4D-flow MRI is used to quantify blood flow in a variety of neurovascular pathologies including intracranial aneurysms [1], but is limited by long scan times as well as moderate spatial and temporal resolution. Conventional slice selective RF pulses have long pulse durations and/or poor slab profiles, which reduce scan efficiency by increasing the field of view (FOV) needed to avoid aliasing in the slab direction. Hargreaves et al. [2] described VariablE Rate Selective Excitation (VERSE) RF pulses which are optimized for short, high time bandwidth product (TBW) excitation by maximizing either the slice select gradient amplitude or the RF amplitude at all times. However, VERSE pulses have not previoulsy been reported for use in 4D flow. Our objectives were: 1) To design and simulate steady-state slab profiles for minimum time, high TBW VERSE pulses; 2) To compare the slice profiles generated by TBW=10 (TBW₁₀) and TBW₂₈ VERSE pulses to a conventional TBW₁₀ windowed sinc (WSinc) pulse used in our clinical 4D flow sequence; and 3) To validate the improvement in sequence efficiency, confirm flow accuracy, and evaluate total SAR deposition for VERSE+4D-flow compared to our clinical 4D flow aneurysm imaging protocol.Minimum time VERSE pulses were created by optimizing high TBW WSinc pulses subject to the following hardware limits: GMax=43mT/m, SRMax=172T/m/s, B1,Max=13.5µT, gradient raster time=10µs and RF raster time=5µs. Pulses were designed and simulated in Matlab (www-mrsrl.stanford.edu/~brian/mintverse/). TBW₁₀ and TBW₂₈ VERSE pulses were implemented on a 3T scanner (Siemens Skyra: 20-channel head coil) in the 4D flow sequence and evaluated in a stationary aqueous copper-sulfate phantom and in healthy subjects. VERSE slice profiles were compared to the TBW₁₀ WSinc pulse in the phantom using a 40mm excited slab thickness (TR=6.41-6.51, TE=3.80-3.87, FOV=200x182x60mm, matrix-size=176x160x60 FA=15°, BW=822 Hz/pixel). In vivo validation between WSinc and the TBW₁₀ VERSE pulse was performed in the carotid arteries of healthy volunteers (N=5) using the following parameters: TE/TR=3.80-3.87/6.41-6.51, FOV=200x182x44-52mm, matrix-size=140x140x44-48, GRAPPA=2, cardiac phases=7-12, scan-time=11:40-19:36, FA=15°, BW=822 Hz/pixel. All volunteers signed statements of informed consent.Bloch simulations of steady-state slab profiles for VERSE pulses spanning TBW2 to TBW100 were used to select two VERSE pulses for comparison to the conventional 900µs, TBW₁₀ WSinc pulse: a fast 510µs TBW₁₀ VERSE pulse and an 870µs TBW₂₈ VERSE pulse with an improved slice profile. Steady-state slab profile simulations (Figure 2a) and phantom validation (Figure 2b-c) demonstrate that a TBW₂₈ VERSE pulse provides a sharper slab profile than either the TBW₁₀ WSinc pulse or TBW₁₀ VERSE pulse. For a 40mm slab a TBW₂₈ VERSE pulse decreases FWHM by 11% compared to the standard WSinc pulse. Using a cutoff of 0.15•Mxy, phantom validation shows that the WSinc pulse excited 9mm out of slab compared to 3mm for the TBW₂₈ VERSE pulse. Of the 40mm desired profile, 38mm and 36mm were excited above 0.75•Mxy for the WSinc and TBW₂₈ VERSE pulse respectively. The results from simulations and phantom validation suggest that the VERSE TBW₂₈ pulse can be used with a 44mm FOV without significant aliasing artifacts. TBW₁₀ VERSE and TBW₂₈ WSinc pulses require a 49-52 mm FOV to avoid slab aliasing artifacts, decreasing acquisition efficiency and extending acquisition times by 2.5 minutes (16%). Table 1 demonstrates that forward flow [mL/min] and peak velocity [cm/s] were similar within the right and left common carotid artery between WSinc and TBW10 VERSE pulses, with a 6.4% difference in through plane flow and an 8.6% difference in peak velocity. Figure 3 shows axial and coronal slices using each pulse from a single volunteer. Initial evaluations show that a decreased slab FOV and reduced number of kz encodes made possible with the VERSE TBW₂₈ pulse resulted in an average total scan time reduction of 2.5 minutes (16%) compared to either the WSinc or TBW₁₀ VERSE pulses. In volunteers averaging the SAR for the TBW₂₈ VERSE pulse was significantly higher than for the TBW₁₀ WSinc pulse (1.45±1.04 W/Kg vs. 0.572±.12 W/Kg, P<0.05), but did not approach SAR limits. High TBW VERSE limits the need for slab oversampling compared to conventional slab selective WSinc pulses. Scan time reductions were determined based exclusively on decreased phase oversampling. Combining the slice select rephasing gradient following the VERSE pulse with the flow compensation waveform will further optimize the sequence and potentially reduce the TR by an additional 400µs. This would improve temporal resolution by 6% and allow the flexibility to either acquire additional cardiac phases or reduce scan time. Further validation and flow evaluation with higher TBW VERSE pulses is warranted.High TBW VERSE excitation in 4D flow provides sharper slab profiles than Wsinc pulses and improves sequence efficiency.