3951

The impact of gradient spoiling on the temporal stability of rapid 2D BOLD EPI
Avery J.L. Berman1,2, Yulin E. Chang3, Jingyuan Chen1, and Jonathan R. Polimeni1,2,4

1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, United States, 2Department of Radiology, Harvard Medical School, Boston, MA, United States, 3Siemens Medical Solutions USA Inc., Charlestown, MA, United States, 4Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States

Synopsis

Short-TR BOLD fMRI studies have increased in popularity due to the advancement of simultaneous multi-slice (SMS) techniques and the many advantages afforded by rapid sampling. However, in rapid imaging of a single slice, disruptions to the steady-state from physiological sources can result in large temporal variability in long-T2 tissues, primarily around cerebrospinal fluid. We investigated the sources of this variability and how they translate from single-slice to whole-brain SMS-based fMRI. Inflow was the primary source of temporal instability, and this was significantly reduced both by the “intrinsic spoiling” from imaging gradients and by the excitation of contiguous slices, indicating that these fluctuations are largely suppressed in most conventional multiple-slice fMRI studies.

Introduction

The popularity of BOLD fMRI studies using short repetition time (TR <1 s) has steadily risen over the past several years as high sampling rates can increase statistical power,1 improve the estimation and removal of physiological noise,2,3 and enable the detection of high-frequency neural activity.4-6 As TR is decreased and becomes comparable to tissue T2, inadvertent steady-state free precession (SSFP) can be induced and the measured signal becomes a combination of the fresh FID and echoes of the transverse magnetization arising from earlier RF pulses. It has been shown that disruptions of the SSFP during rapid BOLD imaging induce significant fluctuations in tissues with long T2s, such as cerebrospinal fluid (CSF), and that these fluctuations can be reduced by applying a constant gradient spoiler after the readout;7 however, to achieve a sufficiently short TR, the acquisition was limited to a single slice. Respiratory-induced B0 fluctuations were proposed as the source of the SSFP disruption7 and the spoiler attenuated the echo formation through diffusion-weighting.8 As CSF bathes the cortex, gradient spoiling could, therefore, be an attractive method to prospectively decrease physiological noise around gray matter in fast fMRI. However, the signal variability reported in the above study may have been due to inflow effects, which were not considered, and these findings may not be as pronounced in contiguous multi-slice imaging due to i) the reduction of inflowing spins’ longitudinal magnetization from neighbouring slices’ excitations, and ii) increased “intrinsic spoiling” contributed by all other slices’ imaging gradients. Here, we investigate the role of gradient spoiling, intrinsic spoiling, and inflow on the temporal stability of rapid BOLD fMRI using single-slice, multiple-slice, and SMS imaging.

Methods

Four healthy volunteers (3 female, age=25±2-years) were scanned on a Siemens MAGNETOM Prisma 3 T system (Siemens Heathcare, Erlangen, Germany) using the manufacturer’s 32-channel head coil. All scans were based on a modification of Siemens’s product BOLD SMS-EPI sequence whereby spoiling was incorporated by adding a 24 mT/m gradient to induce dephasing by an amount Φ along the slice-direction after each readout. All scans used an isotropic resolution of 2.4 mm, 39° flip angle, TE=31 ms, and 700 repetitions.

Experiment 1: To reproduce the earlier observations,7 scans with a single excitation per repetition (“single-RF-per-repetition”) were performed with and without gradient spoiling. This included single-slice imaging and SMS=5, 5-slice imaging with a 500% slice gap. Using single-RF-per-repetition eliminated intrinsic spoiling from neighbouring slices’ excitations/readouts, and the slice gap ensured that inflow was unaffected. For these experiments, TR=450 ms, and Φ=0 or 60π (12-ms duration).

Experiment 2: To determine the impact of spoiling on the temporal stability of a typical SMS acquisition, subjects were scanned with an SMS factor of 5 or 6, acquiring 25 or 30 slices, respectively, no slice gap, TR = 400 or 450 ms, and spoiling Φ=0, 30π or 60π.

Experiment 3: To isolate the “intrinsic spoiling” imparted by the imaging gradients of other slices, the number of slices was increased from 1 through 5 using a 500% slice gap without SMS.

Experiment 4: To isolate the impact of inflow, the slice gap between 3-slice, non-SMS imaging was varied from 0% to 500%.

Data Analysis: For all experiments, the initial 10 repetitions were discarded, volumes were motion corrected (except the single-slice measurements), and each voxel’s temporal drift was removed by second-order polynomial regression. The temporal stability was assessed using the temporal signal-to-noise ratio (tSNR) and power spectral analysis.

Results

Figure 1 shows how, in the single-RF-per-repetition acquisitions, gradient spoiling played a large role in decreasing the temporal variability, particularly within the ventricles. Figure 2 shows how the temporal stability of rapid BOLD imaging, using SMS=5 or 6, in CSF was substantially higher than in the single-RF-per-repetition scans, and additional spoiling did not affect this. Figure 3 shows how applying multiple RF and imaging readouts per repetition intrinsically spoil and stabilize the signal. Figure 4 shows the impact of inflow on 3-slice imaging as the slice gap was increased from 0% to 500%.

