Two-dimensional simultaneous multi-slice (SMS) imaging techniques have recently been applied for arterial spin labeling (ASL) perfusion imaging. Compared to standard 2D ASL, SMS ASL can reduce T1 relaxation effect of the label through shortened acquisition time; improve spatial coverage and resolution with little penalty in SNR^1-2. However, existing 2D SMS ASL techniques are sub-optimal for the background suppression (BS) technique since multiple SMS excitations are required. As a result, the temporal SNR (tSNR) of SMS-EPI ASL is inferior to that of 3D BS GRASE ASL¹. In this study, we propose a novel constrained slice-dependent BS scheme for 2D multi-slice pseudo-continuous ASL (pCASL) with SMS-EPI acquisition, to suppress background signal across a wide range of T1s.Background suppression is generally achieved by suppressing brain tissue signals using several inversion pulses following a pre-saturation of the whole imaging volume³, while leaving the target spins in blood undisturbed. For 2D multi-slice imaging, however, the background tissue signal recovers between multi-slice acquisitions, thereby compromising the efficiency of BS. As shown in Fig. 1, a constrained slice-dependent BS scheme is proposed that employs a group of slice selective pre-saturation pulses and two global inversion pulses, the timings of which are determined based on the pCASL labeling duration and post-labeling delay (PLD) so that the background tissue signal of each imaging slice can be precisely nulled/suppressed when it is excited. As shown in Fig. 2 of simulation results, the signal of static brain tissue of 2 imaging slices can be suppressed at the time of excitation, while the blood signal can be largely preserved under the proposed BS scheme. This BS scheme can be expanded for 2D SMS pCASL scans using multiband pre-saturation and excitation RF pulses. Optimization of BS timing parameters including flip angles (FA₀) of pre-saturation pulses, saturation times (τ_sat) and inversion times (τ_inv) (Fig. 1) can be treated as a nonlinear optimization problem to minimize total residual signal, expressed as the squared sum of background signals across a range of T1 values for all imaging slices. The proposed BS scheme was implemented for an SMS EPI pCASL sequence on a Siemens 3T Prisma scanner with a 20-channel head coil. The imaging parameters were: FOV=220 mm, matrix=64×64, slice thickness=5 mm, TE=14 ms, TR=3540 ms, flip angle=90°, label/control duration=1500 ms, PLD=1000 ms, slice acquisition time=45 ms and 30 acquisitions. 6/12/18 slices were acquired using 2D EPI with single-band (SB) and multiband (MB)-2/3 slice accelerations, respectively, and aliased slices were separated with slice-GRAPPA algorithm⁴. Two hyperbolic secant pulses were implemented as non-selective inversion pulses and pre-saturation was achieved by slice selective RF pulses followed by gradient spoilers. Scans were performed on a phantom and human brain. The phantom was designed as a four-layer (slice) rack with four test tubes in each layer containing gel of different T1 values (1148, 1586, 1908 and 2197 ms from left to right in Fig. 3). Optimization of the BS scheme was performed to suppress all 4 T1 values of the phantom, as well as T1 values of gray and white matter (600 to 1400 ms) for human studies. To evaluate the performance of the proposed BS scheme, voxel-wise tSNR was calculated as the mean value divided by the standard deviation of subtracted perfusion images.Figure 3 shows 4 slices of phantom acquired with and without BS. The signal of the 4 test tubes can be on average suppressed to 10.68% of its original intensity across 4 slices. For human experiment, an optimized BS scheme was designed for six slices considering the timing constraints. The optimized BS timing parameters were: [ FA_0,1 ... FA_0,6 ] = [97, 119, 123, 161, 180, 180]°, [ τ_sat,1 ... τ_sat,6 ] = [108, 293, 6, 6, 0, 0] ms and [ τ_inv,1 τ_inv,2 ] = [84, 795] ms. Average background signal was 4.8±4.4% after suppression based on simulation. Figure 4 shows both background images (a, c, e, g) and corresponding perfusion maps (b, d, f, h) from non-BS SB, BS SB, BS MB2 and BS MB3 sequences of the same scale. As shown in Table 1, an 89% of background signal reduction was achieved with the proposed BS scheme and the mean tSNR across the whole brain increased from 0.43 to 0.95/0.80/0.64 in BS SB/MB-2/3 data respectively, which indicates a more stable perfusion signal in BS acquisitions.The proposed constrained BS scheme for 2D multi-slice pCASL in conjunction with SMS acquisition offers a promising approach for effective suppression of background signals across a wide rang of T1s while ensuring sufficient brain volume coverage.