Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Mar 29;11:158.
doi: 10.3389/fnins.2017.00158. eCollection 2017.

Functional Sensitivity of 2D Simultaneous Multi-Slice Echo-Planar Imaging: Effects of Acceleration on G-Factor and Physiological Noise

Affiliations
Free PMC article

Functional Sensitivity of 2D Simultaneous Multi-Slice Echo-Planar Imaging: Effects of Acceleration on G-Factor and Physiological Noise

Nick Todd et al. Front Neurosci. .
Free PMC article

Abstract

Accelerated data acquisition with simultaneous multi-slice (SMS) imaging for functional MRI studies leads to interacting and opposing effects that influence the sensitivity to blood oxygen level-dependent (BOLD) signal changes. Image signal to noise ratio (SNR) is decreased with higher SMS acceleration factors and shorter repetition times (TR) due to g-factor noise penalties and saturation of longitudinal magnetization. However, the lower image SNR is counteracted by greater statistical power from more samples per unit time and a higher temporal Nyquist frequency that allows for better removal of spurious non-BOLD high frequency signal content. This study investigated the dependence of the BOLD sensitivity on these main driving factors and their interaction, and provides a framework for evaluating optimal acceleration of SMS-EPI sequences. functional magnetic resonance imaging (fMRI) data from a scenes/objects visualization task was acquired in 10 healthy volunteers at a standard neuroscience resolution of 3 mm on a 3T MRI scanner. SMS factors 1, 2, 4, and 8 were used, spanning TRs of 2800 ms to 350 ms. Two data processing methods were used to equalize the number of samples over the SMS factors. BOLD sensitivity was assessed using g-factors maps, temporal SNR (tSNR), and t-score metrics. tSNR results show a dependence on SMS factor that is highly non-uniform over the brain, with outcomes driven by g-factor noise amplification and the presence of high frequency noise. The t-score metrics also show a high degree of spatial dependence: the lower g-factor noise area of V1 shows significant improvements at higher SMS factors; the moderate-level g-factor noise area of the parahippocampal place area shows only a trend of improvement; and the high g-factor noise area of the ventral-medial pre-frontal cortex shows a trend of declining t-scores at higher SMS factors. This spatial variability suggests that the optimal SMS factor for fMRI studies is region dependent. For task fMRI studies done with similar parameters as were used here (3T scanner, 32-channel RF head coil, whole brain coverage at 3 mm isotropic resolution), we recommend SMS accelerations of 4x (conservative) to 8x (aggressive) for most studies and a more conservative acceleration of 2x for studies interested in anterior midline regions.

Keywords: BOLD sensitivity; SMS; SNR; acceleration; fMRI; high frequency noise; simultaneous multi-slice; tSNR.

