Simultaneous multislice (SMS) imaging techniques

Abstract

Simultaneous multislice imaging (SMS) using parallel image reconstruction has rapidly advanced to become a major imaging technique. The primary benefit is an acceleration in data acquisition that is equal to the number of simultaneously excited slices. Unlike in-plane parallel imaging this can have only a marginal intrinsic signal-to-noise ratio penalty, and the full acceleration is attainable at fixed echo time, as is required for many echo planar imaging applications. Furthermore, for some implementations SMS techniques can reduce radiofrequency (RF) power deposition. In this review the current state of the art of SMS imaging is presented. In the Introduction, a historical overview is given of the history of SMS excitation in MRI. The following section on RF pulses gives both the theoretical background and practical application. The section on encoding and reconstruction shows how the collapsed multislice images can be disentangled by means of the transmitter pulse phase, gradient pulses, and most importantly using multichannel receiver coils. The relationship between classic parallel imaging techniques and SMS reconstruction methods is explored. The subsequent section describes the practical implementation, including the acquisition of reference data, and slice cross-talk. Published applications of SMS imaging are then reviewed, and the article concludes with an outlook and perspective of SMS imaging.

Keywords: fast imaging; multiband imaging; simultaneous multislice imaging.

Figure 1

First ever in vivo SMS images obtained from the leg of a healthy volunteer using a four element coil array. The top row shows the images obtained from each coil, the bottom row shows the disentangled slices using a SENSE reconstruction. Figure taken with kind permission from Larkman et al 12.

Figure 2

PINS, multiband and multi‐PINS RF pulses. The RF envelope, its accompanying gradient and the slice profile for a PINS pulse (a,f) and a conventional multiband pulse (b,g). Corresponding k‐space trajectories are shown in (c). Multi‐PINS works by summing a PINS and a multiband pulse, but as they have different gradients VERSE needs to be applied to the multiband pulse to allow it to operate with the blipped gradients (d). Finally, adding them results in the multi‐PINS pulse in (e) with (h) the resulting slice profile. Figure taken with kind permission from Eichner et al 40.

Figure 3

Cross‐talk saturation in SMS imaging. The schematic shows a series of interleaved nonideal multiband slice profiles that overlap slightly due to BWTP limitations. The numbers inside the slices indicate the index of each slice group, the top row shows the first excitation (t = 0) with time being displayed vertically downward. For a slice that is about to be excited, the colored rectangles indicate when its transition bands have last been partially saturated by the neighboring slice profiles. The colorbar on the right indicates the number of TRs that each color represents. The left part of the figure shows the schematic corresponding to an even number of interleaved slice groups, the right shows the same for odd‐numbered excitation schemes. The scheme on the left is highly inhomogeneous with especially severe saturation effects in groups 1 and 6. These effects are nonexistent in the scheme on the right.

Figure 4

Analogy of phase cycled CAIPIRINHA and blipped CAIPIRINHA: 4 Slices are simultaneously excited. a: No phase cycling causes the slices to overlap directly on top of each other. This corresponds to a standard SMS EPI acquisition without using gradient blips along the z direction. b: 180° phase cycling used for slice 2 and 4. Slice 2 and 4 appear shifted by FOV/2 in the FOV. The same aliasing pattern can be realized by using alternating gradient blips along the z direction. The gradient moment of the blips must be chosen such that Δ = 2Δk_z. c: All four slices are shifted by different amounts (0,FOV/4,FOV/2,3/4FOV). By alternating between four different multiband pulses for subsequent phase encoding steps different phase cycles may be imposed for each individual slice. Alternatively, gradient blips may be used on the z‐axes such that subsequent phase encoding steps accumulate a phase according to 1Δk_z. To avoid spin dephasing over the slices the accumulated gradient moments are rephased accordingly. All the sampling schemes can be represented in k‐space with sampling positions in k_y ‐ k_z space in analogy to 2D CAIPIRINHA for 3D imaging.

Figure 5

Simultaneous two slice experiment at acceleration factor R = 2 without sequence modification (a) and using a FOV/2 CAIPIRINHA shift for the second slice (b) either accomplished by RF phase encoding or gradient encoding. The folded slices in A and B can be disentangled using for example standard 1D SENSE. To this end the sensitivity maps of the individual slices are arranged along the phase encoding direction accounting for relative shifts in the case of CAIPIRINHA. As in (a), both slices are directly superimposed on top of each other the parallel imaging algorithm relies on sensitivity variations along the slice direction only. In the case of insufficient coil sensitivities along the slice direction (e.g., small slice distance) the reconstructed images suffer from large g‐factor noise enhancement. In (b), the superimposed slices appear shifted by FOV/2 with respect to each other. This allows the pMRI reconstruction (here SENSE) to use sensitivity variations along both the phase and slice directions resulting in significantly lower g‐factor noise in the reconstructed images.

Figure 6

Various GRAPPA based reconstruction schemes for R = 4 SMS acquisition with acceleration along the slice and the phase encoding (PE) directions including a CAIPIRINHA shift of FOV/4 along PE. a: SENSE‐GRAPPA hybrid along PE: The autocalibration signals (ACS) required for the GRAPPA reconstruction are produced by rearranging the calibration data for the individual slices along an extended FOV along the phase encoding direction. The required large FOV* and the relative slice shifts have to be chosen such that R* times undersampling collapses to the actual aliasing pattern, and such that all the slices appear without overlap in the FOV*. In this case this is accomplished by choosing FOV* = 2/5*FOVPE and by shifting the second slice by 5/4FOVPE with respect to the other slice. The aliasing is finally resolved in 1 single step by using standard 1D GRAPPA along PE at R* = 5. b: 2D SENSE‐GRAPPA hybrid along RO & PE: Similar to A the calibration signals of the individual slices are arranged in an extended FOV, however, along the read‐out (RO) direction providing the required individual slice shifts along the PE direction. As two times in‐plane acceleration is also used, the reconstruction can be performed either in two separate 1D GRAPPA steps or in one step using a 2D GRAPPA kernel (c) Slice‐GRAPPA. Alternatively, the slices can be separated using slice specific GRAPPA kernels to entangle the overlapping slices. The remaining in‐plane aliasing is then resolved by 1D GRAPPA along the phase‐encoding direction in a subsequent step.

Figure 7

Reconstructed volumes and g‐factor analysis for MB‐15 RARE/TSE at 3T. Note, that because two slices remain outside the head, this leads to an effective MB factor of MB_eff = 13. a,b: Blipped‐CAIPI suffers from noise amplification especially in the middle of the volume with g_max = 3.24 and g_avg =1.42. c,d: Wave‐CAIPI yields high quality data and close to perfect SNR retention with g_max=1.34 and g_avg = 1.12. e: Fully sampled MB‐1 product RARE/TSE acquisition is able to cover a very limited FOV (14 slices) in the same acquisition window. f: The 1/g‐factor analysis for MB_eff = 13 reconstruction without FOV shifting and Wave. All images (a,c,e) are scaled identically. [Reproduced with kind permission from Gagoski et al 58].

Figure 8

In vivo 12‐slice myocardial perfusion imaging. a: Simple Fourier transform of one measurement, showing the simultaneous acquisition of two short axis slices. b: Reconstruction of the simultaneously excited slices shown in (a). c: G‐factor maps for the reconstruction shown in (b). d: Enlarged sections of all 12 slices acquired during one first pass experiment. Two slices each were acquired at the same time. The passage of the contrast agent through the myocardium is shown in eight short and four long axis slices. The red boxes indicate the two simultaneously excited slices shown in (a) and (b). (Reproduced with kind permission from Stäb et al 65.