As a recently launched method, spatiotemporally encoded (SPEN) magnetic resonance imaging (MRI) has high robustness to field heterogeneity and chemical shift effect. It has been broadened from single-slice scan to multi-slice scan. In this abstract, segmented spatiotemporal encoding (SeSPEN) pulse sequence was designed for multi-slice SPEN MRI to lower the specific absorption rate (SAR) and improve the image quality.

The multi-slice SeSPEN sequence is shown in Fig. 1. The $$$90_{{\text{ses}}}^{\text{o}}$$$ pulse is segment-selective, so the quadratic phase is only impacted on the selected region along the slice-selective dimension. Subsequently a 180° hard pulse is employed to restore the spins that are not targeted by the $$$90_{{\text{ses}}}^{\text{o}}$$$ pulse back to the thermodynamically stable state for subsequent segment-selective encoding and acquisition. To store the encoding information, a selective 90° pulse ($$$90_{{\text{store}}}^{\text{o}}$$$) as the former one ($$$90_{{\text{ses}}}^{\text{o}}$$$) is employed. With the following slice-selective pulse ($$$90_{{\text{ss}}}^{\text{o}}$$$), we can successively acquire the signals of every slice in the encoded region. The signal of the targeted slice in the selected segment can be calculated using the following integral:

$$$s(t) \propto \int_{ - {L_y}/2}^{{L_y}/2} {\rho (y) \cdot \frac{{ - \exp [i{\varphi _1}(y)] - \exp [ - i{\varphi _1}(y)]}}{2} \cdot \exp [i(\gamma {G_{cr2}}{T_{cr2}}y + {k_{SPEN}}y + \gamma {G_{acq}}yt)] \cdot \exp ({\tau \mathord{\left/ {\vphantom {\tau {{T_1}}}} \right.} {{T_1}}})dy}$$$

where ρ(y) stands for the spatial profile of spin density, γ is the gyromagnetic ratio, τ is the time between the $$$90_{{\text{store}}}^{\text{o}}$$$ pulse and the $$$90_{{\text{ss}}}^{\text{o}}$$$ pulse of the targeted slice, T₁ is the longitudinal relaxation time, G_acq is the amplitude of acquisition gradient, and φ₁(y) is the quadratic phase along the SPEN dimension (y-axis) after the 180° hard pulse, $$${k_{SPEN}} = \gamma \left| {{G_{acq}}{T_{acq}}} \right|/2$$$. A second crusher gradient whose amplitude and duration are G_cr2 and T_cr2 respectively is used to eliminate the echo planar signals arising from the extension of τ. After deconvolution reconstruction of the acquired signals,¹ super-resolved multi-slice images can be obtained.Experiments on phantom and in vivo rat brain were carried out on a Varian 7T MRI system. The results obtained by the SeSPEN sequence were compared to those obtained by the spin-echo EPI sequence, spin-echo SPEN sequence and multi-slice global SPEN sequence proposed by Frydman and coauthors (abbr. GlSPEN sequence)² to demonstrate the advantages of our method. Fig. 2 demonstrates the imaging results of twenty-four anatomical slices of a live rat brain along the axial plane. The FOV was 4.5×4.5 cm² and slice thickness was 1.5 mm. The matrix size was 128×128.As indicated by the red arrows in Fig. 2, the EPI images suffer from severe geometric distortion which the SPEN approaches can effectively alleviate. From the region marked by yellow rectangle, we can see that the GlSPEN image was poor. This is because the phase imparted by field inhomogeneity was not eliminated. Whereas the SeSPEN MRI could efficiently avoid this problem. Beyond that, compared to the GlSPEN sequence, the SeSPEN MRI can effectively remit the signal loss caused by T₁ relaxation, as indicated by the green ellipse. For the GlSPEN sequence, the signal was attenuated greatly with the increase of slice number. At the meantime, compared to the spin-echo SPEN MRI, the SAR of SeSPEN MRI declined significantly, because only three chirp pulses were used for multi-slices.The SeSPEN MRI is not only robust to field heterogeneity and chemical shift effect, but also efficient for multi-slice MRI with high image quality after super-resolved reconstruction.