Blipped-controlled aliasing in parallel imaging for simultaneous multislice echo planar imaging with reduced g-factor penalty

Abstract

Simultaneous multislice Echo Planar Imaging (EPI) acquisition using parallel imaging can decrease the acquisition time for diffusion imaging and allow full-brain, high-resolution functional MRI (fMRI) acquisitions at a reduced repetition time (TR). However, the unaliasing of simultaneously acquired, closely spaced slices can be difficult, leading to a high g-factor penalty. We introduce a method to create interslice image shifts in the phase encoding direction to increase the distance between aliasing pixels. The shift between the slices is induced using sign- and amplitude-modulated slice-select gradient blips simultaneous with the EPI phase encoding blips. This achieves the desired shifts but avoids an undesired "tilted voxel" blurring artifact associated with previous methods. We validate the method in 3× slice-accelerated spin-echo and gradient-echo EPI at 3 T and 7 T using 32-channel radio frequency (RF) coil brain arrays. The Monte-Carlo simulated average g-factor penalty of the 3-fold slice-accelerated acquisition with interslice shifts is <1% at 3 T (compared with 32% without slice shift). Combining 3× slice acceleration with 2× inplane acceleration, the g-factor penalty becomes 19% at 3 T and 10% at 7 T (compared with 41% and 23% without slice shift). We demonstrate the potential of the method for accelerating diffusion imaging by comparing the fiber orientation uncertainty, where the 3-fold faster acquisition showed no noticeable degradation.

Figure 1

A description of the blipped-Wideband and blipped-CAIPI methods for creating FOV/2 inter-slice image shift between two simultaneously excited sliced (the bottom slice at isocenter) and the source of the blurring artifact (tilted voxel) of the blipped-Wideband method.

Figure 2

Generalization of the blipped-CAIPI 2-slice FOV/2 shift method to the case where neither slice is at isocenter. (A) Case 1, where one slice at isocenter (red), and Case 2, the general case with neither slice at isocenter (orange). (B) The phase at the slice center for the two cases. The resultant slice phases for Case 2 are not as desired, which leads to an N/2 PE ghost even for the infinitely thin slice condition. This problem can be resolved by adding +ϕ/2 phase to even k_y lines and −ϕ/2 phase to odd k_y lines prior to reconstruction.

Figure 3

Generalization of the blipped-CAIPI method to FOV/3 and FOV/4 shifts between successive slices. Each panel shows: top, G_z gradient scheme; and bottom, corresponding phase diagrams between excited slices and at slice edges of blipped-CAIPI acquisition for various inter-slice image shifts.

Figure 4

Results from 3× slice-accelerated SE-EPI acquisition with FOV/2 inter-slice shift. (A) Unfolded images of the unaliased 3D volume; left: coronal and sagittal views, right: axial views of an unalised slice group. (B) Left: aliased image of blipped-CAIPI slices. Right: the corresponding Monte-Carlo generated retained SNR maps of the unfolded slices. (C) Aliased slice group and 1/g maps for 3× slice-accelerated acquisition without inter-slice shift. With blipped-CAIPI inter-slice shift, SNR retention is close to 100% in all locations, whereas for the acquisition without the shift the SNR retention drops to as low as 50% in some areas of the image.

Figure 5

Diffusion imaging results from a conventional unaccelerated acquisition (top row) and 3× slice-accelerated acquisition with FOV/2 inter-slice shift (bottom row). Left: directionally encoded diffusion color-maps; right: maps of the 95% uncertainty angle for estimate of the principal (fiber 1) and crossing (fiber2) fiber orientations.

Figure 6

Example 1 mm isotropic whole-brain GRE-EPI blipped-CAIPI acquisition at 7T with 3× slice and 2× inplane acceleration (TR = 2.88s). (A) Coronal, sagittal and axial views of the unaliased 3D volume. (B and C) Corresponding 1/g map of a representative slice group for blipped-CAIPI and non-blipped acquisition, respectively.

Figure 7

The 6 slice per shot GRE-EPI at 3T acquired using the SER method (2×) and 3× slice blipped-CAIPI acceleration with 2× in-plane GRAPPA. (A) Left, coronal and sagittal views of the unfolded slice volume with the 2 SER excited slice groups shown in red and yellow. Right, the corresponding 6 unaliased slices that were acquired with the single EPI readout. (B and C) The 1/g maps of the 6 slice group acquire with the blipped-CAIPI and non-blipped acquisition respectively. Peak 1/g was reduced from 0.33 to 0.57 by the blip scheme resulting in a SNR improvement of 73%.

Figure 8

(A) The underlying relationship in the SENSE/GRAPPA algorithm proposed by Blaimer el al., (23)where an undersampling of a concatenated image’s k-space data results in a slice-collapsed image and, conversely, application of the GRAPPA operator to the aliased data generates a concatenated but unaliased version of the two slices. (B) Phase variations in the PE direction (y) that have to be synthesized in order for the governing equation of the SENSE/GRAPPA algorithm to be satisfied for the non-shift (left) and shifted (right) 2× multi-slice acquisitions. Shown are the desired phase variations in the concatenated image space and in the standard image space for each of the simultaneously acquired slices. For FOV/2 shift acquisition, a discontinuity in phase exists in the desired phase variation profile for one of the slices. This condition is difficult to satisfy with smooth coil sensitivities and small GRAPPA kernels, resulting in an unaliasing artifact.

Figure 9

Unaliasing artifact produced by applying the SENSE/GRAPPA method to slice-shifted data (bottom) compared to data acquired without the slice shift (top). For the shifted acquisition, residual aliasing artifact can be observed corresponding to locations of discontinuity in the desired phase variation (white arrows).

Figure 10

(A) Schematic of the slice-GRAPPA algorithm to obtain the k-space data for the unaliased slices. A set of GRAPPA kernels is applied to the k-space data of the aliased slices to generate k-space data sets for each slice. One kernel set is needed for each slice. The kernels are trained from a conventional fully slice-sampled data set. (B) Tests of the validity of the simplified governing equations of the slice-GRAPPA algorithm in brain data. This relationship exists between the coil sensitivity profiles and GRAPPA kernel, and should be independent of the underlying anatomy in the image.