Perfusion Imaging Using Arterial Spin Labeling

ORIGINAL ARTICLE

Perfusion Imaging Using Arterial Spin LabelingXavier Golay, PhD,* Jeroen Hendrikse, MD,† and Tchoyoson C. C. Lim, MD*

Abstract: Arterial spin labeling is a magnetic resonance method forthe measurement of cerebral blood flow. In its simplest form, the per-fusion contrast in the images gathered by this technique comes fromthe subtraction of two successively acquired images: one with, andone without, proximal labeling of arterial water spins after a smalldelay time. Over the last decade, the method has moved from theexperimental laboratory to the clinical environment. Furthermore,numerous improvements, ranging from new pulse sequence imple-mentations to extensive theoretical studies, have broadened its reachand extended its potential applications. In this review, the multiplefacets of this powerful yet difficult technique are discussed. Differentimplementations are compared, the theoretical background is summa-rized, and potential applications of various implementations in re-search as well as in the daily clinical routine are proposed. Finally, asummary of the new developments and emerging techniques in thisfield is provided.

Key Words: perfusion imaging, arterial spin labeling, spin tagging,cerebral blood flow, quantitative MR imaging

(Top Magn Reson Imaging 2004;15:10–27)

ARTERIAL SPIN LABELING: THE BASICSArterial spin labeling (ASL) is based on the principles of

the indicator dilution theory. This theory was used for the firsttime by Kety and Schmidt to measure cerebral blood flow(CBF) in humans.1,2 Their method was based on the measure-ment of the arteriovenous dynamics of a freely diffusibletracer, nitrous oxide (N2O), passed into the arterial systemthrough the respiration. Since the concentration of arterialtracer, ca(t) increases faster that the venous concentration cv(t),the regional CBF can be estimated using Fick’s principle:

dCT�t�

dt= CBF�ca�t� − cv�t�� (1)

with CT(t) the tissue concentration of the tracer. In this model,the venous concentration of the tracer is corrected for the dif-

ference in blood and tissue water contents: Cv(t) = CT(t)/� with� = blood–brain partition coefficient, which can be differentfor different tissues.3

While the Kety-Schmidt method is rarely used in pa-tients today, their theory has been applied for the estimation ofrCBF in numerous techniques, mainly using radioactive trac-ers, such as H2

15O in positron emission tomography (PET)4,5

or 99mTc-ethylcysteinate-dimer in single-photon emissioncomputed tomography.6 Similar methods using intra-arterialinjection of non-1H NMR-visible tracers have been proposedusing Kety-Schmidt’s quantification theory. For recent reviewarticles of the MR methodologies to measure perfusion, seeBarbier et al,7 Calamante et al,8 or the book chapter by Alsop.9

In ASL, no extrinsic tracer is used. Instead, a selectivepreparation sequence is applied to the arterial water spins, andthe perfusion contrast is given by the difference in magnetiza-tion or apparent relaxation time induced by the exchange ofthese labeled spins with the tissue of interest10–12 (Fig. 1).

The original ASL method for labeling arterial spins wasproposed by Williams et al.13 In this work, the authors used asingle volume coil at the level of the common carotid arteriesto invert the spin adiabatically in a way similar to an earlierpublished angiography technique.14 The degree of labeling ofthis ASL method is highly dependent on the adiabatic labelingcondition13,15–17:

1�T1,1�T2 ��G � v�B1 ��B1 (2)

with T1, T2 the spin-lattice and spin-spin relaxation times re-spectively (in s), G = gradient strength (in mT/m), v = the nor-mal nominal mean blood velocity in the vessels, � = gyromag-netic moment ratio for 1H and B1 is the amplitude (in mT) ofthe applied RF amplitude. Typical values for the degree of in-version achieved with such method vary between 80% and90%, depending on the choice of G and B1.15,16,18 While con-tinuous adiabatic inversion is still used today to measure per-fusion, new sequences appeared soon after, in which the mag-netic labeling of the spins is performed at once over a widespatial range to get better inversion efficiencies.10,19,20 Thesemethods are referred to as pulsed arterial spin labeling (PASL)as opposed to continuous arterial spin labeling (CASL) for theoriginal scheme using flow-driven adiabatic inversion of thearterial spins.

From the *Department of Neuroradiology, National Neuroscience Institute,Singapore; and the †Department of Radiology, University Medical Center,Utrecht, The Netherlands.

