Carotid artery stenting (CAS) is being increasingly used as an alternative treatment to carotid endarterectomy for carotid artery stenosis¹. Variable imaging modalities, including conventional angiogram, dynamic susceptibility contrast perfusion MR, and positron emission tomography, have been applied to investigate cerebral hemodynamic change after CAS and have revealed hemodynamic improvement following CAS^2,3. However, it remains unclear which hemodynamic changes occur in certain areas of the brain following CAS. Pseudo-continuous arterial spin labelling (pCASL) is a recent emerging MRI method for studying cerebral perfusion using radiofrequency (RF) pulses for the noninvasive labeling of endogenous water protons in the blood⁴ instead of injecting exogenous contrast agent or radioactive material. This technique is capable of detecting absolute cerebral blood flow (CBF) at the brain tissue level (ml/100g per minute) permits repeated measurement of CBF in a short interval, which is useful in patients receiving CAS⁵. However, the labeling process of pCASL is not strictly an adiabatic inversion and the labeling efficiency may be dependent on B₀ inhomogeneity or other factors⁶. Therefore, labeling position may need to be carefully applied when patient with CAS. The aim of this study was to experimentally determine the optimal labeling position for pCASL with minimal frequency shift caused by stent in exploring cerebral perfusion in the patient with CAS treatment.

Patients with symptomatic internal carotid artery (ICA) stenosis receiving CAS were included in the study. All MRI acquisitions were performed on a 3T clinical scanner (Discovery MR750, GE Healthcare, Milwaukee, USA) using an HD neurovascular array coil as the signal detection and whole body coil for RF transmission. Perfusion study was conducted using a pCASL with a 3D background suppressed fast-spin-echo stack-of spiral readout module, a TR=5327 ms, an TE=10.5 ms, FOV=24×24 cm, labeling duration=1.5 s, post labelling delay=2525 ms, no flow-crushing gradients, matrix=128×128, NEX=2, slice thickness =4 mm. In addition, labelling plane was 3.35-mm thick and placed 2.18 cm inferior to the lower edge of the scanning coverage. The CBF maps were generated on an Advantage Windows workstation using Functool software (version 9.4, GE Healthcare). For identifying stent position, three-slab time-of-flight magnetic resonance angiography (TOF-MRA) were acquired with TR/TE=21 ms/2.2 ms, flip angle=20°, FOV=23×20 cm, matrix=320×192, slice thickness=1.4 mm. B₀ map for evaluating the field inhomogeneity was acquired using WAter Saturation Shift Referencing (WASSR) method with 0.5 μT, 2×80 ms fermi pulses. To determine the optimal labeling position, the distance between stent and labeling plane was varied to be 2, 2.5, 3, 3.5, 3.75, 4, and 5 mm. Labeling efficiency is the important matter for CBF quantification and is dependent on B₀ inhomogeneity. However, the field homogeneity in the labeling plane was expected to be perturbed by the presence of the carotid stent. The local field inhomogeneity at the labeling plane of the pCASL pulse with flowing spin moving at velocity V along the Z direction can be modeled as a constant shift plus a linear Z gradient, and refer to them as the “off-resonance” (ΔB₀) and “off-resonance gradient” (ΔG), respectively. In a cycle of repeated pattern of RF and slice-selective gradient pulses used for pseudo-continuous inversion, the amount of error introduced into the phase accumulation (Δφ_error) between two RF pulses (duration between pulses denoted as δ) for the pCASL pulse sequence can be calculated⁶:

$$\triangle\phi \scriptsize{error}=\gamma\triangle B_{0}\delta+\frac{1}{2}\gamma V\triangle G\delta^{2}$$

The stent position was determined by TOF-MRA according to the loss of TOF signal caused by susceptibility artifact (Figure 1a). Sagittal brain anatomy with various color lines was used to demonstrate the labeling position for pCASL (Figure 1b). The estimated CBF map (Figure 1c) showed the lower CBF at ipsilateral hemisphere and back to normal compared to contralateral side as the labeling position of pCASL away from stent. Quantitative analysis revealed that the CBF value at contralateral side stays normal (52-66 ml/100g/min) as the change of labeling position, while the value at ipsilateral hemisphere decreased to 27.5 ml/100g/min at the distance of 2 mm and was gradually increase and reached to normal level at 3.75 mm of the distance between stent and labeling position (Figure 1d).Noncontrast pCASL method is very useful for evaluating cerebral hemodynamic change on the patient after CAS. However, prominent frequency shift due to the presence of stent (Figure 2) would lower the labeling efficiency for pCASL if the labeling plane at or close to the stent position. Consequently, this unexpected loss of labeling efficiency results in significant quantification errors. In this study, we demonstrated that perfusion signal will not be compromised when labeling position is 3.75 mm away from stent position. It is advised that the distance may vary with different kind of stent materials (cobalt alloy used in this case). Nonetheless, once the optimal distance has been determined in the beginning for the material, the same condition can be applied to CAS patient with the same stent.