2816

Impact of Autocalibration Method on Accelerated Echo-Planar Imaging of the Cervical Spinal Cord at 7 T
Alan C Seifert1,2,3 and Junqian Xu1,2,3,4

1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 2Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 3Graduate School of Biomedical Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 4Department of Neuroscience, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Synopsis

Respiration-induced B0 fluctuations are significantly greater in the cervical spinal cord than in the brain at 7T, increasing k-space phase inconsistencies and necessitating a separate evaluation of autocalibration scan (ACS) methods for accelerated EPI. We tested four ACS methods (single-shot EPI, segmented EPI, FLEET, and GRE) under three physiological conditions (end-expiration breath-hold, free-breathing, and intentional swallowing). GRE and single-shot EPI ACS methods, which are robust to respiration-induced phase errors between k-space segments, produce images with fewer and less severe artifacts than either FLEET or conventionally segmented EPI ACS methods for accelerated EPI of the cervical spinal cord at 7T.

Introduction

Maximizing temporal signal-to-noise ratio (tSNR) and minimizing image artifacts are fundamental objectives in optimizing BOLD fMRI protocols at 7T. Accelerated EPI at 7T is particularly vulnerable to k-space phase inconsistencies induced by motion or B0 fluctuation [1], during either autocalibration signal (ACS) or timeseries acquisition. For 7T brain fMRI, Polimeni et al. [2] demonstrated reduced sensitivity to motion and B0 fluctuation using a re-ordered segmented EPI ACS based on the fast low-angle excitation echo-planar technique (FLEET) [3]. However, respiration-induced B0 fluctuations (exceeding 100Hz at C7 [4]) are greater, and the number of k-space lines per slice is less, for cervical spinal cord fMRI at 7T, necessitating a separate evaluation of ACS methods.

Methods

Imaging was performed on a 7T actively-shielded whole-body scanner (Magnetom, Siemens) using a 22-channel brainstem/cervical spinal cord RF coil [5]. Axial single-shot gradient-echo (GRE)-EPI images of the cervical spinal cord were acquired spanning vertebral levels C4-C7 in a healthy 38-year-old female volunteer with a well-tested 7T cervical spinal cord fMRI protocol: 0.75x0.75x3mm3, FOV=84x84mm2, TR/TE=1500/20.8ms, flip angle=70°, phase encoding AP, GRAPPA R=2, partial Fourier=6/8, BW=1088Hz/px, echo spacing=1.04ms, coronal saturation slab anterior to the cord. Shimming was performed during free-breathing. Twelve experiments were performed to test four ACS data acquisition methods (24-line single-shot EPI, and 48-line segmented EPI, FLEET, and GRE [6], Figure 1) under three physiological conditions (end-expiration breath-hold, free-breathing, and intentional swallowing). Off-resonance correction of B0 fluctuation was applied [7]. During imaging data acquisition with normal breathing, the subject refrained from swallowing for 20 frames (30s), and then intentionally swallowed every 3-5s for the following 20 frames (30s). Image timeseries were motion-corrected (x- and y-translation) using FSL FLIRT [8]. Temporal SNR was calculated voxelwise and averaged within manually-defined masks of the spinal cord.

Results

Temporal mean images and tSNR maps of data acquired using the various ACS methods and under the various physiological conditions are shown in Figure 2-4. GRE and single-shot EPI ACS consistently produce images free from significant artifacts; tSNR is significantly greater for GRE ACS, particularly in lower slices where through-slice dephasing is most severe. Temporal SNR is plotted slicewise in Figure 5, and is tabulated in Figure 5h. GRE ACS yields the highest average tSNR=11.19 across all experiments, while single-shot EPI ACS yields the second lowest average tSNR=8.62.

Segmented EPI and FLEET ACS produce images with moderate tSNR (8.42 and 9.73, respectively), but severe artifacts in lower slices obliterate the anatomical detail of the spinal cord. These artifacts vary among physiological conditions under segmented EPI ACS, but are consistent under FLEET ACS.

Intentional swallowing during timeseries acquisition decreased tSNR for all ACS methods.

Discussion

In the cervical spinal cord, especially in lower vertebral levels, respiration-induced field shifts are a far more prominent source of image artifacts than motion. Respiration generally produces B0 fluctuations exceeding 100Hz at 7T [4] continuously throughout acquisition of ACS data, and these fluctuations occur on a similar time-scale to TR. Therefore, unlike in whole brain fMRI [2], where motion is the primary concern, severely disruptive phase errors between segments are approximately as likely in FLEET ACS as in conventionally segmented EPI ACS for cervical spinal cord fMRI; these two methods exhibit nearly identical patterns of image artifacts (Figures 2-4). The detrimental effect of these phase errors is illustrated by the fact that image artifacts are more visually apparent in images reconstructed using two 24-line segments of ACS data (segmented EPI and FLEET) than a single 24-line segment (single-shot EPI).

