Seong Dae Yun1, Erhan Genc2, and N. Jon Shah1,3,4,5
1Institute of Neuroscience and Medicine 4, INM-4, Forschungszentrum Juelich, Juelich, Germany, 2Department of Psychology and Neurosciences, Leibniz Research Centre for Working Environment and Human Factors (IfADo), Dortmund, Germany, 3Institute of Neuroscience and Medicine 11, INM-11, JARA, Forschungszentrum Juelich, Juelich, Germany, 4JARA - BRAIN - Translational Medicine, Aachen, Germany, 5Department of Neurology, RWTH Aachen University, Aachen, Germany
Synopsis
Keywords: Pulse Sequence Design, fMRI, 7T, improved fMRI mapping accuracy, EPI, geometric-distortion correction, opposite phase-encoding and reduced acquisition time
In
high-resolution fMRI using EPI, geometric distortions typically seen in reconstructed
images significantly hinder the accurate mapping of activated voxels. One
method to correct for distortions is to acquire EPI data with an opposite
phase-encoding direction. However, this method is usually implemented with an
additional run of the same protocol, leading to a redundant measurement of
parallel imaging calibration scans. Here, we present an EPI scheme that measures
the opposite-direction data in a single fMRI session, substantially reducing
the total acquisition time. We demonstrate more accurate functional mapping with
the distortion correction in submillimetre whole-brain visual fMRI at 7T.
Introduction
The
detection of fMRI signals requires high spatiotemporal imaging to effectively
capture the dynamic haemodynamic responses and specify their activation site. As
this imaging condition is well satisfied with EPI, the method has been widely
used in numerous fMRI studies. The performance of EPI has now been far improved
by virtue of the development of hardware and imaging techniques, which allows an
fMRI acquisition with a submillimetre voxel size at ultra-high fields.1,2,3
However, the geometric deformation of tissue structures typically seen in the
EPI images significantly hinders accurate mapping of activated voxels in
high-resolution fMRI. One widely used method to correct the distortion is the
acquisition of EPI data with an opposite phase-encoding direction.4,5,6
However, this method is usually implemented by applying an additional run of
the same EPI protocol only with the change of the phase-encoding direction,
which can lead to a redundant measurement of parallel imaging (PI) calibration scan.
Therefore,
this work aims to present a novel EPI scheme that additionally measures EPI
data with the opposite phase-encoding direction in a single fMRI session. The
method substantially reduces total acquisition time, particularly for fMRI with
a large imaging matrix size for high spatial-resolution imaging and whole-brain
coverage. Methods
Figure
1a depicts a schematic diagram of a typical fMRI run using the standard EPI
scheme. Before the start of the actual fMRI scans, two preceding measurements, i.e.
calibration and dummy scans, are performed, which are necessary for the
reconstruction of acceleration techniques (e.g. PI)7,8 and for reaching
a steady-state. Here, the PI calibration scans can be acquired with EPI.
However, at ultra-high fields, they are usually acquired using a more
field-inhomogeneity tolerant sequence, such as a conventional gradient-echo, to
improve the reconstruction quality, as is the case of the present work.
In
the standard scheme, EPI with an opposite phase-encoding is performed by
applying an additional run of the same protocol with a change in the encoding
direction (Fig. 1b). However,
this strategy still requires another complete acquisition of the PI calibration
scan which makes the acquisition time and specific absorption rate (SAR)
redundant. This issue can become more severe when the PI calibration scan is
acquired with a gradient-echo sequence.
To
avoid this conflict, the proposed scheme adjusts the EPI acquisition such that
the first two EPI scans in the dummy stage are acquired with the opposite
phase-encoding direction (Fig. 1c). The feasibility of using the proposed scheme
for fMRI was verified with whole-brain, submillimetre fMRI (0.73 ×
0.73 mm2,
117 slices), designed with a visual checkerboard paradigm; the detailed imaging
configuration is described in Fig. 2a. Data sets from a healthy volunteer
screened with a standard safety procedure were acquired at a Siemens Magnetom Terra 7T scanner with a
1-Tx/32-Rx head coil.Results
Figure
2b shows reconstructed original- and opposite-phase-encoding images from the
proposed method. The different distortion characteristics of the two images
were effectively computed in ANTs (https://github.com/ANTsX/ANTs)9, and
a distortion-corrected image was generated, depicting no significant visible
loss of spatial resolution. Figure 3 shows the distortion-corrected results for
the entire slices, demonstrating reliable correction performance for all slice
locations. The distortion-corrected EPI images were co-registered to the anatomical
scan (MP2RAGE), and its co-registration accuracy was assessed in comparison to
that of the original, uncorrected EPI images. Figure 4 shows the results
presented in three different sectional views. The MP2RAGE and EPI images are
displayed alongside, each depicting the counterpart of the brain; the
background noise of the unified MP2RAGE scan was cleaned by the script.10 It can be seen that the degree of
matching between the MP2RAGE and EPI images is significantly enhanced in the distortion-corrected
case when compared to the non-corrected one. This is observed particularly
around the regions marked by yellow circles (i.e. ventricles) or red arrows.
