Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Magn Reson Imaging. Author manuscript; available in PMC 2013 July 1.
Published in final edited form as:
PMCID: PMC3368036

CBF Measurements using Multidelay Pseudocontinuous and Velocity-Selective Arterial Spin Labeling in Patients with Long Arterial Transit Delays: Comparison with Xenon CT CBF



To test the theory that velocity-selective arterial spin labeling (VSASL) is insensitive to transit delay.

Materials and Methods

Cerebral blood flow (CBF) was measured in ten Moyamoya disease patients using xenon CT (xeCT) and MR imaging, which included multiple pseudo-continuous ASL (pcASL) with different post-label delays, VSASL, and dynamic susceptibility contrast (DSC) imaging. Correlation coefficient, root-mean-square difference, mean CBF error between ASL and gold-standard xeCT CBF measurements as well the dependence of this error on transit delay (TD) as estimated by DSC time-to-peak of the residue function (Tmax) were determined.


For pcASL with different PLD, CBF measurement with short PLD (1.5–2 s) had the strongest correlations with xeCT; VSASL had a lower but still significant correlation with a mean coefficient of 0.55. We noted the theoretically predicted dependence of CBF error on Tmax and on PLD for pcASL; VSASL CBF measurements had the least dependence of the error on TD. We also noted effects suggesting that the location of the label decay (blood vs. tissue) impacted the measurement, which was worse for pcASL than for VSASL.


We conclude that VSASL is less sensitive to TD than conventional ASL techniques and holds promise for CBF measurements in cerebrovascular diseases with slow flow.

Keywords: ASL, VSASL, perfusion, Moyamoya, Xenon CT, DSC


Arterial spin labeling (ASL) has become a valuable technique for cerebral blood flow (CBF) measurements, as it is completely non-invasive and does not employ ionizing radiation (1,2). This technique not only allows measurement of baseline CBF, but can also be used to study cerebral function (3). It uses magnetically tagged water protons in the arterial blood by magnetically inverting or saturating their spins. In conventional ASL such as pulsed ASL (4,5), continuous ASL (2,6), and pseudocontinuous (sometimes called pulsed continuous) ASL (pcASL) (7), tagging is performed at a location proximal to the region of interest (ROI). Then, an image is acquired after allowing a post-label delay time (PLD) during which the blood flows from the tagged region into the ROI. The CBF is proportional to the modulation of longitudinal magnetization (Mz) caused by the labeled blood within the measurement slice. Due to the nature of spatial separation of tagging and imaging region, there is inevitably a delay between the time arterial blood is tagged and the time it reaches the ROI, termed transit delay (TD). Variations in TD between different brain regions are a major source of error in CBF quantification using ASL. These problems can be mitigated somewhat by using longer PLD (8) and/or creating a tagging scheme with well-defined bolus length (9). However, both approaches ultimately fail in regions with very long TD because of significant SNR reduction related to the T1 decay of the tagged blood during the PLD. This problem has limited clinical enthusiasm for the use of ASL in pathologies with long TDs, particularly acute ischemic stroke and large vessel vasculopathies.

Recently, velocity-selective ASL (VSASL) has been proposed, which tags water protons based on their velocity rather than their spatial location, and has been postulated to overcome the problem of long TD (10). However, this advantage has not yet been demonstrated clinically. Moyamoya is a cerebrovascular disease caused by progressive stenosis of large arteries at the base of the brain, which leads to extensive collateral formation. However, these collaterals are imperfect, and patients are prone to both acute ischemic stroke and brain hemorrhage (11). Because of these collaterals, there are large regions of the brain with markedly prolonged TD, as measured by angiography qualitatively and bolus contrast dynamic susceptibility contrast (DSC) (quantitatively) (12). Being a chronic disease of otherwise healthy young patients, Moyamoya disease is well suited for demonstrating the effects of TD on ASL CBF measurements as well as to determine whether VSASL can eliminate or mitigate this problem.

