Simultaneously multi-slice (SMS) has been a powerful tool for single-shot EPI (ssEPI) DWI acceleration, but still not well established for accelerating interleaved EPI (iEPI) DWI. The main challenge is how to effectively combine slice un-folding and inter-shot phase correction. Recent studies have proposed several k-space based methods for iEPI DWI phase correction ^1-3, for which inter-shot phase variances are corrected by interpolation in k-space, based on extra acquired navigators. This characteristic makes it straightforward to integrate such methods into the SMS 3D reconstruction framework ⁴. In this study, a 3D k-space reconstruction method for SMS accelerated iEPI DWI is proposed. A new acquisition strategy for navigators is designed to accurately estimate inter-shot phase variance for all simultaneously excited slices. Then slice un-folding and inter-shot phase correction are performed in a SMS 3D k-space. The performance of the proposed method is compared with ssEPI and un-accelerated iEPI DWI, and has been shown to be applicable for both 4-sh and 8-sh iEPI DWI.

Sequence A previously reported generalized blipped-CAIPI sequence ^1,3 is used in this study, as shown in Fig. 1. The navigator is optimized to reduce its distortion in both phase encoding and slice selective direction, by using CAIPI under-sampling patterns and reducing its echo spacing (ESP).

Reconstruction The pipeline of the proposed 3D k-space reconstruction is shown in Fig. 2. Data with SMS acceleration factor MB = 3, 2-shot acquisition are shown as an example. For simplicity, the kx and coil dimension are not shown here. First, the under-sampled navigator of each shot is recovered to a full SMS 3D k-space (Fig. 2(a)) using traditional GRAPPA ⁴. Then the weighting matrix (Fig. 2(b)) for slice un-folding and inter-shot phase correction is calculated from the recovered navigator. For the image echo, data from different shots are divided into different k-space (from Fig. 2(c) to (d)). Then the missing data is recovered using k-space interpolation (arrows in Fig. 2(d)), using the calculated weighting matrix in Fig. 2(b). Here, both the inter-shot phase variation and coil sensitivity information are used. The fully recovered 3D k-space data (Fig. 2(e)) are inversely Fourier transformed and combined over different channels and shots.

Experiments All scans were performed on a Philips 3.0T Achieva TX MRI scanner (Philips Healthcare, Best, The Netherlands) using a 32-channel head coil. All human studies were performed under IRB approval from our institution. The imaging parameters for different acquisitions were summarized in Table 1. For diffusion imaging, diffusion preparation was applied in 16 directions with b = 800 s/mm². For un-accelerated imaging, 4 slices with gap = 16mm were acquired, while for SMS accelerated imaging, the same slices were covered by simultaneously exciting 2 slices with gap = 36 mm. In this study, the ESP of the navigator was reduced to half of the image echo, which induced a navigator size of 80×29 (kx×ky), while the size of the image echo was 216×216.

Data analysis The SNR for msEPI DWI images with or without SMS acceleration were calculated using the pseudo-multiple replica method with 128 repetitions ⁶. FA maps were calculated using DTIstudio ⁷ and compared.

Fig. 3 shows the DWI image comparisons among ssEPI (a), un-accelerated 8-sh EPI (b), accelerated 4-sh (c) and 8-sh EPI (d) with MB=2. As shown, the msEPI DWI images show higher spatial resolution and less distortion (red arrow heads) than ssEPI. The proposed 3D reconstruction has successfully reconstructed SMS accelerated 4-sh and 8-sh EPI DWI images. The SNR drop from SMS acceleration is also shown from (e) to (g).

Fig. 4(a) to (d) show the corresponding color-coded FA (cFA) maps, with ssEPI as references. Both un-accelerated and accelerated 8-sh EPI show more detailed structures than ss-EPI and 4-sh EPI, especially where severe distortion exists (yellow arrow heads).

In this study, a 3D k-space reconstruction method for SMS accelerated iEPI DWI has been proposed for the first time. A new strategy is first proposed to acquire SMS 3D navigator, which can estimate the inter-shot phase variance for all simultaneously excited slices. For the reconstruction, slice unfolding and inter-shot phase correction are effectively combined into one SMS 3D k-space, which is novel in concept and easy to implement. The performance of the proposed method has been demonstrated in iEPI DWI with 4 shots and 8 shots. This method is promising for high resolution iEPI DWI imaging acceleration, which is valuable for clinical diagnosis and neuroscience study.