Figures

Figure 1
Figure 1
High frequency content and tSNR maps for all SMS factors, averaged over all volunteers. The first column shows the level of high frequency content for each SMS factor, obtained from the full data before filtering. SMS 1 is not shown as this data does not have frequency content above 0.18 Hz. The image in top left corner shows the three ROIs for V1, PPA, and vmPFC overlaid on the 10-vounteer average T1-weighted anatomical image. The second and third columns show average tSNR maps for the eight conditions of the factorial design.
Figure 2
Figure 2
G-factor maps. (A–C) G-factor maps for SMS factors 2, 4, and 8, averaged over all volunteers. White circles indicate the outlines of the V1, PPA, and vmPFC ROIs as shown in Figure 1. (D) Bar plots of average g-factor values for each SMS factor within the three different ROIs. Mean and standard error over volunteers shown.
Figure 3
Figure 3
(A–C) Maps of percent change in tSNR due to the effect of changes in relative image SNR. The Downsampled data is used, with percent changes calculated for each SMS factor against SMS 1. White circles indicate the outlines of the V1, PPA, and vmPFC ROIs as shown in Figure 1. The inset images are the group average T1-weighted anatomical image with changes of >30% and < −30% overlaid. (D) Histogram plots of tSNR percent change over all voxels in the brain for each SMS factor.
Figure 4
Figure 4
(A–C) Bar plots of average tSNR value within an ROI for the Downsampled data, mean and standard error over all volunteers. Significant differences between conditions based on a two-tailed t-test are indicated by *p < 0.05 or **p < 0.01. (D) Percent change in the tSNR values for each SMS factor against SMS 1, plotted for each ROI.
Figure 5
Figure 5
(A–C) Maps of percent change in tSNR due to the effect of high frequency noise removal. The Decimated data is compared against the Downsamled data, with percent changes calculated for each SMS factor. White circles indicate the outlines of the V1, PPA, and vmPFC ROIs as shown in Figure 1. (D) Histogram plots of tSNR percent changes over all voxels in the brain for each SMS factor.
Figure 6
Figure 6
(A–C) Bar plots of average tSNR value within an ROI comparing the Decimated data against the Downsampled data for SMS 2, SMS 4, and SMS 8. Data presented as mean and standard error over all volunteers. Significant differences between conditions based on a two-tailed t-test are indicated by *p < 0.05 or **p < 0.01. (D) Percent change in the tSNR values for each ROI, Decimated data against Downsampled data.
Figure 7
Figure 7
(A–C) Maps of percent change in tSNR due to the combined effects in the Decimated data. Percent changes in tSNR are calculated for each SMS factor against SMS 1. White circles indicate the outlines of the V1, PPA, and vmPFC ROIs as shown in Figure 1. (D) Histogram plots of tSNR percent changes over all voxels in the brain for each SMS factor.
Figure 8
Figure 8
(A–C) Bar plots of average tSNR value within an ROI for the Decimated data, mean and standard error over all volunteers. No significant differences were observed between the conditions. (D) Percent change in the tSNR values for each SMS factor against SMS 1, plotted for each ROI.
Figure 9
Figure 9
(A–C) Bar plots of the mean of the highest 10% of t-scores within an ROI calculated from the Decimated data, mean and standard error over all volunteers. (D) Percent change in the t-score values for each SMS factor against SMS 1, plotted for each ROI. (E–G) Bar plots of the number of voxels within an ROI that had a t-score above 3.1 (corresponding to a significance level of p < 0.001, uncorrected). (H) Percent change in the number of activated voxels for each SMS factor against SMS 1, plotted for each ROI. Significant differences between conditions based on a two-tailed t-test are indicated by *p < 0.05.
Figure 10
Figure 10
Group-level t-scores overlaid on the group-average anatomical image for all SMS factors of the Decimated data. The presented contrast tested for stronger activation during scene or object trials than during baseline trials, displayed at a significance threshold of p < 0.001, uncorrected. The white outlines depict the union of all individual ROIs for V1 and the vmPFC.

Similar articles

See all similar articles

Cited by 8 articles

See all "Cited by" articles

References

    1. Ashburner J., Friston K. J. (2005). Unified segmentation. Neuroimage 26, 839–851. 10.1016/j.neuroimage.2005.02.018 - DOI - PubMed
    1. Boyacioglu R., Schulz J., Koopmans P. J., Barth M., Norris D. G. (2015). Improved sensitivity and specificity for resting state and task fMRI with multiband multi-echo EPI compared to multi-echo EPI at 7 T. Neuroimage 119, 352–361. 10.1016/j.neuroimage.2015.06.089 - DOI - PubMed
    1. Breuer F. A., Blaimer M., Heidemann R. M., Mueller M. F., Griswold M. A., Jakob P. M. (2005). Controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA) for multi-slice imaging. Magn. Reson. Med. 53, 684–691. 10.1002/mrm.20401 - DOI - PubMed
    1. Breuer F. A., Kannengiesser S. A., Blaimer M., Seiberlich N., Jakob P. M., Griswold M. A. (2009). General formulation for quantitative G-factor calculation in GRAPPA reconstructions. Magn. Reson. Med. 62, 739–746. 10.1002/mrm.22066 - DOI - PubMed
    1. Cauley S. F., Polimeni J. R., Bhat H., Wald L. L., Setsompop K. (2014). Interslice leakage artifact reduction technique for simultaneous multislice acquisitions. Magn. Reson. Med. 72, 93–102. 10.1002/mrm.24898 - DOI - PMC - PubMed

LinkOut - more resources

Feedback