In the present study, we compared the performance of pcASL (with and without suppression of vascular signals from large vessels) acquired at multiple PLD times and VSASL in a group of Moyamoya disease patients using xenon CT (xeCT) CBF measurement as a gold standard. In particular, we examine the CBF error as a function of TD, which is estimated using the bolus dynamic susceptibility contrast (DSC) measurement time-to-peak of the residue function (Tmax). We hypothesized that VSASL would provide a more accurate measurement of CBF than pcASL that is independent of arterial transit delay.



The study was approved by the institutional review board and was Health Insurance Portability and Accountability Act (HIPAA) compliant. Ten adult patients with Moyamoya disease were included in the current study. Four of them were male, and the age ranged from 25 to 58 years, with mean (standard deviation) of 43.0 (13.3) years. All were symptomatic, four had unilateral and six had bilateral disease. Patients were enrolled in the study prospectively as part of their pre-operative assessment for possible external-to-internal carotid (EC-IC) artery bypass procedure.

MR Imaging

All MR imaging was performed at 3T (GE Healthcare MR750, Waukesha, WI, USA) and included anatomic imaging, multiple ASL sequences as described below, and a bolus perfusion sequence using dynamic susceptibility contrast (DSC) imaging.

A high-resolution T1-weighted image was acquired using a 3D T1-weighted inversion recovery SPGR sequence with parameters of: TR/TE/TI = 9.2/3.7/400ms, flip angle = 13°, matrix = 256×204×144, voxel size = 0.94×0.94×1.2mm, parallel imaging reduction factor = 2. These images were used for segmentation of the brain into gray and white matter as described below.

pcASL with and without vessel suppression (the former referred to as pcASL(sup)) were performed using a labeling period (TL) of 1500 ms, followed by a PLD of 1s, 1.5s, 2s, 2.5s, or 3s, with a 3D background-suppressed fast-spin-echo stack-of-spiral readout module with 8 in-plane spiral interleaves. Thirty-six 4mm slices were acquired with an in-plane resolution of 3mm. Multiple tag/control pairs were performed to improve the signal-to-noise ratio (SNR) (NEX = 3 for PLD = 1s, 1.5s and 2s; 4 for PLD = 2.5s; 5 for PLD = 3s). For the vessel suppression, a T2 preparatory module was inserted immediately prior to the excitation pulse (13). This was designed to dephase vessels with a velocity above approximately 10 cm/s, based on the gradient-time value of 0.3 s/mm (14); this also had the effect of adding very minimal diffusion-weighting to the image (b=10 s/mm2). CBF quantification (in ml/100 g/min) was performed in the same manner for both pcASL and pcASL(sup), using an automated script, as follows:


where λ is the brain:blood partition coefficient (0.9 ml/g), PLD is the post-label delay, T1blood is the T1 of arterial blood at 3T (1.664 s) (15), α is the labeling efficiency (0.85), TL is the labeling duration (1.5 s), ΔS is the ASL difference signal, and S0 is the proton-density signal intensity. The term (1−exp(−2.0s/1.5s)) in the numerator reflects the presence of a saturation pulse that is applied in the proton-density images and allows conversion between measured MR signal (S0) and the unperturbed longitudinal gray matter magnetization. The decay of the blood during the TE period is not included, since the effective TE is very short (2.5 ms). More details can be found in (16).

