Multiband Echo-Shifted EPI (MESH) fMRI
Rasim Boyacioglu1,2, Jenni Schulz1, and David G. Norris1,3

1Radboud University, Donders Institute for Brain, Cognition and Behaviour, Nijmegen, Netherlands, 2Radiology, Case Western Reserve Univesity, Cleveland, OH, United States, 3Erwin L. Hahn Institute, University Duisberg-Essen, Essen, Germany

Synopsis

Multiband Echo-Shifted EPI (MESH) is a combination of echo shifted 2D multi-slice EPI, in-plane and multiband acceleration. An additional EPI readout is inserted in the dead-time between slice selection and the multiband EPI readout. It is useful especially for low static magnetic field strengths (long optimal TE) and lower spatial resolutions (short EPI readout). It is shown that echo shifting gradients do not affect tSNR. Compared to standard and multiband EPI similar RS fMRI results are obtained at the group and individual subject level. MESH offers a further acceleration in image acquisition for fMRI at no loss in sensitivity.

Purpose

Multiband Echo-Shifted EPI (MESH) was presented; combining the principles of echo shifted acquisition for 2D multi-slice EPI, with both in-plane and multiband acceleration by means of partial parallel imaging techniques1. Here it is compared to standard 2D EPI and multiband (MB) EPI in terms of temporal stability (tSNR) and resting state (RS) fMRI.

Methods

Figure 1 illustrates the sequence diagram of MESH. The typical EPI sequence is adapted by replacing the RF pulse with a standard MB pulse which is a complex-sum of individually modulated RF pulses2,3. Blipped CAIPI4 was also incorporated to shift the slices in the FOV along the phase-encoding direction. The echo-shifting is achieved with the additional crushers and slice rephase gradients along the slice direction5 (see Figure 1). Data were collected from 6 healthy subjects at a 1.5 T Avanto scanner (Siemens Healthcare, Erlangen, Germany) with the product 32 channel head coil, after previously obtaining informed consent. Three protocols with two different effective TEs were compared in terms of tSNR (100 volumes): standard EPI, MB-EPI (MB factor 3, blipped CAIPI FOV/2 shift, no echo-shift) and MESH (MB factor 3, blipped CAIPI FOV/2 shift, Echo shift factors (ES) 1 and 2). To achieve an unbiased comparison, the TR was kept constant and the total number of slices was adjusted accordingly for each protocol. This way, the comparison is independent of square root of TR corrections and effects arising from different Ernst angles. The two echo shifting factors correspond to effective TEs of 30 and 49 ms. In the case of fMRI data the protocols with 49 ms TE were adjusted for whole brain RS acquisition (36 slices, 7 min, eyes open) with corresponding TRs and Ernst angles: EPI TR = 2030 ms & FA = 74°, MB-EPI TR = 688 ms & FA = 49° and MESH TR = 232 ms & FA = 30°. All protocols have 3 mm isotropic resolution, in plane acceleration factor of 3 and BW of 2480 Hz/Px. Reconstructions in the phase and slice directions were done in MATLAB with GRAPPA6 and Leak-Block Slice-GRAPPA7 algorithms, respectively. MELODIC (FSL, http://www.fmrib.ox.ac.uk/fsl/) was used to perform ICA on the group level RS data with 30 components and the following standard preprocessing steps: spatial smoothing (5 mm kernel), drift removal, MCFLIRT motion correction and registration to T1 weighted anatomical images. After dual regression analysis8 with the most common 8 RSNs (http://www.fmrib.ox.ac.uk/analysis/royalsoc8/) and mixture modeling to correct for inflated z-scores of low TR protocols DICE scores were calculated for all subjects and networks.

