Background:
Simultaneous multislice imaging(SMS) (aka multiband imaging) is the latest approach to accelerated imaging reconstruction based on the linear parallel imaging techniques, SENSE and GRAPPA, and the two reviews (1,2), provides an extensive list of literature on the background and application of the technique.
The conceptual ideas effectively predate parallel imaging for accelerated image acquisitions. The earlier approaches for multi-band imaging, used either temporal modulation (e.g. Hadamard imaging) or higher phase-encoding sampling (e.g. POMP), with the former providing a powerful temporal sampling strategy, especially when combined with processing techniques such as UNFOLD for removing temporal aliased signal.
The SMS technique as it is known today, was initially shown for leg imaging(3) and also simultaneously investigated by many researchers [http://www.ismrm.org/workshops/MultiSlice15/] who opted for a variety of reasons to not pursue the technique. The reasons for this leap in exploration can be attributed to several cohort effects, including lacking a clear applications need, the application of 3D vs. 2D techniques for large volume coverage and the need for additional reference calibration data limiting the net acceleration for anatomical imaging.
The application initially also suffered from an encoding challenge, where close simultaneous slices could not be sufficiently well separated using the spatial encoding of coil sensitivity profiles. This challenge was reduced by applying an orthogonal encoding to the conventional readout and phase encoding, whereby the FOV of adjacent slices effectively was shifted and the received signal allowed to wrap-around. This was demonstrated for GRE type imaging in the CAIPIRINHI technique(4,5), which was developed for sheared volumetric (3D) encoding, and quickly evaluated for a variety of applications (e.g. Vibhas (6). For 3D imaging, a gradient encoding was leveraged, and for the 2D equivalent imaging an RF encoding was leveraged.
For single-shot 2D-EPI imaging the RF techniques for CAIPIRINHI are not applicable, and it was recognized by Nunes et al (7) that the flexibility of using gradients orthogonal to the slice plane, could be used to perform spatial encoding of different slices; even when slices where spatially close. The approach was to add a blipped slice-encoding gradient simultaneously with the phase-encoding gradient. Working on the problem for just two weeks, a proof of concept was established but the approach suffered from reduced SNR due to increased spin dephasing across the individual slices.
Independently of the works by Nunes, the application of the blipped slice-encoding gradient simultaneously with the phase-encoding was developed by setsompop(8), using a balanced blipped approach which mostly mitigated the spin dephasing that limited the approach by Nunes, and also provided the same encoding capability when using multiple receivers. The FOV shift provides a significant universal encoding, but for maximal encoding it should not be used independently of the slice-specific RF-phase and spatial B0 distribution (9), where it was shown that the effect of suitable choice of DC phase-difference between adjacent can provide a substantial decrease in spatial noise.
With the blipped encoding being established as a suitable application, the concept of SMS 2D encoding could then be recast into a 3D encoding framework (10), enabling a better understanding of how to leverage the Cartesian encoding in a non-cartesian framework(11), and also how to use either a fully balanced blipped slice-scheme compared with random but approximately balanced blipped scheme (12).
The reconstruction of the SMS data was initially approached with a SENSE algorithm. Using the duality of image and k-space, the SENSE-GRAPPA algorithm was introduced in (13), including the subsequent concept of k-space gapping(9) to reduce the effect of edge discontinuities when using the FOV shifting between aliased slices. In parallel to these, a RO-SENSE-GRAPPA algorithm was used in (14) for absorbing effect of slice-specific eddy-currents, and when approaching the initially blipped-CAIPI GRE-EPI SMS algorithm in (8) the slice-GRAPPA algorithm was proposed as a projection reconstruction in contrast to the existing k-space interpolation algorithms. The freedom of a projection reconstruction, allowed for tailored reconstructions to correctly handle systems with slice-dependent eddy-currents; which can be detrimental for SMS-EPI. The k-space approach has also been incorporated into image-space based techniques (12,15).
