Divya Varadarajan1,2, Mukund Balasubramanian2,3, Daniel J. Park1, Thomas Witzel4, Jason P. Stockmann1,2, and Jonathan R. Polimeni1,2,5
1Martinos Center for Biomedical Imaging, Charlestown, MA, United States, 2Department of Radiology, Harvard Medical School, Boston, MA, United States, 3Boston Children’s Hospital, Boston, MA, United States, 4Qbio Inc., San Carlos, CA, United States, 5Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States
Synopsis
Maps of the B0 field are routinely used for MRI scanner
calibration and for post-processing corrections of geometric distortions.
However, several sources of bias are present in conventional field estimates,
which will result in uncorrected image artifacts, yet it is unclear what the
magnitude of these biases are, whether these are large enough to warrant
concern, and how to reduce these errors to extract more accurate field
estimates. Here we investigate the accuracy of the standard B0 field map
acquisition, demonstrate that the estimated fields vary with several
acquisition parameters, and investigate sources of these errors.
Introduction
B0 fieldmaps are routinely used both for scanner calibration
to improve spatial uniformity of the B0 field and for correcting distortions
due to spatial inhomogeneity of susceptibility [1-5]. Imaging readouts such as
EPI—used in functional, diffusion, perfusion
and, increasingly, in anatomical [6], ultra-high field (7 Tesla) and clinical [7]
imaging—are particularly vulnerable to artifacts due to nonuniform B0 fields, and
conventional imaging readouts at higher resolutions. However, several sources
of bias corrupt conventional field estimates, and it is unclear what the
magnitude of these biases are, whether these are large enough to warrant concern,
and how to extract accurate field estimates. Here we investigate the accuracy
of the standard two-echo B0 fieldmap acquisition, demonstrate that the
estimated fields vary with several acquisition parameters indicating
protocol-specific biases, and investigate sources of these errors.
Specifically, we evaluate the dependence on TE, TR, flow compensation and fat
saturation, and evaluate if these artifacts are attributable to either eddy
currents or incomplete spoiling.Method
Data were acquired using a clinical 7T MRI scanner on an
anthropomorphic agar-gel phantom and on two healthy volunteers (with written
informed consent). Fieldmaps were acquired using the standard two-echo
gradient-echo-based acquisition, which acquires one line of k-space per
excitation [8].
We acquired fieldmaps after sweeping several acquisition parameters:
- Multiple TRs (211 ms, 500
ms, 850 ms, 3000 ms) with TE of 10 ms
- Multiple TEs (10 ms, 30
ms, 50 ms) with TR of 850 ms
- Repeated 1. and 2. with
reversed phase-encode (PE) and frequency-encode (FE) gradient directions.
- Reversed slice-select
(SS) gradient directions, similar to our previous method [5].
- Flow compensation on and
off.
- Fat saturation on and off
We first quantified the total bias in the fieldmap by
subtracting with the fieldmap with the largest TR and TE; these were selected references
because they are least affected by spoiling or eddy currents.
Next, we removed short-term eddy
current effects by averaging acquisitions with reversed PE/FE gradients
and quantified the TR-related bias. The residual fieldmap was then analyzed to
see if flow compensation or fat saturation setting affected the TR-related
bias. Spoiling issues were investigated via
Bloch-simulation of transverse magnetization phase immediately after
the RF pulse for TRs of 200 and 2000 ms; stable phase over time would suggest
negligible error due to incomplete spoiling. Lastly, we investigate reversing
the SS gradient to check if the eddy currents due to those gradients can
explain any remaining bias.
Results
TR variation:
Figure 1a rows 1-2 show the difference between fieldmaps at several
TRs and the 3-sec TR fieldmap, for both gradient directions. Row 3 shows
the fieldmap generated from summing the PE/FE reversed data, which cancels much
of the spatial nonuniformity, suggesting that this was due to short-term eddy
currents from these image-encoding gradients, however TR variation remains. Figures
1b-1d present the statistics of the absolute differential error. Fieldmaps show
a bias of upto 39.8 Hz at 211 ms TR, reducing to ~12 Hz after PE/FE eddy
current cancellation.
TE variation: Figure 2 shows fieldmaps across multiple TEs, relative
to the 50 ms TE fieldmap. Fig. 2a shows the reversed PE/FE polarities, while Fig.
2b shows the variation of the averaged fieldmap. All difference maps show
variation in TE which is reduced after averaging.
Flow compensation and
Fat sat: Figure 3a shows the difference maps in a phantom when flow
compensation (FC) and fat saturation (FS) are switched on and off for multiple
TRs. FC and FS difference does not show TR decreasing trend. The in-vivo
result shows bias amounting to 25–30 Hz, which aligns with our phantom measurements. The 12 Hz residual thus could not be explained by
these parameters.
