Cross-sectional imaging of the pediatric brain is often performed for a targeted assessment of the size of the ventricles after a neurosurgical intervention (shunt catheter placement) or shunt revision in hydrocephalus. Secondary imaging goals can include detection of cysts and other masses, intracranial hemorrhage, and venous thrombosis. For these indications, head computed tomography is often performed, with attendant risk of cancer attributable to radiation, especially when serial imaging is required. MRI can be used as an alternative modality for these indications. Single shot fast spin echo (SSFSE) is frequently used because of its T2 contrast, speed and robustness to motion, off-resonance and other system imperfections^1,2. However, SSFSE involves high rates of RF energy deposition, particularly at 3T, which translate in delays between the acquisition of consecutive slices in order to remain within the recommended guidelines for SAR (Specific Absorption Rate). A recent study showed that a variation of SSFSE with variable refocusing flip angles (vrfSSFSE) can reduce RF energy deposition and hence delays between the acquisitions of successive slices³. In this study we hypothesize that a fast brain MRI approach that incorporates vrfSSFSE can decrease scan time and inter-slice motion artifacts while maintaining or improving image quality (IQ) compared to conventional SSFSE.33 consecutive children (median age: 4.9 years; range: 6 weeks-17 years) referred for non-contrast sedation-free brain MRI at 3T (GE MR750, Waukesha, WI), who had both SSFSE and vrfSSFSE scans performed between April 2014 and October 2014, were retrospectively identified. Subjects were scanned with a receive-only 8-channel brain coil. The imaging protocol consisted of three orthogonal planes of SSFSE (constant 130° refocusing flip angle) and a coronal vrfSSFSE (first refocusing flip angle 130°, ramped down to 90°, increased to 100° at the center of k-space, finally ramped down to 45° at the end of the echo train). Imaging parameters were: TE=86 ms, 256x256 matrix, 83 kHz bandwidth, consecutive 4-mm thick slices with interleaved slice ordering. FOV (18-22 cm) was adjusted to each patient’s anatomy. Both SSFSE and vrfSSFSE used a parallel imaging factor of 2 and half Fourier acquisition with homodyne reconstruction. Two neuroradiologists independently and blindly evaluated SSFSE and vrfSSFSE images for motion, perceived resolution (sharpness), contrast, and lesion conspicuity according to the following five-point scale: 1-nondiagnostic, 2-poor, 3-acceptable, 4-above average, and 5-outstanding. Mean scores and proportions of cases with acceptable IQ (score no less than 3) for both readers were calculated for both SSFSE and vrfSSFSE. The null hypothesis of no significant difference in IQ between the two sequences was assessed with a two-sided Wilcoxon signed rank test, with the Holm-Bonferroni correction for multiple comparisons. Inter-observer agreement was assessed using Cohen’s kappa. Differences in the minimum repetition time, used as a proxy for scan time, were evaluated using Student’s two-tailed paired t-test. p<0.05 was considered significant. The mean scores and proportions of cases with diagnostically acceptable IQ are reported in Fig.1 and Fig.2. All image quality metrics scored on average higher than 3 with the exception of image sharpness and contrast, which were assigned lower scores by reader 2. There were significantly less motion artifacts in vrfSSFSE than SSFSE according to reader 1 (p<0.001), while reader 2 found the same level of motion-related artifacts in both sequences (p=0.25). Similarly, only one of the readers found significantly reduced blurring in vrfSSFSE with respect to SSFSE (p=0.79 for reader 1; p<0.001 for reader 2). In terms of image contrast and lesion conspicuity, both readers scored vrfSSFSE as having a significant advantage over conventional SSFSE (p<0.01). Representative images for vrfSSFSE and SSFSE are shown in Figs.3-5. Inter-reader agreement ranged between poor and moderate, with the best agreement found in the evaluation of motion artifacts. There was substantial disagreement in the evaluation of blurring and image contrast in SSFSE and lesion conspicuity on vrfSSFSE images. For both sequences and each category, high prevalence was observed, which resulted in low kappa. In all cases marginal homogeneity (total number of positive and negative scores for each reviewer) and symmetrical disagreements were observed. The minimum TR (and corresponding scan duration) was significantly shorter for vrfSSFSE than conventional SSFSE (min TR vrfSSFSE = 543±157ms vs. min TR SSFSE = 1115±383ms, p<0.0001).vrfSSFSE is significantly faster than conventional SSFSE and gives equivalent or improved IQ as well as potentially reduced motion artifacts due to the overall shorter scan duration.