With the slice-GRAPPA algorithm it was established that as a performance metric, the g-factor did not exhibit sufficient separation between different choices of encoding, and as a supplement the L-factor was introduced (16). This approach was leveraged in(17) to improve the signal fidelity, and the slice-blocking terminology was introduced. The need for slice-blocking is a creation due to the approach taken in the slice-GRAPPA algorithm, which has turned out to be a biased calibration. With an unbiased calibration, the utility of slice-blocking is cancelled, and a full equivalence between the RO-SENSE-GRAPPA and the slice-GRAPPA technique is established, including absorbing the dual-kernel and joint slice and in-plane undersampling reconstruction.
The application of SMS was as a first proof-of-principle applied to imaging of a leg (3), where the received signal from each channel was almost fully separated, similar to the benefits that can be obtained when performing bi-lateral breast imaging. The SMS-EPI was revived in (14), where again a dedicated coil-array with good inherent spatial encoding of the jointly acquired slices was used, and at 7T using a 16 channel array, a 16 fold (4 slices, and 4 fold in-plane undersampling) was feasible with acceptable g-factors and limited signal leakage using the aforementioned RO-SENSE-GRAPPA technique.
The extension to standard hardware was challenging, and simultaneously the combination of SMS with SIR (18), which temporally separates the signals from spatially close slices, and the enhanced encoding from adding FOV shifting (8,19), was published, providing two different approaches for enabling SMS with general purpose hardware – which have subsequently been combined too. The application of SIR has a lower phase-encoding bandwidth, and different SIR slices can have different TE’s depending on implementation, yet can still provide utility for high enough SMS factors (20). The SIR technique although independent of SMS, also highlights the challenge of a spatial comb- excitation versus a 3D encoding. The comb excitation in itself is not a problem, but the spin-history of adjacent slices provides new issues; challenges which 3D excitations do not have. For applications such as rsfMRI and diffusion, these can best be mitigated by choosing the ratio of total number of slices relative to the simultaneously excited slices, to be odd. For applications such as ASL, where the temporal signal properties, during the time that a volume is acquired, are significant requires that this knowledge is incorporated into existing analysis tools. Also, such concepts as background suppression in ASL imposes timing constraints on a 2D sequence that is different from the 3D sequences.
For the GRE-SMS, there are still unmeet reconstruction needs, one of which has been shown in (19). In this case, the reconstruction was not able to stably separate aliased signals in the case when there temporally is a change in the relative ratio between slices. It is unknown if the approach from (21), which used a biased slice-GRAPPA reconstruction to incorporate the spatial phase variation into the reconstruction can be used and how to do this in a stable algorithm, and whether this should be a low or high resolution approach etc.
Discussion:
Any SMS acquisition can be considered in a higher-dimensional encoding space, and reconstructions utilizing prior information can be fully incorporated into SMS reconstructions similar to 2D reconstructions, and SMS can and has been incorporated to standard acquisitions being it readout or phase-encoding segmented acquisitions, radial or spiral readouts or finger-printing excitation patterns.1. Setsompop K, Feinberg DA, Polimeni JR. Rapid brain MRI acquisition techniques at ultra-high fields. NMR Biomed 2016;29(9):1198-1221.
2. Barth M, Breuer F, Koopmans PJ, Norris DG, Poser BA. Simultaneous multislice (SMS) imaging techniques. Magn Reson Med 2016;75(1):63-81.
3. Larkman DJ, Hajnal JV, Herlihy AH, Coutts GA, Young IR, Ehnholm G. Use of multicoil arrays for separation of signal from multiple slices simultaneously excited. J Magn Reson Imaging 2001;13(2):313-317.
4. Breuer FA, Blaimer M, Heidemann RM, Mueller MF, Griswold MA, Jakob PM. Controlled aliasing in parallel imaging results in higher acceleration (CAIPIRINHA) for multi-slice imaging. Magn Reson Med 2005;53(3):684-691.