Address correspondence and reprint requests to Xavier Golay, PhD, Depart-ment of Neuroradiology, National Neuroscience Institute, 11 Jalan TanTock Seng, Singapore, 308433 (e-mail: [email protected]).

Copyright © 2004 by Lippincott, Williams & Wilkins

10 Top Magn Reson Imaging • Volume 15, Number 1, February 2004

Indeed, it was the development of PASL techniques thatmoved the whole field from laboratory animals to the hu-mans.11 The successful application of such methods to nonin-vasively determine brain perfusion imaging in humans at-tracted a larger number of people to work in this field, and thenumber of scientific or clinical papers yearly published on thistopic exploded from a few papers published yearly until 1996to over 20 from the year 1999 onwards (Fig. 2).

MR TECHNIQUES FOR ASL

Continuous Arterial Spin Labeling (CASL)Techniques

The very first implementation of ASL was done usinglong RF pulses in combination with a slice-selection gradientto adiabatically invert the arterial magnetization.11,13,15 It wasimmediately recognized that a potential confounding factor ofthis method is that it induces magnetization transfer (MT) ef-fects.21 Indeed, the application of long off-resonance RFpulses will induce a saturation of the backbone of large mac-romolecules in the tissue included in the imaging coil, due totheir broad resonance frequencies.22 Once saturated, this largemacromolecular pool will then exchange its magnetizationwith that of the “free” water, inducing in that way changes inboth T1 relaxation time and magnetization similar to those pro-duce by ASL and therefore leading to an overestimation of theperfusion effects.11 To overcome this problem, a distal label-ing of the magnetization is performed in the control experi-ment to produce identical MT effects. This method works well

for single-slice perfusion imaging; however, the long RFpulses used in such techniques are applied concurrently with agradient pulse and therefore render MT effects dependent onthe slice position.

Other techniques are necessary to control for these ef-fects in a global way. The simplest one is to use a second coilto perform the adiabatic continuous inversion.23,24 In such asituation, the RF saturation pulse does not affect the tissue ofinterest due to the localized B1 field of the second labeling coil.The use of both separate gradient and RF coils for ASL has alsobeen proposed and would be more useful for human brain per-fusion imaging in head-only scanners, or other systems withrestricted gradient coils.25 Furthermore, the use of additionalcoil for labeling has the added feature of being able to selec-tively label the left or right common carotid artery indepen-dently, therefore allowing some kind of perfusion territorymapping.26,27

Another possibility, as originally proposed by Alsop andDetre, is to control globally for MT effects using a sinusoidalmodulation of the RF waveform.28 The application of such anRF pulse together with a gradient will continuously saturatetwo planes at the same time, resulting theoretically in a net zerolabeling during the control phase. The RF power is adjusted sothat the root mean square power is identical in both labelingand control acquisitions to produce identical MT effects. Themain drawback of this method is that imperfections in thedouble-inversion plane usually result in a lower labeling effi-ciency (∼70%).28 An example of an image acquired with this

FIGURE 1. Schematic description ofthe principles of freely diffusibletracer theory. Inverted magnetiza-tion (white) comes from the arterialtree (white arrow) and diffusesthrough the blood–brain barrier atthe capillary level. There, spins areexchanged with the tissue magneti-zation (gray), and reduce its local in-tensity. The degree of attenuation isa direct measure of perfusion. Re-maining tagged magnetization aswell as exchanged water moleculesflow out of the voxel of interestthrough the venous system (gray ar-row).

Top Magn Reson Imaging • Volume 15, Number 1, February 2004 Perfusion Imaging Using Arterial Spin Labeling

© 2004 Lippincott Williams & Wilkins 11

technique is shown in Figure 3, in which three slices (out of 15acquired) are shown for both a healthy 10-year-old girl and afemale patient suffering from Rett syndrome, a rare neuroge-netic disease.29

Pulsed Arterial Spin Labeling (PASL)TechniquesSymmetric PASL Techniques

Apart from the continuous ASL techniques, pulsed ASL(or PASL) techniques have met with increasing success,mainly due to the ease of implementation and reduced practicaldifficulties as compared with those encountered in CASL,such as those related to the long (2–4 seconds) continuouswave-like excitation pulse required. PASL techniques can beseparated in two categories, whether the labeling is appliedsymmetrically or asymmetrically with respect to imaging vol-ume. Generally, the arterial labeling is performed much closerto the acquisition plane in PASL than in CASL, to get an ad-equate signal-to-noise ratio.