Investigating spoiling
via Bloch simulations: Figure 4 plots the phase immediately after RF
excitation assuming T1=1500 ms and T2=50 ms (for brain
parenchyma at 7T), for TRs of 200 and 2000 ms. The phase stabilizes after a few
excitations, suggesting that spoiling is unlikely to cause TR-dependent
bias. Few dummy scans could remove any errors occurring in the
approach to steady state.
Slice select gradient: Lastly,
we tested whether variations are due to long-term eddy currents caused by the SS
gradient. Figure 5a. plots the reversed-polarity fieldmaps with PE, FE, and SS
directions reversed. Remaining fieldmap
nonuniformity is canceled by averaging these data. Figure 5b. shows the post-correction difference
maps with fieldmap at TR=851 ms. We observe no TR variation, mean bias of 0.5
Hz and 97th percentile bias of ~2 Hz. Therefore, we conclude that
most of the 12-Hz residual was due to long-term eddy currents from the SS gradients.Discussion and Conclusion
Here we disentangled contributors to fieldmap
inaccuracy on a modern 7T scanner. We identified both short-term and long-term
eddy current bias, with the SS gradient contributing to the TR
dependence. We ruled out incomplete spoiling as a contributor to the
errors. The presence of eddy current errors in fieldmaps suggests that either similar
gradients to the imaging acquisition of interest, such as EPI-based fieldmaps [1],
or steps similar to those performed here are necessary for accurate fieldmap
estimates. Distortion correction approaches that combine data-driven correction
with a soft fieldmap constraint may be best to compensate for low accuracy [9].Acknowledgements
We thank Mr. Kyle Droppa for help with scanning, and Drs. Yulin
Chang and John Kirsch for helpful discussions. This work was supported in part by the
NIH NIBIB (grants P41-EB030006 and R01-EB019437), the NIH NIMH (grant R01-MH124004),
by the BRAIN Initiative (NIH NIMH
grants R01-MH111419, U01-EB026996 and U01-EB025162), and by the MGH/HST
Athinoula A. Martinos Center for Biomedical Imaging; and was made
possible by the resources provided by NIH Shared Instrumentation Grant S10-OD02363701.References
[1] Reese TG, Davis TL, Weisskoff RM. Automated shimming at 1.5 T
using echo-planar image frequency maps. J Magn Reson Imaging.
1995;5(6):739-745. PMID: 8748496.
[2] Dymerska B, Poser BA, Bogner W, Visser E, Eckstein K, Cardoso
P, Barth M, Trattnig S, Robinson SD. Correcting dynamic distortions in 7T echo
planar imaging using a jittered echo time sequence. Magn Reson Med.
2016;76(5):1388-1399. PMID: 26584148.
[3] Reber PJ, Wong EC, Buxton RB, Frank LR. Correction of off
resonance-related distortion in echo-planar imaging using EPI-based field maps.
Magn Reson Med. 1998;39(2):328-330. PMID: 9469719.
[4] In M-H, Speck O. Highly accelerated PSF-mapping for EPI
distortion correction with improved fidelity. Magn Reson Mater Physics, Biol
Med. 2012;25(3):183-192. PMID: 21814756.
[5] Błażejewska AI, Witzel T, Wald LL, Polimeni JR. Correction of
EPI geometric distortion in slice direction using reversed slice-select
gradients and topup. Proc Intl Soc Mag Reson Med. 2017:1650.
[6] Skare S, Sprenger T, Norbeck O, Rydén H, Blomberg L, Avventi E,
Engström M. A 1-minute full brain MR exam using a multicontrast EPI sequence.
Magn Reson Med. 2018;79(6):3045-3054. PMID: 29090483.
[7] Prakkamakul S, Witzel T, Huang S, Boulter D, Borja MJ,
Schaefer P, Rosen B, Heberlein K, Ratai E, Gonzalez G, Rapalino O. Ultrafast
Brain MRI: Clinical Deployment and Comparison to Conventional Brain MRI at 3T.
J Neuroimaging. 2016;26(5):503-510. PMID: 27273370.
[8] Bernstein MA, King KF, Zhou XJ. Handbook of MRI Pulse
Sequences. Academic Press; 2004.
[9] Varadarajan D, Frost R, van der Kouwe A, Morgan L, Diamond B,
Boyd E, Fogarty M, Stevens A, Fischl B, Polimeni JR. Edge-preserving B0
inhomogeneity distortion correction for high-resolution multi-echo ex vivo MRI
at 7T. International Society for Magnetic Resonance in Medicine. 2020 p.664.