5. Breuer FA, Blaimer M, Mueller MF, Seiberlich N, Heidemann RM, Griswold MA, Jakob PM. Controlled aliasing in volumetric parallel imaging (2D CAIPIRINHA). Magn Reson Med 2006;55(3):549-556.
6. Vibhas Deshpande DN, Randall Kroeker, Stephan Kannengiesser, and Gerhard Laub. Optimized Caipirinha acceleration patterns for routine clinical 3D imaging. ISMRM 2012 2012:104.
7. RG Nunes JH, X Golay, DJ Larkman. Simultaneous slice excitation and reconstruction for single shot EPI. ISMRM 2006;2006:293.
8. Setsompop K, Gagoski BA, Polimeni JR, Witzel T, Wedeen VJ, Wald LL. Blipped-controlled aliasing in parallel imaging for simultaneous multislice Echo Planar Imaging with reduced g-factor penalty. Magn Reson Med 2012;67(5):1210-1224.
9. Blaimer M, Choli M, Jakob PM, Griswold MA, Breuer FA. Multiband phase-constrained parallel MRI. Magn Reson Med 2013;69(4):974-980.
10. Zahneisen B, Poser BA, Ernst T, Stenger VA. Three-dimensional Fourier encoding of simultaneously excited slices: generalized acquisition and reconstruction framework. Magn Reson Med 2014;71(6):2071-2081.
11. Zahneisen B, Poser BA, Ernst T, Stenger AV. Simultaneous Multi-Slice fMRI using spiral trajectories. Neuroimage 2014;92:8-18.
12. Kangrong Zhu AK, John M. Pauly. Autocalibrating CAIPIRINHA: Reformulating CAIPIRINHA as a 3D Problem. ISMRM 2012;2012:518.
13. Blaimer M, Breuer FA, Seiberlich N, Mueller MF, Heidemann RM, Jellus V, Wiggins G, Wald LL, Griswold MA, Jakob PM. Accelerated volumetric MRI with a SENSE/GRAPPA combination. J Magn Reson Imaging 2006;24(2):444-450.
14. Moeller S, Yacoub E, Olman CA, 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(5):1144-1153.
15. Koopmans PJ. Two-dimensional-NGC-SENSE-GRAPPA for fast, ghosting-robust reconstruction of in-plane and slice-accelerated blipped-CAIPI echo planar imaging. Magn Reson Med 2016.
16. Xu J, Moeller S, Auerbach EJ, Strupp J, Smith SM, Feinberg DA, Yacoub E, Ugurbil K. Evaluation of slice accelerations using multiband echo planar imaging at 3 T. Neuroimage 2013;83:991-1001.
17. Cauley SF, Polimeni JR, Bhat H, Wald LL, Setsompop K. Interslice leakage artifact reduction technique for simultaneous multislice acquisitions. Magn Reson Med 2014;72(1):93-102.
18. Feinberg DA, Moeller S, Smith SM, Auerbach E, Ramanna S, Glasser MF, Miller KL, Ugurbil K, Yacoub E. Multiplexed echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging. PLoS One;5(12):e15710.
19. Schmitter S, Moeller S, Wu X, Auerbach EJ, Metzger GJ, Van de Moortele PF, Ugurbil K. Simultaneous multislice imaging in dynamic cardiac MRI at 7T using parallel transmission. Magn Reson Med 2017;77(3):1010-1020.
20. Chen L, A TV, Xu J, Moeller S, Ugurbil K, Yacoub E, Feinberg DA. Evaluation of highly accelerated simultaneous multi-slice EPI for fMRI. Neuroimage 2015;104:452-459.
21. Steen Moeller EJA, Junqian Xu, Christophe Lenglet, kamil Ugurbil, Essa S. Yacoub. Dynamic Multiband Calibration for Improved Signal Fidelity. ISMRM 2013(2013):2661.