The symmetric PASL sequences are based on a schemeoriginally proposed by Kwong et al,10 and independently pub-lished by Kim20 and Schwarzbauer et al,30 and are referred toas flow-sensitive alternating inversion recovery (FAIR) se-quences, according to Kim’s acronym.20 In this particularscheme, an inversion-recovery sequence is performed twice:once with and once without slice-selection gradient to label thearterial spins. The perfusion-weighted signal comes then fromthe difference in signal intensity due to the absence of invertedarterial spins in the slice-selective experiment with respect tothe non-slice-selective inversion recovery sequence.31,32 Since

FIGURE 2. Bar plot of the number of papers provided by a simple Medline search using “Arterial Spin Labeling” or “Arterial SpinTagging” as key words.

FIGURE 3. Comparison of absolute CBF in a normal girl (A) anda female patient with Rett Syndrome (B) of identical ages (10.3years). Images were acquired at 1.5 T in 8 minutes. The aver-age whole brain CBF is 105 mL/min/100 g for the volunteerand 59 mL/min/100 g for the patient, demonstrating a generalreduction in brain metabolism. Notice also the microcephalyof the patient, one of the typical clinical manifestations for thisdisease.29

Golay et al Top Magn Reson Imaging • Volume 15, Number 1, February 2004

12 © 2004 Lippincott Williams & Wilkins

both labeled and nonlabeled images are acquired with identicalRF pulses, MT effects are counterbalanced over the centerslice. However, problems in the control for these effects mayarise when using multislice schemes.33,34 MT-related prob-lems will, however, be less important than in CASL35 becauseof the reduced RF power needed in such sequences.

Soon after the publication of the FAIR sequence, severalother methods based on similar symmetric labeling were de-veloped:• In the UNFAIR sequence, meaning “uninverted flow-

sensitive alternating inversion recovery,” an additional non-selective pulse is applied just after the first inversion pulse toyield noninverted static signal in the volume of interest.36,37

This sequence, developed independently by Berr et al,38,39

is also called EST (extra-slice spin tagging).• The FAIRER acronym refers to two different sequences,

aimed at solving different problems: in the first one,FAIRER means “FAIR excluding radiation damping,” andconsists of a regular FAIR sequence after which a small con-tinuous gradient is turned on to avoid radiation damping ef-fects, affecting mostly animal experiments at high fieldstrength40,41; in its second definition, FAIRER means“FAIR with an extra radiofrequency pulse” and consists of aFAIR sequence followed or directly preceded by a 90° RFpulse to null the static signal in the volume of interest. Thissequence was designed to be used in pulmonary perfusionimaging, in which cardiac triggering is necessary.42,43

• Another version of the FAIR sequence is the FAIRESTtechnique or “FAIR exempting separate T1 measurement,”in which a slice-selective saturation recovery acquisition isadded to the standard FAIR scheme and used mostly in com-bined BOLD and CBF fMRI experiments.44

• To achieve perfusion quantification with a short repetitiontime (TR), Pell et al modified the original implementationof the FAIR sequence by performing a global (nonselective)saturation pulse prior to the actual FAIR sequence.45

• Finally, in the BASE sequence, meaning “unprepared BAsisand SElective inversion,” a selective inversion-recovery se-quence and a sequence without any preparation pulse arealternatively acquired, and perfusion is calculated from theexpected signals coming from both experiments.46

Asymmetric PASL Techniques

Apart from these symmetrical schemes, another type ofPASL techniques has been developed, based on an originalsequence called EPISTAR for “echo-planar imaging with sig-nal targeting by alternating radiofrequency pulses.”19 InEPISTAR, the tagging of arterial water magnetization is per-formed using an inversion pulse in a slab proximal to the vol-ume of interest, after (or before) saturation of the static spins toavoid any contamination of the labeling slab into the measuredvolume. In the control experiment, the slab is positioned at thesame distance distally to the volume of interest, using a simple

polarity inversion of the slab-selective gradient. This controlsfor magnetization transfer effects in a way similar to the origi-nal CASL sequence.11 Based on this original sequence, the fol-lowing other schemes have been proposed:• First, the original EPISTAR sequence has been modified by

its authors to render it multislice capable.47 In this secondversion of EPISTAR, the labeling inversion pulse is re-placed by an adiabatic fast passage inversion pulse of twicethe length of a conventional 180° pulse, leading to an inver-sion of the arterial magnetization, while the control scan iscomposed of two 180° pulses of half the length of the label-ing one canceling each other and therefore leaving the spinsuninverted in a way similar to the sinusoidal CASL control.