For VSASL, velocity-selective (VS) tagging was performed using two 90-degree hard pulses and a pair of hyperbolic secant refocusing pulses. Gradient pulses were played between the RF pulses to perform flow encoding and VS spin saturation. These pulses are essentially transparent to stationary spins, as the spins are tipped down to the transverse plane by the first 90-degree pulse, refocused by the adiabatic pulses, and brought back to the longitudinal position by the second 90-degree pulse. For moving spins, they will have accrued a phase at the time immediately before the application of the second 90-degree pulse, and therefore will not be brought back to the longitudinal position. The amount of phase accrued depends on the velocity of the spin and parameters of the gradient pulses; for laminar flow, the spins are essentially saturated above a certain velocity, determined by the gradient parameters (10). In the current study, the cutoff velocity was set to 2cm/s (aligned in the slice-select direction), and the inflow time was 1.6s. Background suppression was performed using two non-selective inversion pulses played at 1550ms and 450ms before the acquisition to minimize signal from spins with T1 between 700 and 1600ms. A single-shot 2D spin-echo spiral imaging module was used with the following parameters: TR/TE = 3000/15ms, FOV = 22cm, Matrix size = 64×64, slice thickness/gap = 6mm/2mm, 5–9 slices in the axial plane. Fifty pairs of tag/control images were acquired, in addition to an image with minimal contrast (TR/TE 2 s/15 ms) for coil sensitivity correction and a proton density image (TR/TE=infinite/15ms) for absolute CBF calibration. The total acquisition time was about 5 min. CBF quantification is performed using the following equation:


where S0B is the MR signal of fully relaxed arterial blood, TI is the delay between the VS pulse and the excitation pulse (1.6 s), T1blood is the T1 of the arterial blood (1.664 s at 3T), TE is the echo time (15 ms), T2 is the T2 of arterial blood (275 ms), and the Mz,blood is the magnetization of the tagged blood (10).

DSC was performed using gradient-echo echo planar imaging during passage of 0.1 mmol/kg of gadobenate dimeglumine (Multihance; Bracco Diagnostics, Princeton, NJ) delivered using a power injector at 4 mL/sec. Image readout was performed using a single-shot echo-planar imaging sequence with pulse repetition time/echo time of 1800/40 ms and flip angle of 60°. Twenty axial slices of 5mm thickness with no gap covered the entire supratentorial brain. In-plane resolution was 1.7mm (matrix 128×128, field of view 220mm). Sixty dynamic scans were acquired. The slices were aligned with the superior orbitomeatal axis. Automated arterial input function (AIF) and venous output function detection followed by a delay-insensitive deconvolution, using a regularization threshold of 15% of the maximum singular value (17,18), to create maps of CBF, cerebral blood volume (CBV), mean transit time, and AIF corrected time-to-peak of the residue function (Tmax).

Xenon CT

Xenon CT (xeCT) perfusion imaging was performed using a GE 8-detector scanner, using previously described parameters (16). XeCT is considered a gold standard for CBF measurement, as inhaled xenon gas is a freely diffusible, stable tracer (19). The xeCT protocol includes four slices of 10-mm thick beginning at the level of the basal ganglia, aligned with the superior orbitomeatal axis. Two baseline images were acquired, followed by 6 images spaced at 45 sec intervals during which the patient breathes 28% xenon gas through a facemask. The mask has a sensor that measures exhaled xenon concentration, which is taken to represent the arterial input function, which is a good approximation in young patients without an arterial-alveolar gradient. CBF was calculated using the Kety-Schmidt method implemented by the manufacturer’s dedicated commercial software (Diversified Diagnostic Products, Inc., Houston, TX, USA). Routine noncontrast CT was also acquired, which covered the entire brain with a resolution of 0.49mm and slice thickness of 4mm, and was used for the purpose of co-registration.

Image Analysis

Image analysis was performed using SPM8 (Wellcome Trust Center for NeuroImaging, UCL, UK) and custom code running on Matlab 2009b (Mathworks, Natick, MA, USA). All ASL and xeCT CBF images as well as Tmax maps were registered to the T1-weighted image. The xeCT CBF image was registered to T1-weighted image using the whole brain noncontrast CT during an intermediate step. All MR images were then re-sliced into the space of xeCT CBF image for all further analysis. The T1-weighted image was segmented to produce gray matter (GM) and white matter (WM) probability images, which give the probability of being GM and WM respectively for each voxel. These probability images were also resliced to the space of xeCT CBF image. The GM mask was created by assigning a voxel to GM if it had the greatest probability of being GM using the formula (pGM>pWM) & (pGM>(1-pGM-pWM)), where pGM and pWM represent the probability of a voxel being GM and WM respectively.

