Arterial spin labeling (ASL) is a non-invasive MRI technique that utilizes magnetically labeled arterial water as an endogenous tracer for quantitative measurement of the cerebral blood flow (CBF). Compared to the other perfusion imaging methods, such as 15O-water PET or dynamic susceptibility contrast (DCS) MRI, ASL does not involve exposure to ionizing radiations, does not involve injection of contrast agents, and can be repeated for multiple acquisitions in a single imaging session. Thanks to these known advantages, the technique has been widely used in various research and clinical applications [[1], [2], [3], [4]].

Because of the intrinsically low signal-to-noise ratio (SNR), three-dimensional (3D) imaging with segmented acquisition scheme in combination with background suppression (BS) has been recommended for perfusion imaging with ASL [1]. Currently, acquisition with gradient and spin echo (GRASE) [5,6] or fast spin echo (FSE) [7] readout is widely used for ASL perfusion imaging. Balanced steady-state free precession (bSSFP), which is known to provide the highest SNR per unit time [8], has also been proposed for high SNR, distortion-free ASL perfusion imaging [9]. However, bSSFP is sensitive to off-resonance effects, which result in banding artifacts in regions with severe B0 inhomogeneity [8], and effects from motion and flow when Cartesian sampling scheme is used for segmented acquisition scheme [10].

Radial sampling is a non-Cartesian data sampling scheme that can be beneficial for ASL imaging with bSSFP readout. Radial sampling provides robustness to effects from motion. In the case of radial sampling, over-sampling can be inherently applied in all radial directions without much increase in scan time, which increases the robustness to aliasing or wrap-around artifacts as well as Gibbs ringing artifacts [11]. Additional robustness to pulsatile flow artifacts can be achieved with radial sampling since it allows to distribute the effects of the artifacts in multiple radial directions, which can be especially beneficial for ASL perfusion imaging with bSSFP readout at 3 T [12]. Radial sampling scheme also displays incoherent aliasing artifacts from under-sampling [[12], [13], [14]] that are desirable for sparsity-constrained image reconstruction. Up to date, the radial sampling scheme has been applied to 2D ASL angiography and perfusion imaging using spoiled gradient-recalled acquisition (SPGR) readout [14] and ASL angiography imaging with bSSFP readout [13,15,16]. However, the usefulness of radial sampling scheme has not been thoroughly investigated for ASL perfusion imaging with bSSFP readout.

This work presents a new ASL perfusion imaging method that utilizes bSSFP readout with radial sampling for improved SNR and robustness to motion and off-resonance effects. More specifically, an ASL imaging sequence was developed with pseudo-continuous arterial spin labeling (pCASL) and bSSFP readout following a stack-of-stars sampling trajectory. Multiple phase-cycling technique was utilized to improve the robustness to off-resonance effects. Parallel imaging with sparsity-constrained image reconstruction was used to accelerate imaging or increase the spatial coverage. In vivo studies were performed on healthy volunteers at 3 T to demonstrate the advantage of the proposed method in terms of SNR, robustness to B0 inhomogeneity-induced banding artifacts, and respiratory motion-related artifacts.