Arterial spin labeling (ASL) is a robust MR technique for assessment of cerebral blood flow (CBF), and adds value across a wide range of pathologies, including acute ischemic stroke, brain tumors, infection, demyelinating disease and dementia¹. While ASL is a powerful technique to quantitatively assess an individual’s cerebrovascular status, visual inspection, region-of-interest analysis, and CBF values show considerable variance. This variance in quantification arises partly due to variability in acquisition protocols. In this work, we sought to evaluate the contribution of various readout schemes on specific imaging metrics as assessed by statistical 3D mapping, and a pseudocontinuous labeling method that was conserved across all evaluated sequences.Experiment: Seven volunteer subjects underwent a ASL scans at 3T (Ingenia, Philips Healthcare, Best, The Netherlands) using a 32-channel SENSE receive coil during this pilot study. All subjects underwent 5 consecutive ASL sequences with an identical pseudocontinuous ASL preparation using five distinct sequence readouts: 1) 2D echo planar imaging (EPI) acquired with product background suppression (BS) and label duration (LD)/post label delay (PLD) = 1.65/1.6 sec; 2) A 3D GRAdient and Spin Echo (GRASE) with product BS and LD/PLD = 1.8/2.0 sec (GRASE-A); 3) An alternate protocol 3D GRASE (GRASE-B) with BS and LD/PLD = 1.5/1.5 s (based on¹); 4) A 3D Turbo Spin Echo (TSE) stack of spirals (SOS) trajectory with spiral-out readout (3D SOS) and BS and LD/PLD = 1.5/1.5 s (based on²); and 5) 3D spiral TSE sequence employing a cylindrical distributed spiral (CDS) trajectory with a spiral-in/spiral-out readout and the same LD and PLD as 4². All sequences were acquired with 21 slices (to match vendor product protocol) and matrix: 148 x 148; FOV: 260mm in-plane and 84mm through-plane, yielding pixel size=1.76mm in-plane and 5.7mm through-plane. Analysis: All intrasubject ASL 3D images were first coregistered to the 2D EPI dataset to perfectly align slice locations³, and subsequently all readout methods (2D and 3D readouts) were transformed by stereotactic anatomic standardization to the Talairach atlas⁴. Sample difference maps are provided in Fig. 1. All readout methods were subsequently evaluated using descriptive 3D statistics, calculated on a per-readout basis across all subjects, with fixed distribution. Furthermore, relative CBF maps were obtained be dividing the difference map by the averaged difference in all gray matter voxels. Relative CBF maps in the frontal (FRT), temporal (TMP), parietal (PRT) and occipital (OCT) lobe were evaluated. Inter-subject Coefficient of Variance (COV) was also calculated for each of the four lobes as the ratio of the mean (MN) to the standard deviation (SD) of the lobar CBF values. Sample images depicting COV, MN and SD are illustrated in Fig. 2, calculated per readout across all seven subjects on a per-voxel basis. MN values, on a per-voxel basis, are depicted in Fig. 3 (Red = 4 SD from MN). SD values, on a per-voxel basis, are depicted in Fig. 4 (Red = 4 SD from MN). COV (SD/MN) values are shown in Figure 5. Overall, EPI and GRASE-B showed high COV (>0.4) while GRASE-A and SOS had low COV (0.3<COV<0.4). COV for CDS ranged from 0.3 -0.5. Since different LD and PLD values will likely introduce differences in the relative CBF values, we assessed only those Bland-Altman plots where LD/PLD are identical and only the readouts are different ie, for GRASE-B, SOS, CDS. Representative plots for the frontal, temporal, parietal, and occipital lobes are shown in Figure 5. In general, relative CBF values with GRASE-B were higher in the frontal, temporal, and parietal lobe and lower in the occipital lobe when compared to SOS and CDS (Figure 5), although not significantly different. With 7 subjects, no significant systematic bias was observed.In this pilot project, the 2D EPI, GRASE-A, GRASE-B and 3D SOS, and 3D CDS readouts were compared, although note should be made of the non-uniform LD/PLDs used in this pilot work. Relative CBF values with all five methods were similar and not significantly different in the 7 subjects. GRASE-A and SOS readouts provided the lowest inter-subject COV and will likely allow group comparisons with fewer subjects. Overall, the 3D CDS performance is in keeping with that of other currently used methods. Descriptive statistical 3D mapping can offer insight into the performance of the employed pseudo-continuous labeling scheme coupled with five differing readout methods. In this pilot project, the GRASE-A and 3D SOS readouts provided the lowest inter-subject COV, suggesting group comparisons with fewer subjects might be possible, although no significant systematic bias was observed across readout performance.