Pulsed Arterial Spin Labeling (PASL) ^1-4 has a short labeling time and consistently high labeling efficiency. However, the temporal width of the label bolus (or bolus duration, BD) generated by an inversion pulse is limited and typically shorter than that in Pseudo-Continuous ASL (PCASL) ⁵, resulting in lower signal-to-noise ratio (SNR). In this study, we propose using multiple inversion modules in PASL (MM-PASL) to increase the total BD for improved SNR and SNR efficiency.

In PCASL, a longer bolus can be achieved by increasing the duration of the labeling. However, in PASL the maximal size of the bolus is limited by the coverage of the RF coil. In addition, a slightly shorter BD, e.g., 700-800 ms, is typically chosen to ensure the accuracy of quantification with the QUIPSS II ⁶ (Q2) technique. As previously shown with the Multi-Module Velocity-Selective ASL (MM-VSASL) method ⁷, multiple labeling modules can be used to lengthen the total BD in a single TR. However, in MM-VSASL overlapped boluses do not cancel each other as the labeling modules are saturation pulses. In PASL, though, boluses that are generated by consecutive inversion pulses may interfere with each other if they overlap due to poorly designed timing, which may result in suboptimal improvement or even a decrease of total BD and SNR, and impaired quantitative accuracy. In this study, a wedge-shaped (WS) inversion ⁸ and a regular slab inversion with a Q2 pulse are combined to generate two boluses with known temporal widths. The timing between the two inversion pulses is matched to the BD generated by the WS inversion, so there should be no overlapping nor gap between the two boluses. To minimize the interference of the labeling pulses to the imaging volume, saturation pulses are applied to the imaging volume before and immediately after the WS inversion pulse, but not with the second slab inversion pulse.

A healthy young volunteer was scanned in a GE 3T scanner using the following sequences: 1) regular PASL (PICORE) ³ with Q2: inversion thickness=15cm, TI₁=700ms, TI=2.1s; 2) two-module PASL (WS inversion + slab inversion with Q2): the BD of the WS inversion adjusted and estimated ⁸ as 600ms (Fig. 1), TI₁=600ms for the slab inversion with Q2, TI=2.6s; 3) PCASL, labeling duration=1.2s, PLD=1.4s. The delays after the tail of the boluses left the labeling region were matched for all the scans. Other imaging parameters were: FOV=22×22cm, 7 slices, slice thickness=6mm, gap=4mm, TR/TE=3s/6ms, spectral-spatial excitation, single-shot GRE, spiral readout with matrix size=64×64, 30 pairs of tag/control images with 2 dummy repetitions, no background suppression. Theoretical ASL SNR and SNR efficiencies (SNR/√2TR) were simulated for the above sequences. Labeling efficiencies of 0.98 and 0.85 were assumed for PASL and PCASL ⁹, respectively, both in simulation and in in vivo experiments. A standard ASL signal model ¹⁰ was used for CBF quantification with an assumed T₁ of blood of 1.65s 9. A common gray matter (GM) ROI was generated from the averaged ASL intensity maps and used in the analysis.

The measured ASL and CBF maps are shown in Fig. 2. The results from the simulation and the in vivo experiments were summarized in Table 1. Assuming similar noise levels across these experiments, the ASL signal should be a suitable indicator of SNR. The in vivo results showed an SNR and an SNR efficiency improvement of 54% in GM using two inversion modules in PASL, matched well with that predicted by the simulation, while keeping the quantification accurate. These results were also consistent with that from the reference PCASL experiment.As an alternative to the current implementation, two WS inversion modules can be used. Due to the geometry of the WS inversion, additional caution is needed to ensure that the interference between the labeling and the imaging regions is minimized. In principle, multiple (>2) modules can be used to further increase the total BD and improve the SNR gain while keeping the measurement quantitative. In the limit, the SNR of MM-PASL approaches that of PCASL and can even exceed it if the SNR difference due to higher labeling efficiency outweighs the stronger T₁ decay in MM-PASL. MM-PASL may be favored in applications such as ultra-high field ASL, where the SAR issue and possible strong off-resonance effects may be challenging for PCASL, and applications where the desired duty cycle of imaging acquisition is high.Multiple inversion modules can be used to increase the total bolus duration and the SNR of PASL, while keeping ASL signal quantitative.