Introduction

Vascular suppression (VS) is a technique used to selectively eliminate intra-arterial signals, which are residual spin signals in a vascular compartment on acquisition, in ASL. Intra-arterial signals are observed when the PLD is shorter than the arterial transit time (ATT), leading to a quantification error in CBF. There are two types of VS techniques for flowing spin signal suppression. One is a diffusion-based technique, such as motion-sensitized driven-equilibrium (MSDE)¹; the other is the delays alternating with nutation for tailored excitation (DANTE) preparation pulse^2,3, which consists of a series of low flip angle (FA) nonselective radiofrequency pulses and gradients along the flow direction. Matsuda et al. reported that DANTE is a promising VS module in terms of image uniformity and may be an alternative method to MSDE³. It has also been reported that DANTE can suppress the flowing spin signal in relation to velocity⁴. Although DANTE has many advantages, the kind of spin compartment that is eliminated by DANTE has not been clarified. Meanwhile, it has been demonstrated that T₂ measurements of ASL signals can separate the spin compartments based on the differences in transverse relaxation times between the compartments⁵. We indirectly compared the signal characteristics between DANTE-ASL and the T₂ measurement method in a previous study⁶. In the current study, we directly determined the T₂ value of the ASL signal under the application of DANTE to verify its characteristics for VS.

Methods

Five healthy volunteers (5 men, 26 ± 3 years old) were scanned using a 3.0-T MRI system (Discovery MR750, GE Healthcare, USA). Two types of VS, i.e., DANTE and T₂ preparation pulses, were used (Fig. 1a and 1b). DANTE was set as the optimal condition (FA of 12.5° and gradient area of 10 µs T/m), as previously reported⁴. The effective TEs (eTEs) of T₂ preparation were 0, 40, 80, and 120 ms. The labeling duration was fixed (2.4 s), and the PLDs were set to 0.4, 1.2, and 2.0 s without DANTE and 0.8, 1.2, and 1.6 s with DANTE, respectively (Fig. 1c). Tissue T₂ values were calculated using saturation-recovery T₂-prepared proton-density images as a reference (T_{2_ref}). 3D T₁-weighted images were also obtained. After the application of the Gaussian kernel⁷, T₂ maps were computed on a voxel-by-voxel basis from multiple eTE images using the least-squares solution. In addition, multi-delay ASL signals were fitted to a single-compartment model to estimate the ATT and CBF. All images were spatially normalized to the MNI-space template using individual 3D-T₁ images. Gray matter was extracted to remove signal contamination from white matter and cerebrospinal fluid regions. T₂ values and the ATT were determined using a vascular territory atlas. We evaluated changes in T₂ values along the PLD and relation between T₂ values and the ATT in cases with and without DANTE, respectively.

Results

Fig. 2 shows the averaged ASL images of each PLD with different eTEs and calculated T₂ maps. In all PLDs, ASL signals decreased as the eTE increased. While T₂ values without DANTE (T_{2_noVS}) decreased as the PLD increased, T₂ values with DANTE (T_{2_DANTE}) did not change with the PLD. These findings were also supported by Fig. 3, which shows a relationship between the T₂ value and PLD. Moreover, T_{2_DANTE} was equivalent to T_{2_ref} in all PLDs. The averaged CBF and TT maps are shown in Fig. 4, indicating that CBF reduction and TT prolongation were observed with DANTE compared to those without DANTE. Although there is a positive correlation between T_{2_noVS} and the ATT, T_{2_DANTE} was neither dependent on the ATT nor PLD (Fig. 5).

Discussion

T_{2_noVS} decreased as the PLD increased. This result is similar to those of other ASL transverse relaxometry studies^2,7. Therefore, the feasibility of T₂ measurements with a 3D fast spin-echo-based ASL sequence was demonstrated. However, T_{2_noVS} at the short PLD was systematically smaller than that of the reported values because of the difference in read-out sequences. Since signal characteristics between gradient-echo- and spin-echo-based sequences are different, the proton population to be imaged using each sequence depends on its characteristics. While spins in larger vessels are imaged using gradient-echo-based sequences, spin-echo sequences can only image spins in small vessels⁸. These differences are attributed to the differences in baseline T_{2_noVS} with short PLDs. On the other hand, T_{2_DANTE} was not different among the PLDs and was equivalent to T_{2_ref} at any PLD. Therefore, DANTE sufficiently eliminates the spin signal in the vascular compartment, resulting in CBF reduction and TT prolongation. T_{2_noVS} was positively correlated with the ATT at short PLDs because in the longer ATT region, labeled spins were retained in the vascular compartment at the time of imaging with short PLDs. The correlation between the ATT and T_{2_noVS} became weaker as the PLD increased because longer PLDs enabled the labeled spins to enter the tissue compartment. However, T_{2_DANTE} was not significantly correlated with the ATT in either PLD. Therefore, DANTE may sufficiently eliminate spin signals in the vascular compartment at any PLD without tissue signal suppression. Namely, DANTE separates the vascular and tissue compartments.

Conclusion

DANTE can selectively eliminate intra-vascular signals from total ASL signals. We can use DANTE to separate the spin compartments in ASL.

Acknowledgements

No acknowledgement found.

References

1. Wang J, et al., Magn Reson Med, 2007; 58: 973-981. 2. Li L, et al., Magn Reson Med, 2012; 68: 1423-1438. 3. Matsuda T, et al., Magn Reson Imaging, 2018; 49: 131-137. 4. Fujiwara Y, et al., MAGMA, 2020; 33: 367-376. 5. Liu P, et al., Magn Reson Med, 2011; 65: 120-127. 6. Ishida S, et al., Proc ISMRM, 2020, 3285. 7. Schmid S, et al., Neuroimage, 2015; 123: 72-79. 8. Weisskoff RM, et al., Magn Reson Med, 1994; 31: 601-610.