GRE and single-shot EPI ACS acquisition methods, on the other hand, appear largely free of image artifacts. This suggests that incoherence of phase errors across acquisition of many spin-warp GRE ACS k-space lines (with minimum TE) produces far less detrimental effects on images than phase errors of a similar magnitude but across a smaller number of shots as in segmented EPI and FLEET ACS. Although tSNR in segmented EPI and FLEET ACS exceeds that of single-shot EPI ACS, much of the image signal in segmented EPI and FLEET arises from artifacts rather than the spinal cord, confounding the tSNR measurements.

GRE ACS produces images qualitatively similar to those reconstructed using single-shot EPI ACS, with the least sensitivity to intentional swallowing during timeseries acquisition, and with the highest tSNR, particularly in lower slices where through-slice B0 gradients are strongest.

Conclusion

GRE and single-shot EPI autocalibration signal acquisition methods, which are robust to respiration-induced phase errors between k-space segments, produce images with fewer and less severe artifacts than either FLEET or conventionally segmented EPI for accelerated EPI of the cervical spinal cord at 7T.

Acknowledgements

This study was supported by National Institutes of Health (NINDS) award K01NS105160 (ACS) and Department of Defense MSRP grant W81XWH-17-1-0204.

References

[1] Triantafyllou C, Hoge RD, Krueger G, Wiggins CJ, Potthast A, Wiggins GC, et al. Comparison of physiological noise at 1.5 T, 3 T and 7 T and optimization of fMRI acquisition parameters. NeuroImage 2005;26:243-250.

[2] Polimeni JR, Bhat H, Witzel T, Benner T, Feiweier T, Inati SJ, Renvall V, Heberlein K, Wald LL. Reducing sensitivity losses due to respiration and motion in accelerated echo planar imaging by reordering the autocalibration data acquisition. Magnetic Resonance in Medicine 2016;75(2):665-79.

[3] Chapman B, Turner R, Ordidge RJ, Doyle M, Cawley M, Coxon R, Glover P, Mansfield P. Real-time movie imaging from a single cardiac cycle by NMR. Magn Reson Med 1987;5:246–54.

[4] Vannesjö SJ, Miller KL, Clare S, Tracey I. Spatiotemporal characterization of breathing-induced B0 field fluctuations in the cervical spinal cord at 7T. NeuroImage 2018;167:191-202.

[5] Zhang B, Seifert AC, Kim JW, Borrello J, Xu J. 7 Tesla 22-Channel Wrap-Around Coil Array for Cervical Spinal Cord and Brainstem Imaging. Magnetic Resonance in Medicine 2017;78(4):1623-1634.

[6] Talagala SL, Sarlls JE, Liu S, Inati SJ. Improvement of temporal signal-to-noise ratio of GRAPPA accelerated EPI using a FLASH based calibration scan. Magnetic Resonance in Medicine 2016;75(6):2362-2371.

[7] Pfeuffer J, Van de Moortele P-F, Ugurbil K, Hu X, Glover GH. Correction of physiologically induced global off-resonance effects in dynamic echo-planar and spiral functional imaging. Magnetic Resonance in Medicine 2002;47:344-353.

[8] Jenkinson M, Smith SM. A global optimisation method for robust affine registration of brain images. Medical Image Analysis, 5(2):143-156, 2001.

Figures

Figure 1: Schematic of k-space data acquisition order of single-shot EPI (a), segmented EPI (b), FLEET (c), and spin-warp GRE (d) autocalibration signal data acquisition. Single-shot EPI acquires each ACS image at a single, discrete phase of the respiratory cycle (e), while segmented EPI and FLEET acquire multiple k-space segments per imaging slice at different phases of the respiratory cycle, leading to phase discontinuity between k-space segments. GRE ACS data also contain phase errors, but these errors are distributed incoherently across a large number of k-space phase-encode lines. The imaging FOV (yellow) and shimming volume (green) are also displayed (f-h).

Figure 2: Temporal mean images and tSNR maps of timeseries data reconstructed using ACS data acquired by multiple methods during an end-expiration breath hold. Timeseries data were acquired during free-breathing.

Figure 3: Temporal mean images and tSNR maps of timeseries data reconstructed using ACS data acquired by multiple methods during free breathing. Timeseries data were also acquired during free breathing.

Figure 4: Temporal mean images of timeseries data reconstructed using ACS data acquired by multiple methods (rows) and under various physiological conditions (columns). Timeseries data were acquired during free breathing.

Figure 5: Slice-wise plots of temporal SNR of timeseries data reconstructed using ACS data acquired by multiple methods (a-d) and under various physiological conditions (e-g). Timeseries data were acquired with and without intentional swallowing during free-breathing. Average tSNR in the entire spinal cord is tabulated (h). Relative to free breathing during timeseries acquisition, swallowing during timeseries acquisition decreased GRE tSNR by 4.9%, single-shot EPI tSNR by 7.4%, FLEET tSNR by 16.5%, and segmented EPI tSNR by 23.6%.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2816