For
the original and distortion-corrected fMRI data, the first-level analysis was individually
performed based on a GLM model using SPM12 (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/).
Figure 5 shows activated voxels obtained with an uncorrected p-value < 0.001,
which are overlaid on the co-registered MP2RAGE scan. The results at the representative
slice locations show that the activated voxels from distortion-corrected EPI
are more precisely localised along the cortical ribbon when compared to the original,
uncorrected EPI. Discussion and conclusions
This work demonstrates the use of the proposed
EPI scheme for fMRI, which acquires EPI data with the opposite phase-encoding
direction in a single fMRI session. This strategy can effectively spare another
acquisition time for the PI calibration scans, which was 95.94 seconds for the
submillimetre-protocol tested here (0.73
× 0.73 mm2,
117 slices), making the method particularly applicable for time-critical
clinical applications. The spared time can become even greater for an fMRI application employing a higher spatial resolution and a larger number of slices. The presented work also successfully verifies the EPI
distortion correction using the data obtained from the proposed scheme.
Consequently, more accurate functional mapping onto the co-registered
anatomical scan was achieved. Moreover, the proposed scheme can also open a
possibility of a more direct distortion correction of fMRI data via the online
reconstruction platform (e.g. ICE- Gadgetron11), enabling its
straightforward use in clinical diagnosis. Acknowledgements
No acknowledgement found.References
- Kay
K, Jamison KW, Vizioli L, Zhang R, Margalit E, Ugurbil K. A critical assessment
of data quality and venous effects in sub-millimeter fMRI. Neuroimage. 2019 Apr
1;189:847-869. doi: 10.1016/j.neuroimage.2019.02.006. Epub 2019 Feb 5. PMID:
30731246; PMCID: PMC7737092.
- Sharoh
D, van Mourik T, Bains LJ, Segaert K, Weber K, Hagoort P, Norris DG. Laminar
specific fMRI reveals directed interactions in distributed networks during
language processing. Proc Natl Acad Sci U S A. 2019 Oct 15;116(42):21185-21190.
doi: 10.1073/pnas.1907858116. Epub 2019 Sep 30. PMID: 31570628; PMCID:
PMC6800353.
- Yun
SD, Pais-Roldán P, Palomero-Gallagher N, Shah NJ. Mapping of whole-cerebrum
resting-state networks using ultra-high resolution acquisition protocols. Human
Brain Mapping. 2022;43(11):3386-3403. https://doi.org/10.1002/hbm.25855.
- de
Hollander G, van der Zwaag W, Qian C, Zhang P, Knapen T. Ultra-high field fMRI
reveals origins of feedforward and feedback activity within laminae of human
ocular dominance columns. Neuroimage. 2021 Mar;228:117683. doi:
10.1016/j.neuroimage.2020.117683. Epub 2020 Dec 30. PMID: 33385565.
- Navarro
KT, Sanchez MJ, Engel SA, Olman CA, Weldon KB. Depth-dependent functional MRI
responses to chromatic and achromatic stimuli throughout V1 and V2. Neuroimage.
2021 Feb 1;226:117520. doi: 10.1016/j.neuroimage.2020.117520. Epub 2020 Nov 1.
PMID: 33137474; PMCID: PMC7958868.
- Marquardt
I, De Weerd P, Schneider M, Gulban OF, Ivanov D, Wang Y, Uludağ K. Feedback
contribution to surface motion perception in the human early visual cortex.
Elife. 2020 Jun 4;9:e50933. doi: 10.7554/eLife.50933. PMID: 32496189; PMCID:
PMC7314553.
- Pruessmann
KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast
MRI. Magn Reson Med. 1999 Nov;42(5):952-62. PMID: 10542355.
- Griswold
MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J, Kiefer B, Haase A.
Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn
Reson Med. 2002 Jun;47(6):1202-10. doi: 10.1002/mrm.10171. PMID: 12111967.
- Avants
BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. A reproducible evaluation of
ANTs similarity metric performance in brain image registration. Neuroimage. 2011
Feb 1;54(3):2033-44. doi: 10.1016/j.neuroimage.2010.09.025. Epub 2010 Sep 17.
PMID: 20851191; PMCID: PMC3065962.
- https://github.com/JosePMarques/MP2RAGE-related-scripts
- https://github.com/gadgetron/gadgetron