The following analyses were performed to assess the performance of ASL methods against the gold standard xeCT CBF method: 1) correlation coefficients, root-mean-square (RMS) CBF difference and Bland-Altman statistics (including mean difference, and standard deviation of the difference) were calculated between ASL methods and xeCT using a large voxel size of 1×1×1cm3, which was made up by averaging the individual smaller voxels. The analysis was repeated for the entire imaged brain (WB) region and GM region. The 1×1×1cm3 voxel size was used to account for the effects of imperfect co-registration and differences in the acquisition resolution of different CBF measurement methods, and has the basic effect of spatial smoothing of the data. 2) Mean CBF values measured with the ASL methods were calculated for the WB and GM regions and were compared with that of xeCT CBF measurements. 3) To study the dependence on arterial transit delays, the GM mask generated from the segmentation of T1-weighted image was divided into four regions according to the Tmax value (region A, 0–2s; region B, 2–4s; region C, 4–6s; region D, >6s), and the mean CBF values in the four regions were calculated for all the CBF measurement methods. Also, the number of voxels falling into each of the above groups was determined. ΔCBF was calculated as the difference between ASL and xeCT CBF measurements, as in part (2) above.


Qualitative Observations

An example of the pcASL, pcASL(sup), and xeCT CBF maps obtained is shown as Figure 1. VSASL CBF maps, however, had relatively lower SNR due to the use of saturation rather than inversion labeling pulses, poorer background suppression, and artifacts related to stimulated echoes. Visually, the expected PLD dependence of pcASL and pcASL(sup) CBF measurements was obvious in the regions with prolonged TD, usually the territories of the anterior and middle cerebral arteries. CBF underestimation was observed in pcASL and pcASL(sup) with short PLD (e.g. 1s), and this discrepancy was somewhat mitigated at longer PLD’s. The pcASL sequences also demonstrated the expected arterial transit artifact (ATA) in regions with slow flow, which was also somewhat mitigated with long PLD. Given the lower SNR, it was difficult to be confident by visually assessing the images regarding the question of whether slow flow artifacts remained present in the VSASL images, though they were definitely observed in some patients (Figure 2).

Figure 1
CBF maps; from left to right: xeCT, VSASL, pcASL without (top) and with (bottom) vessel suppression for PLD of 1s, …, 3s (left to right). CBF is underestimated by pcASL or pcASL(sup) with short post-label delay (PLD) in regions with slow blood ...
Figure 2
CBF map of xeCT (a), pcASL(sup) with post-label delay (PLD) of 1s, 2s and 3s (b–d), and VSASL (e), as well as Tmax image (f). Arterial transit artifacts were still visible in some cases with long PLD of 3s (arrows). At long PLDs, regions with ...

Correlation, RMS, and Bland-Altman Analysis

A typical set of scatterplots relating ASL and xenon CT CBF measurements are shown in Figure 3. The mean correlation coefficients between ASL methods and xeCT ranged between 0.50 and 0.65, and pcASL with PLD of 1.5–2s having the strongest correlation with xeCT (Table 1), regardless of whether vessel suppression or not was used. VSASL had a lower but still significant correlation with xeCT with a mean coefficient of 0.55.

Figure 3
CBF images (bottom panel, from left to right: xeCT, pcASL with PLD of 1s, 1.5s, … 3s, VSASL); scatterplots (top panel, the horizontal axes are xeCT CBF value, vertical axes are ASL CBF values).
Table 1
The mean and standard deviation of CBF values in the entire image brain (WB) and gray matter (GM) of the different CBF measurements techniques, as well as the correlation coefficients and root-mean-square (RMS) difference, mean difference (Mean Diff) ...

