Introduction

Fast spin-echo 3D stack-of-spirals is often used for the readout of a 3D pseudo-continuous arterial spin-labeling (3DpCASL) pulse sequence. The total scan duration of 3DpCASL acquisition is largely dependent on the number of interleaves (NI) used for the acquisition. The achievable maximum gradient amplitude (G_max) and slew rate (SR_max) of the gradient can limit the minimum NI used for the 3DpCASL acquisition and therefore the overall scan time for a given spatial resolution. This is because of the need to reduce the spiral duration per shot ($$$\rm{\tau}$$$) that is necessary to minimize the accrual of erroneous phases due to susceptibility, static B₀ inhomogeneity and eddy currents. A recently developed head gradient coil (MAGNUS - GE Global Research, Niskayuna, NY, USA) simultaneously achieves 200 mT/m and 500 T/m/s on each axis using a standard 1 MVA gradient driver in a clinical whole‐body 3.0 T system[1]. The MAGNUS has a higher peripheral nerve stimulation (PNS) threshold compared to conventional clinical whole-body 3.0 T system (CWBS)[2] with typical G_max and SR_max of 33-40 mT/m and 120 T/m/s respectively. In this work, we demonstrate that the spiral acquisition can be made more efficient on the MAGNUS to decrease the total scan time of a 3DpCASL acquisition without compromising the image quality compared to a CWBS acquisition. Perfusion signal-to-noise ratio (pSNR) efficiency and noise power per pixel ratio (NPPR) were used to compare the image quality of the CWBS and MAGNUS acquisitions.

Methods

All experiments were carried out on the MAGNUS gradient system with patches for zeroth and first order eddy current and concomitant gradient fields correction[3]. The G_max, SR_max, and receiver bandwidth (rBW) of the readout spiral acquisitions on MAGNUS and CWBS satisfied the azimuthal Nyquist criterion that prevents undersampling for the 3DpCASL spiral k-space[4]. At an rBW of 500 kHz, and spatial resolution of 3.5 x 3.5 x 4 mm³, the spiral readout of the 3DpCASL acquisition on the MAGNUS configuration was limited to G_max of 49 mT/m (far below the system G_max of 200 mT/m), and slew rate of 450 T/m/s. This was in contrast to CWBS configuration which was limited to G_max of 24 mT/m and utilized a slew rate of 110 T/m/s at an rBW of 250 kHz. The lower slew rate was chosen to mimic the lower PNS threshold for simultaneous x and y gradient waveforms on a CWBS. Acquired raw k-space phantom and volunteer 3DpCASL data were processed for image reconstruction using theoretically calculated spiral trajectories (nominal trajectories). Separately, a dynamic field camera composed of 16 ¹H probes (Skope, Zurich, Switzerland) was used to measure the true zeroth and first-order transverse k-space trajectories (measured trajectories) for 3DpCASL reconstruction as previously described[4]. Image reconstruction was carried out in MATLAB (Mathworks, Natick, MA, USA) using the Orchestra software development kit (GE Healthcare, Waukesha, Wisconsin, USA). Cerebral blood flow (CBF) was calculated using the perfusion-weighted (PW) and proton-density (PD) images as previously described in the ASL consensus[5]. pSNR efficiency was calculated as[4]: $$\mathrm{pSNR\ Efficiency = \frac{PW}{Noise\times\sqrt{NI\times \tau}}}$$ Wilcoxon signed-rank test for matched pairs was used to assess CBF and pSNR efficiency differences between the CWBS and MAGNUS acquisitions.

Phantom and Volunteer Experiments: Phantom images were acquired using the American College of Radiology (ACR) MRI quality control phantom. To verify the trajectory correction, 3DpCASL images of the ACR phantom were acquired using the CWBS and MAGNUS sequence parameters (Table 1). Five healthy volunteers (mean age: 37 years, 2 females) were imaged under an IRB approved protocol after a written informed consent was received. A two-minute wait time was observed between the CWBS and MAGNUS 3DpCASL scans for data consistency. Regional CBF and pSNR efficiency were quantified using semi-automatically drawn regions of interests (ROIs) in the lateral, anterior, and posterior cerebral regions of the brain.

Results and Discussions

The total scan time of the MAGNUS acquisitions were 48.3% lower than those of the CWBS (Table 1). Implementing k-space undersampling strategies can further reduce the 3DpCASL scan time, thereby facilitating dynamic 3DpCASL studies. The calculated NPPR was 0.96, indicating similar noise power per pixel on the CWBS and MAGNUS acquisitions. Reconstructed PD images using the nominal and measured spiral trajectories for the CWBS and MAGNUS acquisitions show comparable image quality (Figure 1). The measured trajectories only corrected for image rotations that become evident at higher SR_max[4]. Such image rotations can also be corrected by co-registration of the 3DpCASL acquisition with an anatomical image acquired in the same volume. Representative volunteer PD, PW and CBF images for the CWBS and MAGNUS acquisitions are shown in Figure 2. The calculated regional CBF values for the CWBS and MAGNUS acquisitions (Figure 3) were in the normal range for a healthy volunteer (40–78 mL/100g/min)[6]. The pSNR efficiency was consistently higher for all brain regions on the MAGNUS acquisition compared to the CWBS acquisition with a mean increase of 28.7% across all ROIs and subjects (Figure 4), which can be largely attributed to the reduced NI on the MAGNUS acquisitions.

Conclusions

The higher performance of the MAGNUS gradient system can be utilized to halve the scan time (compared to a CWBS acquisition) of a clinical 3DpCASL acquisition while preserving the image quality of the scan.

Acknowledgements

We gratefully acknowledge the expert knowledge of Dr. Marc Lebel in helping us understand the implementation of the 3D Pseudo-Continuous Arterial Spin Labelling pulse sequence.