Discussion and Conclusion

This study reproduced previous findings whereby single-slice rapid BOLD imaging showed large temporal instability in CSF.7 However, our findings indicate that inflow, rather than B0 fluctuations, is the primary source of this instability. As shown by Zhao et al.,7 our data show that gradient spoiling significantly attenuated the fluctuations, resulting in increased tSNR. Furthermore, both the intrinsic spoiling created by multiple-slice acquisitions and the inflow attenuation created by acquiring multiple contiguous slices was sufficient to significantly reduce the fluctuations. We conclude, therefore, that additional gradient spoiling provides little benefit to the temporal stability for the majority of BOLD studies that apply several RF excitations and readouts per repetition, including SMS and conventional multiple-slice fMRI.

Acknowledgements

This work was supported in part by the NIH NIBIB (grants P41-EB015896 and R01-EB019437), by the BRAIN Initiative (NIH NIMH grant R01-MH111419), and by the MGH/HST Athinoula A. Martinos Center for Biomedical Imaging.

References

1 Feinberg, D. A. et al. Multiplexed echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging. PloS one 5, e15710, doi:10.1371/journal.pone.0015710 (2010).

2 Tong, Y. & Frederick, B. D. Studying the Spatial Distribution of Physiological Effects on BOLD Signals Using Ultrafast fMRI. Frontiers in human neuroscience 8, 196, doi:10.3389/fnhum.2014.00196 (2014).

3 Narsude, M., Gallichan, D., van der Zwaag, W., Gruetter, R. & Marques, J. P. Three-dimensional echo planar imaging with controlled aliasing: A sequence for high temporal resolution functional MRI. Magn Reson Med 75, 2350-2361, doi:10.1002/mrm.25835 (2016).

4 Chen, J. E. & Glover, G. H. BOLD fractional contribution to resting-state functional connectivity above 0.1 Hz. Neuroimage 107, 207-218, doi:10.1016/j.neuroimage.2014.12.012 (2015).

5 Lewis, L. D., Setsompop, K., Rosen, B. R. & Polimeni, J. R. Fast fMRI can detect oscillatory neural activity in humans. Proceedings of the National Academy of Sciences of the United States of America 113, E6679-E6685, doi:10.1073/pnas.1608117113 (2016).

6 Lewis, L. D., Setsompop, K., Rosen, B. R. & Polimeni, J. R. Stimulus-dependent hemodynamic response timing across the human subcortical-cortical visual pathway identified through high spatiotemporal resolution 7T fMRI. Neuroimage 181, 279-291, doi:10.1016/j.neuroimage.2018.06.056 (2018).

7 Zhao, X., Bodurka, J., Jesmanowicz, A. & Li, S. J. B(0)-fluctuation-induced temporal variation in EPI image series due to the disturbance of steady-state free precession. Magn Reson Med 44, 758-765 (2000).

8 Wu, E. X. & Buxton, R. B. Effect of diffusion on the steady-state magnetization with pulsed field gradients. Journal of Magnetic Resonance (1969) 90, 243-253, doi:10.1016/0022-2364(90)90131-r (1990).

9 Setsompop, K. et al. Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planar imaging with reduced g-factor penalty. Magn Reson Med 67, 1210-1224, doi:10.1002/mrm.23097 (2012).

10 Elster, A. D. Gradient-echo MR imaging: techniques and acronyms. Radiology 186, 1-8, doi:10.1148/radiology.186.1.8416546 (1993).

Figures

Figure 1: The impact of gradient spoiling on the temporal stability of the SMS=1/1-slice (a,c) and SMS=5/5-slices (b,d) acquisitions (i.e., single-RF-per-repetition). Without spoiling, variability in the ventricles is high, as indicated by the low tSNR and high spectral power over the respiration and cardiac frequency bands. Gradient spoiling significantly reduces the variability. In the whole-brain spectra (c,d), note the broadly elevated noise between the respiratory and cardiac bands in the unspoiled signal. The shaded bands indicate the frequency ranges that were averaged across to generate the power maps in (a) and (b). Results from a single subject are shown with the slice common to both acquisitions.

Figure 2: The impact of gradient spoiling on the temporal stability of SMS-5 or SMS-6 whole-brain imaging (5 RF pulses per repetition). In both cases, the temporal stability is elevated relative to the single-RF cases (Fig. 1) but is relatively unaffected by additional spoiling, suggesting the magnetization is well-spoiled. Also, in this whole-brain acquisition with contiguous slices, inflow effects are minimal. Data from two representative subjects are shown, with similar results found in all cases.

Figure 3: Impact of “intrinsic” gradient spoiling on temporal stability. Intrinsic spoiling was increased by increasing the number of acquired slices from 1 (left) to 5 (right). The tSNR appears to have plateaued for ≥2 slices, yet the spectral power maps appear to plateau for >3 slices. In addition to the respiration and cardiac frequency bands, the mid-range band is displayed as it, too, is sensitive to the reduction in temporal variability. Applying multiple RF and imaging readouts per repetition intrinsically spoil and stabilize the signal. Only the slice that is common to all five acquisitions is displayed.

Figure 4: Impact of inflow on temporal stability. The magnitude of inflow was adjusted by increasing the slice gap of a 3-slice multiple-slice acquisition from 0% (left, lowest probability of inflow) to 20% (middle) to 500% (right, greatest probability of inflow). All four maps show that the temporal stability is greatest for the 0% gap, intermediate for the 20% gap, and lowest for the 500% gap. Despite the small gap in the 20% gap case, the tSNR and spectral power still demonstrate slightly poorer stability (evidenced by lower tSNR) around the ventricles as compared to the 0% gap, most noticeably around the anterior portion of the ventricles.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3951