Introduction

Spatially-selective Arterial Spin Labeling (ASL), such as pseudo-continuous ASL (PCASL), is sensitive to arterial transit times (ATT) because arterial blood is labeled in arteries outside the imaging volume (1). This can lead to artifactually decreased perfusion if the labeled bolus is not entirely delivered, yet can also result in macrovascular signal that provides arterial transit information regarding proximal pathology or collateral pathways. In contrast, velocity selective ASL (VSASL) is insensitive to ATT because it labels blood inside the image volume (2), allowing immediate label delivery without delay. Thus, VSASL accurately reflects cerebral blood flow (CBF) even in cases of delayed blood arrival. It has been shown that acquiring both VSASL and spatially-selective ASL yields complementary, clinically useful information about perfusion and arterial transit delays, but requires twice the scan time (3).

In this work, we introduce the VElocity and SPAtially selective (VESPA) ASL sequence, which combines VSASL and PCASL, in order to label two distinct arterial blood pools. A linearly independent control/label encoding scheme permits quantitative measurement of absolute CBF and ATT from decoded VSASL and PCASL signals. Other operating modes flexibly trade off macrovascular ASL signal, quantitation of CBF, and SNR.

Methods

Figure 1 provides the VESPA sequence timing diagram. The PCASL module is inserted in time between the first and second VSASL modules. The first module is spatially-selective to label spins within the imaging slab. The second module is non-selective and designed to suppress ASL signal flowing above the VSASL velocity cutoff (Vcut). A global pre-saturation resets the magnetization after the previous TR. The PCASL labeling plane is positioned directly below the inferior edge of the VSASL labeling volume. Label and control conditions are alternated in a linearly independent order (Figure 1, center). The different contrasts are encoded in the acquired images and decoded using simple image arithmetic (Figure 1, right). VSASL inflow time (TI) roughly equals PCASL labeling duration (LD).

Two time periods are considered for quantification: TI<Δt and TI>Δt, where Δt is the ATT. When TI<Δt, no PCASL signal has arrived and VSASL bolus duration is TI. In this case, CBF can be estimated but the ATT is only known to be >TI. When TI>Δt, the PCASL bolus accumulates in the tissue and the VSASL bolus duration is Δt. CBF and ATT are estimated by fitting the following signal model to the data:

$${\Delta}M_{VSASL}=M_0{\cdot}{\alpha}{\cdot}{f}{\cdot}{TI}{\cdot}{e^{-\frac{TI}{T_1}}}{\cdot}(1-e^{-\frac{T_{sat}}{T_1}}),{\qquad}{\qquad}{\qquad}{\qquad}{\quad}{TI<{\Delta}t}\\{\qquad}{\quad}=M_0{\cdot}{\alpha}{\cdot}{f}{\cdot}{{\Delta}t}{\cdot}{e^{-\frac{TI}{T_1}}}{\cdot}(1-e^{-\frac{T_{sat}}{T_1}}),{\qquad}{\qquad}{\qquad}{\qquad}{\quad}{{\Delta}t}<TI\\~\\{\Delta}M_{PCASL}=0,{\qquad}{\qquad}{\qquad}{\qquad}{\qquad}{\qquad}{\qquad}{\qquad}{\qquad}{\;\,}{TI<{\Delta}t}{\;}\\{\qquad}{\quad}=2{\cdot}{M_0}{\cdot}{\alpha}{\cdot}{f}{\cdot}[({\Delta}t-TI){\cdot}e^{-\frac{T_{sat}+TI}{T_1}}+{T_1}{\cdot}({e^{-\frac{{\Delta}t}{T_1}}}-{e^{-\frac{TI}{T_1}}})],{\quad}{{\Delta}t}<TI$$

M0 is the equilibrium magnetization of blood, α is the labeling efficiency (0.85 for VSASL (simulation), 0.85 of PCASL (3), f is CBF in ml/ml/s, T1 is the longitudinal relaxation time of blood (1.65s), and Tsat is the time between pre-saturation and VS1.

VESPA data were acquired on a healthy volunteer at 3T (GE MR750) with TI/LD = 1800ms and Vcut = 2cm/s. Accuracy of the VESPA signal decoding was evaluated by comparing to separate VSASL and PCASL scans, acquired by disabling the appropriate labeling module. CBF and ATT were calculated by fitting the VESPA signal model to the decoded VSASL and PCASL data. VESPA data were also acquired at three different TIs (1s, 1.25s, 1.5s) with vascular suppression partially disabled in the second VSASL module, to preserve macrovascular signal in PCASL. Other parameters: scan duration 5:20 minutes, TR 4s, Vcut 2cm/s, 2D-SE spiral readout, TE 15.8ms, 5 slices, voxel size 3.1x3.1x8mm3, 40 averages.

Results

Figure 2 demonstrates accurate decoding of the VESPA signal via comparison to distinct VSASL and PCASL acquisitions.

Figure 3 shows CBF and ATT maps after voxelwise fitting VESPA data with the proposed signal model.

Figure 4 demonstrates VESPA decoded VSASL and PCASL contrasts across three TIs with vascular suppression partially disabled. While VSASL signal is constant across TIs, the PCASL signal remains intravascular and flows deeper into the vascular tree as TI increases .

Discussion

VESPA ASL is a novel sequence that combines PCASL and VSASL into a signal scan. VESPA involves inserting PCASL labeling between two VSASL modules, which allows labeling of two distinct arterial blood pools. The bolus delivery delay between the PCASL and VSASL labels allow simultaneous quantitative measurement of CBF and ATT. A 50% scan time reduction relative to acquiring the data separately is achieved, in addition to obviating the need for multi-delay PCASL to estimate ATT.

Other modes of VESPA operation may also be of interest. Disabling vascular suppression of the second VSASL module preserves macrovascular signal (so-called “arterial transit artifact”, ATA), which has significant clinical value. Acquiring only dual label and dual control scans (A and D in Figure 1) provides the highest SNR, but only allows approximate calculation of CBF, and no direct information about ATT. Disabling vascular suppression in this mode retains macrovascular signal/ATA in addition to high SNR CBF signal.

Given the clinical need for complementary CBF and arterial transit information in short imaging times (for example, in acute stroke), VESPA has potential to become a robust clinical ASL technique, possibly even replacing PCASL as the criterion-standard.

Conclusion

VESPA ASL introduces a novel and efficient way to simultaneously acquire VSASL and PCASL data. A VESPA signal model to quantify CBF and ATT is presented with generation of human CBF and ATT maps.

Acknowledgements

Divya S. Bolar is supported by an RSNA Resident/Fellow Grant.