The RMS difference with xeCT was the smallest for pcASL(sup) with PLD of 1.5–2s, and the largest for VSASL. A decrease in the RMS difference, connoting more accurate CBF measurements, was observed for pcASL with increasing PLD. Bland-Altman standard deviation of the difference showed that CBF measured using pcASL(sup) with PLD of 1.5–2s was most similar to xeCT. The standard deviation of the difference for pcASL ranged 11.7 – 13.0 ml/100g/min, whereas it was 15.0 ml/100g/min for VSASL

Mean CBF Values

Table 1 shows the mean CBF values in the WB and GM regions of the different CBF measurements techniques. The difference between ASL and xeCT CBF values, ΔCBF, is plotted in Figure 4. The mean CBF from all ASL techniques with all PLD values were within +/− 5 ml/100g/min of the mean xeCT CBF, except for pcASL(sup) with a PLD of 1 s, which underestimated xeCT CBF by 5–10 ml/100g/min. For the same PLD, pcASL(sup) showed closer estimation of CBF to xeCT than pcASL, except for PLD of 1s, For pcASL, PLD of 1s underestimated the mean CBF in WB and GM regions, while larger PLD’s overestimated these values. VSASL also overestimated the mean CBF values, with a mean error of about 2 ml/100g/min.

Figure 4
Mean and standard error of ΔCBF values in the entire imaged brain (WB) and gray matter (GM) regions of the different CBF measurements techniques. Δ CBF represents the difference between the ASL CBF value and the xeCT CBF value in the same ...

Voxels were classified to regions with different Tmax values (0–2 sec, 2–4 sec, etc.) as shown in Figure 5. CBF values in GM and WB from the different ASL measurement methods were calculated and are shown in Table 2. XeCT CBF decreased monotonically with increasing Tmax. The differences in CBF values between ASL methods and xeCT were plotted in Figure 6 for GM. For pcASL with PLD of 1–1.5s, overestimation of CBF in GM regions with small Tmax value was found, with expected CBF underestimation in regions with long Tmax. With increasing Tmax the overestimation became smaller and eventually becomes underestimation of the true CBF value. For higher PLD’s (>2.5s), a reverse trend was observed, with CBF underestimation in regions with smaller Tmax and overestimation of CBF in regions with larger Tmax. Similar trends were observed for pcASL(sup), except that consistent underestimation of CBF was found in GM regions with Tmax<2s for all PLD’s. For VSASL, underestimation of CBF in regions with Tmax <=2s and overestimation in regions with Tmax >2s were found, though the magnitude of the Tmax dependence was the smallest of any of the ASL sequences tested. Figure 2 shows an example of overestimation of CBF of the ASL methods in region with long TD.

Figure 5
(a) Tmax image (in units of seconds). (b) shows the gray matter matter ROI, with different Tmax values: blue, 0–2s; green, 2–4s; yellow 4s–6s; red, >6s.
Figure 6
Mean and standard error Δ CBF values (ASL measurement – xeCT measurement) in GM regions as a function of transit delay, estimated from Tmax values. For pcASL with short post-label-delay (PLD) of 1–1.5s, CBF is underestimated in ...
Table 2
The mean and standard deviation of CBF values in the gray matter (GM) and whole brain (WB) with different transit delay, as measured by the bolus DSC parameter Tmax. The unit for CBF value is ml/100g/min.


In this study, we compared pcASL (with and without vessel suppression) at multiple PLD’s and VSASL with gold standard xeCT CBF measurements in patients with Moyamoya disease, a steno-occlusive disease of the large arteries at the base of the brain. Different metrics were used, including correlation coefficients, RMS ΔCBF difference, Bland-Altman statistics, and mean CBF values. In particular, we examined the dependence of the CBF levels based on TD as measured by the bolus DSC parameter Tmax. A previous study has shown that there is a relationship between prolonged Tmax and lower CBF values measured using xeCT, a finding also seen in this study (20). We also observed reduced sensitivity to TD with increasing PLD for pcASL, as evidenced by a progressive reduction in regions with ATA and progressive increase in regions showing parenchymal ASL signal (Figure 1). Alsop et al. first proposed the use of long PLD’s to mitigate transit delay errors (8); our findings support this idea. However, as longer PLD’s are used (and in this study we used PLD values up to 3 sec, longer than any reported in the literature to our knowledge), the SNR of the measurement decreases due to the longitudinal decay of the label with T1. For this reason, we increased the amount of signal averaging in the current study to try to maintain a near constant SNR in the images. This required longer imaging times (8 min for the PLD 3 s scans), which may not be feasible in clinical practice.

