Previous work has shown that dual-venc k-t GRAPPA accelerated 4D flow MRI provides low-noise, non-aliased flow angiograms showing time-resolved flow velocity throughout the cardiac cycle¹. This method requires long acquisition times and advanced reconstruction methods such as k-t GRAPPA to acquire data sets with acceptable spatial and temporal resolution. However, for evaluation of cerebral arteriovenous malformations (AVM), key clinical information is contained in the flow distribution among cerebral vessels, suggesting that it may be possible to use a time-averaged rather than time-resolved dual-venc flow angiogram for this application. In this study we compare time-resolved and time-averaged dual-venc 4D flow MRI in healthy volunteers to determine the feasibility of the time-averaged method and its fidelity in capturing flow distribution in the cerebral vasculature.Time-averaged and time-resolved dual-venc 4D flow MRI data was acquired for 10 healthy volunteers (mean age = 54.0±17.6 years, 7 male, 3 female) on a 3T Siemens MAGNETOM Skyra MRI scanner, with imaging parameters shown in Figure 1. Both data sets were corrected for background phase offset error: static tissue was defined using standard deviation over time in time-resolved acquisitions²; time-averaged scans used a manually created 3D mask of static tissue. For both scans, high-venc acquisitions were corrected for aliasing and used to correct aliased voxels in the low-venc acquisition³ in order to achieve a combined dual-venc data set with high velocity to noise ratio. The phase-contrast MR angiogram (PC-MRA) was calculated and provided the basis for a refined angiogram segmentation. In both scans, net flow and peak velocity were determined at major cerebral vessels: superior sagittal (SAG), straight (STR), and left/right transverse sinuses (TS); basilar artery (BA); and left/right internal carotid (ICA), anterior cerebral (ACA), middle cerebral (MCA), posterior cerebral (PCA) and posterior communicating arteries (PCOM). Streamlines were used to visualize blood flow within vessel boundaries as defined by the segmented angiogram. Streamline quality was graded 0-3 on continuity of vessels and uniformity of flow direction in each vessel (“0”=no streamlines, “1”=<50% uniformity, “2”=<100% uniformity, “3”=100% uniformity). Absolute values and ratios among vessels of net flow and peak velocity were used to compare data collected using the time-averaged and time-resolved methods. High resolution time-resolved and time-averaged dual-venc 4D flow MRI data was successfully acquired in all 10 subjects. Time-averaged dual-venc 4D flow MRI was acquired in an average 15.3±1.9min scan time with three times larger imaging volume than time-resolved data (similar acquisition time of 14.7±2.2min with additional 18.1±5.6min for k-t reconstruction; field-of-view large enough to cover circle of Willis). Motion artifacts were observed on some time-averaged data sets. The combined dual-venc data sets provided the high velocity range and low velocity noise desired in high spatial resolution imaging. Results are summarized in Figure 3. While absolute values of net flow and peak velocity are significantly different between the two scanning methods for most planes, net flow and velocity distributions among vessels are largely preserved, as shown by ratios between independent branches in vessel architecture. Within-subject Pearson correlation between time-resolved and time-averaged net flow for all vessels was significant (R=0.31, p=1.1x10-11). The most reliable measure was flow distribution, meaning ratio of flow between left and right hemisphere ICA, MCA, ACA, PCA, PCOM and TS: R=0.72 and p=1.2x10-8. Peak velocity values are less consistent between methods (R=0.46, p=0.2x10-7), but still significant. Visual grading of streamline quality showed higher quality when using the time-averaged approach (2.1±0.67 versus 1.8±0.65 for time-resolved, p=0.0003).This work provides an evaluation of the accuracy of a new method for imaging flow hemodynamics in cerebral vasculature. The benefit of time-averaged rather than time-resolved sequences with similar scan time is significant reduction in data volume, potentially increased spatial resolution, and increased FOV, which enabled the simultaneous visualization of more vessel branches than time-resolved scans. This may increase ease of evaluation and coverage of large AVMs. Flow distribution is well-correlated between the two scan types. Preliminary analysis suggests that time-averaged acquisition could be an effective way of acquiring essential cerebral flow information that could be applicable in patients with AVM; this application requires evaluation by future studies and sequence optimization.