2485-2500
[4] Yu Z. et al., Exploring the sensitivity of magnetic resonance
fingerprinting to motion, MRI. 2018; 54: 241-248
[5] Jiang Y. et al., MR fingerprinting using fast imaging with
steady state precession (FISP) with spiral readout, MRM. 2015; 74(6): 1621-1631
[6] Chung S. et al., Rapid B1+ mapping using a preconditioning RF
pulse with TurboFLASH readout, MRM. 2010; 64(2): 439-446
[7] Körzdörfer G. et al., Effect of spiral undersampling patterns
on FISP MRF parameter maps, MRI. 2019; 62: 174-180
[8] Bogner W. et al., 3D GABA imaging with real-time motion
correction, shim update and reacquisition of adiabatic spiral MRSI, NeuroImage. 2014; 103: 290-302
[9] Heckova E. et al., Real-time Correction of Motion and Imager
Instability Artifacts during 3D γ-Aminobutyric Acid–edited MR Spectroscopic
Imaging, Radiology. 2018; 286(2): 666-675
[10] Badve et al., MR Fingerprinting of Adult Brain Tumors:
Initial Experience, AJNR. 2017; 38(3): 492-499
[11] De Blank et al., Magnetic Resonance Fingerprinting to
Characterize Childhood and Young Adult Brain Tumors, Pediatr. Neurosurg. 2019; 54: 310-318