Furthermore, we present initial findings in these patients using VSASL, a methodology that theoretically removes the sensitivity of ASL to transit delay; as such, it would address one of the major concerns of quantitative ASL – errors incurred due to unknown and variable transit delay. Standard VSASL suffers from poor SNR due to the ability to saturate rather than invert moving spins, though there is active research in this area (21). We also found that VSASL implemented in this patient population suffered from poor SNR beyond what would be expected. This may reflect insufficient background suppression (22), slice-profile effects, and stimulated echoes created between slices. While single-slice acquisitions or widely-spaced slices can avoid these issues, they are not applicable to the need to acquire volumetric CBF data sets which can be co-registered to compare with other modalities.

Voxel-based correlations between all of the ASL sequences and xeCT were moderate, and higher than that seen in a previous study performed at 1.5T (16), likely due to the higher field strength at which the current study was performed. pcASL with PLD of 1.5s was found to have the highest correlation coefficient with xeCT; however, visual inspection indicated that CBF underestimation, manifested as low signal or signal voids, was still found in regions with long Tmax, which was confirmed by quantitative analysis. The explanation of why this relatively short PLD had the best correlation is probably because it had the highest SNR while still allowing a reasonable amount of time for the blood to reach the capillary of most normal tissue. A post-label delay of between 1.5 and 2 s has been suggested as appropriate for pulsed ASL measurements in both young and elderly normal subjects (23). For a shorter PLD of 1s, both pcASL and pcASL(sup) underestimate CBF, because a large portion of the tagged blood does not reach the capillary bed within this time frame. For PLD greater than 1s, pcASL(sup) showed more accurate estimation of CBF than pcASL, suggesting that suppressing signal from blood in large vessels improves CBF quantification. The identification of ATA in clinical patients can be a helpful marker of collateral flow, but if quantitation is the goal, it is probably best to suppress large vessel signal.

The accuracy of ASL measurements varies with different Tmax values. In the ideal case, Tmax represents the time interval between the peak of the residue function in the tissue compared with that of the AIF, and therefore, it indicates how fast the blood reaches the tissue. pcASL with PLD lower than 1.5 s underestimated the xeCT CBF levels for regions with long Tmax (>4 sec). This is consistent with the fact that the tagged blood did not have enough time to reach these regions during the short PLD. Interestingly, for PLD >=2 s, pcASL sequences tended to overestimate the flow in these regions. We believe this paradoxical overestimation might be due to differences in the relaxation of the label in these regions, which will be primarily within the blood rather than the tissue. Thus, in these regions, the label experiences relaxation with a T1 value of the arterial blood, which is longer than that of the tissue. Consistent with this hypothesis is that in the early arriving regions, CBF is slightly underestimated, since the label spends proportionally longer time in the tissue and therefore experiences faster decay. To account for the differences in the decay rate of the label, an estimate of the exchange time is required, which, while likely to be correlated with arrival, is very challenging to measure on a voxel-by-voxel basis. During the current study, we used the general kinetic model (24) and assumed that the spins remain in the blood and experience relaxation with T1 value of the blood for the entire course of the acquisition.