Results & Discussion

Figure 2 shows the reconstructed images and the corresponding tSNR maps for two representative subjects. Multiband lowers tSNR9 when compared to standard EPI (rows 1&2), however the additional crusher gradients used for the echo shifting do not affect tSNR (rows 2&3). Standard group level RSNs such as visual, motor, default mode and (fronto-)parietal obtained with the three protocols are shown in Figure 3. Similarity of RSN maps between the protocols indicates that functional results do not suffer from the reduced tSNR of MB and MESH. Dual regression results of DMN in Figure 4 provide the individual subject level results which are linked to the same standard space maps and thus still comparable between methods. Qualitatively it can be seen that the maps are spatially similar but with the increased average z-scores for MESH. On the other hand, DICE scores in Figure 5 quantitatively show the match between the subject level and the standard results. The scores correspond with the typically reported values in the literature10.

Conclusion

MESH is suitable for fMRI in situations where there is sufficient time to insert an additional EPI readout in the dead-time between slice selection and the multiband EPI readout. In which situation it can further accelerate data acquisition compared to standard multiband techniques. The method is particularly well suited to low static magnetic field strengths (where the optimal TE is long) and lower spatial resolutions (where the EPI readout is short). MESH provides homogeneous spatial resolution and PSF. In conclusion, MESH offers a further acceleration in image acquisition for fMRI at no loss in sensitivity.

Acknowledgements

No acknowledgement found.

References

1. Norris DG, Schulz J, Boyacioglu R. Multiband Echo-Shifted (MESH) EPI for Improved Acquisition Efficiency of T2* Weighted EPI. ISMRM Proc. 2014:2988.

2. Larkman DJ, Hajnal J V, Herlihy a H, Coutts G a, Young IR, Ehnholm G. Use of multicoil arrays for separation of signal from multiple slices simultaneously excited. J Magn Reson Imaging 2001;13:313–317.

3. Moeller S, Yacoub E, Olman C a, Auerbach E, Strupp J, Harel N, Ugurbil K. Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn Reson Med 2010;63:1144–1153.

4. Setsompop K, Gagoski BA, Polimeni JR, Witzel T, Wedeen VJ, Wald LL. Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planer imaging with reduced g-factor penalty. Magn Reson Med 2012;67:1210–1224.

5. Gibson A, Peters AM, Bowtell R. Echo-shifted multislice EPI for high-speed fMRI. Magn Reson Imaging 2006;24:433–42.

6. 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;47:1202–1210.

7. Cauley SF, Polimeni JR, Bhat H, Wald LL, Setsompop K. Interslice leakage artifact reduction technique for simultaneous multislice acquisitions. Magn Reson Med 2014;72:93–102.

8. Beckmann CF, Mackay CE, Filippini N, Smith SM. Group comparison of resting-state FMRI data using multi-subject ICA and dual regression. Neuroimage 47 (Supplement 1) 2009:OHBM, 39–41.

9. Chen L, Vu A, Xu J, Moeller S, Ugurbil K, Yacoub E, Feinberg D. Evaluation of Highly Accelerated Simultaneous Multi-Slice EPI for FMRI. Neuroimage 2015;104:452–459.

10. Boyacioglu R, Beckmann CF, Barth M. An Investigation of RSN Frequency Spectra Using Ultra-Fast Generalized Inverse Imaging. Front. Hum. Neurosci. 2013;7:156.

Figures

Figure 1. MESH sequence diagram. ‘n’ represents the echo shifting factor. Slice rephase gradients crush the echo and the crusher gradient before the RF pulse recovers the echo. Excitation is done with a multiband pulse, please note the blips in the slice direction for blipped CAIPI.

Figure 2. Reconstructed images and the corresponding tSNR maps. Please note the different maximum values set for tSNR maps of ES1 and ES2 cases. Multiband acquisition results in a tSNR drop (rows 1&2). The introduction of echo shifting does not further decrease tSNR above that introduced by multiband (rows 2&3).

Figure 3. Standard RSNs from the group level ICA for all the protocols are shown in representative slices.

Figure 4. Dual regression maps for DMN are overlaid on MNI standard space images for four representative slices. Please notice the increased z-scores for MESH. Thus, mixture modeling is crucial to estimate the true thresholds for a fair comparison.

Figure 5. DICE score results for all subjects and most commonly found 8 RSNs. Reported values are comparable with those reported in the literature.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3255