New type of ultrafast imaging techniques have recently been developed using spatiotemporal encoding (SPEN) [1, 2]. In SPEN imaging, a frequency-modulation pulse like a chirp pulse is used for spin excitation and produces a quadratic phase which sequentially localizes a signal in both time and space along the SPEN direction. SPEN imaging can provide higher immunity to image artifacts than conventional EPI counterparts such as distortions in the presence of local field inhomogeneities, and Nyquist ghosting along the phase-encoding direction. Besides, it fully refocuses T_²* effects by virtue of application of π refocusing pulses. While RASER (Rapid Acquisition by Sequential Excitation and Refocusing) sequence acquires all the echoes with same TE by using two refocusing pulses [1], other SPEN techniques using a single refocusing pulse offers a shorter effective TE than RASER but with varying TE, thereby causing signal variation in the SPEN dimension [2]. Here, we propose an effective way of overcoming this problem of TE variation in single-refocusing SPEN imaging. The proposed method is demonstrated by phantom and in-vivo rat brain imaging at 9.4 T.In single-refocusing SPEN imaging, spin isochromats are sequentially refocused but in reverse order when compared to excitation. Accordingly, signal variation due to varying TE appears in the SPEN direction. One simple way to solve this issue is to compensate the signal variation by using exp(-TE/T₂). However, this approach seems not very effective since T₂ and B₁ variations also need to be considered. In contrast, our proposed method is simple but much more effective: Two images are obtained with an opposite polarity of gradients along the SPEN direction, thereby having an opposite direction of TE variation (Fig. 1A and 1B) and the two images are then averaged for compensating the TE variation. Super-resolved algorithm was used for image reconstruction in the SPEN direction [3]. To validate the proposed method, a spherical phantom (with a trace of Gd) and in-vivo rat brain were scanned at Bruker 9.4 T using a single-channel transceiver volume coil and a 4-channel surface coil, respectively. Two single-shot 2D SPEN sequences were implemented with an opposite polarity of gradients in the SPEN direction (Fig. 1). For all experiments, slice thickness was 2mm and, except the rat-brain SPEN imaging, FOV=30x30mm² and Matrix size=64x64. For phantom imaging, scan parameters were: TE/TR=36/1000ms, (Pulse)bandwidth×length=256, Pulse length=26.88ms. For rat-brain imaging, RARE and 2D spin-echo EPI images were also acquired for reference. For RARE, TE/TR=36/2500ms and RARE factor=8. For spin-echo EPI, TE/TR=36/1000ms. For 2D SPEN imaging, TE/TR=36/1000ms, FOV=30x15mm², Matrix size=64x32, Pulse length=13.44ms.Figure2 shows the results of phantom imaging using the proposed method. As expected, signal decay due to TE variation along the SPEN direction appears in each case of applying positive (A and B) and negative gradients (C and D) in the SPEN direction. As a result of averaging the two images of A and C, the signal decay induced by varying TE was effectively removed in E and F. For better representation of signal variation and averaging effect, Figs. 2B, D, and F are displayed in color scale and Fig.2G shows 1D profiles obtained along the center line. Figure 3 shows the coronal images of in vivo rat brain. Figures 3A and B are acquired with RARE and spin-echo EPI, respectively, for reference, and figures 3C, D, and D are the results of using the proposed method. SPEN imaging shows much better images than the spin-echo EPI owing to its robustness in the presence of local field inhomogeneities. Figure 3E shows that signal variation along the SPEN direction is effectively compensated with image quality almost comparable to RARE.We here suggested a simple and effective way of overcoming the problem of TE variation that happens when a single-refocusing SPEN sequence is used. In addition to the intrinsic advantages of being robust in local field inhomogeneities, the single-refocusing SPEN imaging combined with the proposed method not only enables data acquisition with constant TE, but also provides a shorter TE by a half than a double-refocusing SPEN sequence like RASER, reducing the T₂ signal decay related to long TE. We expect that this approach find its use in many applications, especially, at ultrahigh fields (e.g., 7 T and above). As very recently reported by Luisa et al [4], one of promising applications of the proposed method is the fMRI studies that focus on the brain areas nearby air-tissue interfaces with large magnetic susceptibility such as the orbitofrontal and the inferior temporal lobes.