The error of the CBF measurement with VSASL had the least dependence on Tmax, supporting the notion that VSASL does lead to improved quantification of CBF over a wide spectrum of transit delays. This is likely due to the closer spatial location between the labeled spins and the imaged voxel. VSASL also suffers from errors due to the unknown time at which the label leaves the blood and exchanges with the tissue. However, as VSASL uses velocity-selective tagging, a shorter PLD can be used, which minimizes the time over which the error accumulates. Although undesirable, the CBF overestimation of VSASL in regions with large Tmax suggests VSASL with moderate TI allows measurement of CBF in regions that may be otherwise inaccessible using conventional ASL using similar PLD.

VSASL uses saturation tagging as opposed to inversion tagging, used in pcASL and pcASL(sup), which reduces the signal by 50%. To compensate for this, thicker slices were prescribed, which reduced the resolution of VSASL. Current implementation of VSASL also employs a 2D readout module, which limits the performance of background suppression. These factors substantially reduced the image quality of CBF maps from VSASL. This is reflected as moderate correlation coefficient, the largest RMS difference and standard deviation of the mean difference between VSASL and xeCT CBF measurements among all the ASL methods. Recently, there are attempts to develop velocity-selective inversion tagging as well as to employ 3D readout modules (25). Finally, VSASL tags spins based on velocity rather than spatial location and is expected to tag spins at the imaging plane with the typical cut-off of velocity of 2cm/s. In extreme pathological cases where the blood flow is so slow that tagging only occurs at upstream arteries that are spatially separated from the imaging region, CBF underestimation may still occur because of non-zero TD (26). Lower cut-off velocity can be used in this situation; however, this increases the diffusion weighting and may introduce additional errors. How much this affects CBF quantification of VSASL and what is the best imaging strategy in pathology is a subject of further study.

In conclusion, we compared VSASL, pcASL, and pcASL(sup) with gold standard xeCT CBF measurement in a group of patients with long arterial transit delays. VSASL allows the use of moderate post-label delay for the measurement of CBF in regions with slow flow and had the least dependence of the error on Tmax. Two sources of quantification errors exist for pcASL: variable arterial arrival delays of different regions, and faster T1 decay of spins in tissues than in the blood, which appears to be important for very long PLD. The latter source of error also applies to VSASL, but to a lesser degree. Further development of VSASL holds promises for its application in cerebrovascular diseases with slow flow.


Grant support: NIH (P41RR009783, R01-NS066506, R01-NS047607-05, R01-EB002711), Lucas foundation, Oak Foundation.

The authors thank Eric Wong, PhD, from University of California, San Diego for providing the VSASL sequence and reconstruction scripts.


1. Detre JA, Wang J, Wang Z, Rao H. Arterial spin-labeled perfusion MRI in basic and clinical neuroscience. Current opinion in neurology. 2009;22:348–355. [PubMed]
2. Williams DS, Detre JA, Leigh JS, Koretsky AP. Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci U S A. 1992;89:212–216. [PubMed]
3. Feng CM, Narayana S, Lancaster JL, et al. CBF changes during brain activation: fMRI vs. PET Neuro Image. 2004;22:443–446. [PubMed]
4. Edelman RR, Siewert B, Darby DG, et al. Qualitative mapping of cerebral blood flow and functional localization with echo-planar MR imaging and signal targeting with alternating radio frequency. Radiology. 1994;192:513–520. [PubMed]
5. Kwong KK, Chesler DA, Weisskoff RM, et al. MR perfusion studies with T1-weighted echo planar imaging. Magn Reson Med. 1995;34:878–887. [PubMed]
6. Zaharchuk G, Ledden PJ, Kwong KK, Reese TG, Rosen BR, Wald LL. Multislice perfusion and perfusion territory imaging in humans with separate label and image coils. Magn Reson Med. 1999;41:1093–1098. [PubMed]
7. Dai W, Garcia D, de Bazelaire C, Alsop DC. Continuous flow-driven inversion for arterial spin labeling using pulsed radio frequency and gradient fields. Magn Reson Med. 2008;60:1488–1497. [PMC free article] [PubMed]
8. Alsop DC, Detre JA. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab. 1996;16:1236–1249. [PubMed]
9. Wong EC, Buxton RB, Frank LR. Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II) Magn Reson Med. 1998;39:702–708. [PubMed]
10. Wong EC, Cronin M, Wu W-C, Inglis B, Frank LR, Liu TT. Velocity-selective arterial spin labeling. Magn Reson Med. 2006;55:1334–1341. [PubMed]
11. Suzuki J, Takaku A. Cerebrovascular “moyamoya” disease. Disease showing abnormal net-like vessels in base of brain. Archives of neurology. 1969;20:288–299. [PubMed]
12. Calamante F, Ganesan V, Kirkham FJ, et al. MR perfusion imaging in Moyamoya Syndrome: potential implications for clinical evaluation of occlusive cerebrovascular disease. Stroke. 2001;32:2810–2816. [PubMed]
13. Zaharchuk G, Shankaranarayanan A, Alsop DC. Removing large vessel contamination from arterial spin label MR perfusion images using T2 preparation. Proceedings of Radiological Society of North America; Chicago, IL. 2008. pp. VN31–10.
14. Ahn CB, Lee SY, Nalcioglu O, Cho ZH. The effects of random directional distributed flow in nuclear magnetic resonance imaging. Med Phys. 1987;14:43–48. [PubMed]
15. Lu H, Clingman C, Golay X, van Zijl PC. Determining the longitudinal relaxation time (T1) of blood at 3. 0 Tesla. Magn Reson Med. 2004;52:679–682. [PubMed]
16. Zaharchuk G, Straka M, Marks MP, Albers GW, Moseley ME, Bammer R. Combined arterial spin label and dynamic susceptibility contrast measurement of cerebral blood flow. Magn Reson Med. 2010;63:1548–1556. [PMC free article] [PubMed]
17. Wu O, Ostergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. Tracer arrival timing-insensitive technique for estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magn Reson Med. 2003;50:164–174. [PubMed]
18. Straka M, Albers GW, Bammer R. Real-time diffusion-perfusion mismatch analysis in acute stroke. J Magn Reson Imaging. 2010;32:1024–1037. [PMC free article] [PubMed]
19. DeWitt DS, Fatouros PP, Wist AO, et al. Stable xenon versus radiolabeled microsphere cerebral blood flow measurements in baboons. Stroke. 1989;20:1716–1723. [PubMed]
20. Olivot JM, Mlynash M, Zaharchuk G, et al. Perfusion MRI (Tmax and MTT) correlation with xenon CT cerebral blood flow in stroke patients. Neurology. 2009;72:1140–1145. [PMC free article] [PubMed]
21. Wong EC, Guo J. Velocity Selective Inversion Pulse Trains for Velocity Selective Arterial Spin Labeling. Proceeding of 17th Meeting of International Society of Magnetic Resonance in Medicine; 2009. p. 620.
22. Garcia DM, Duhamel G, Alsop DC. Efficiency of inversion pulses for background suppressed arterial spin labeling. Magn Reson Med. 2005;54:366–372. [PubMed]
23. Campbell AM, Beaulieu C. Pulsed arterial spin labeling parameter optimization for an elderly population. J Magn Reson Imaging. 2006;23:398–403. [PubMed]
24. Buxton RB, Frank LR, Wong EC, Siewert B, Warach S, Edelman RR. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med. 1998;40:383–396. [PubMed]
25. Wong EC, Wu W-C, Bykowski J. Whole Brain 3D Velocity Selective Arterial Spin Labeling using a Background Suppressed Flow Weighted Spiral Fast Spin Echo Acquisition. Proceeding of 14th Meeting of International Society of Magnetic Resonance in Medicine; 2006. p. 3427.
26. Wu W-C, Wong EC. Intravascular effect in velocity-selective arterial spin labeling: The choice of inflow time and cutoff velocity. Neuro Image. 2006;32:122–128. [PubMed]