• In the PICORE sequence (proximal inversion with controlfor off-resonance effects), the labeling experiment is similarto that of EPISTAR while the slab-selective labeling gradi-ent is turned off during the control experiment, thereforecontrolling for off-resonance MT effects in the central sliceof the volume of interest.35

• In the TILT sequence (transfer insensitive labeling tech-nique), both labeling and control experiments are performedusing two self-refocusing 90° pulses; however, with thephase of the second pulse shifted by 180° in the control ex-periment, therefore yielding a 90° − 90° = 0°.48,49 Thisscheme allows controlling for MT artifacts over a widerange of frequencies, as both 90° + 90° and 90° − 90° pro-duce identical MT effects, in a way similar to the sinusoidalmodulation in continuous arterial spin labeling28 or to themodified EPISTAR technique.47

• Finally, another scheme has been developed recently, usinga “Double Inversion with Proximal Labeling of bOth taggedand control iMAges” (DIPLOMA).50 In this sequence, thetagging is performed by two consecutive inversion pulses:the first one applied off-resonance, without gradient selec-tion, and therefore inducing only MT effects, and the secondone being a slab-selective on-resonance pulse inverting thespins in the region of interest. In the control experiment, twoconsecutive inversion pulses are played out on-resonance,inducing identical MT effects but no net change in the mag-netization, in a way similar to the BASE sequence.46

Technological Improvements: A Clear Definition ofthe Bolus

In most of the PASL techniques described above, one ofthe major technical difficulties in getting an absolute quantifi-cation of the CBF (see next section) comes from the potentiallyvariable delay or arterial transit time �t taken by the blood totravel between the labeling slab and the measurement volume.In an attempt to improve the definition of the bolus and there-fore simplifying perfusion quantification, Wong et al35,51,52

came up with simple modifications of PASL sequences thatcircumvent the problem of variable �t by the introduction ofadditional saturation pulses. Their methods, dubbed QUIPSS



(I & II) for “QUantitative Imaging of Perfusion using a SingleSubtraction” simply add a slab-selective 90° dephasing pulseafter a time TI1 in the volume of interest (for the first version)or in the same position as the labeling bolus (in the 2nd ver-sion). In the QUIPSS II approach in particular, the bolus widthand transit time can be regulated in a way similar to CASL,using the saturation pulse to limit the length of the tag. Bycarefully timing the signal detection after the saturation pulse,transit-time insensitive perfusion images can be acquired, evenfor multiple slices. Furthermore, the use of a saturation pulse inboth labeled and control acquisitions reduces the artifactuallyhigh perfusion-like signal coming from remaining labeled in-travascular spins.

A further improvement has recently been introduced byLuh et al,53,54 in which the saturation pulse is repeatedly ap-plied at the distal end of the labeled volume to improve both thedefinition of the tagged slab, as well as the cancellation of in-travascular signal. The method is called Q2TIPS for “QUIPSSII with Thin-slice TI1 Periodic Saturation.”

Dynamic and Pseudo-ContinuousASL Techniques

One of the drawbacks of ASL, especially when appliedto functional MRI (see next Section on Neuroscience Applica-tions) is its relative low inherent temporal resolution, due to thesubtraction process involved. Based on this consideration,Wong et al55 showed that by acquiring the images at an earlyTI, one gets the perfusion signal from the previous repeti-tion, thereby improving the nominal temporal resolution ofPASL35 by a factor of 2. The conditions to perform such anexperiment can be summarized as follows: TR > �, TI > � +�tmax, TI − TT < �tmin, with � = temporal width of the bolus,�tmin (resp �tmax) = minimum (resp. maximum) arterial transittime from the bolus slab to the imaging volume.

Based on a similar observation, Silva and Kim pro-posed a modified CASL technique, in which short acquisi-tions are interleaved with an RF labeling pulse in such away that the labeling duty cycle is still large enough to get adecent labeling efficiency. The control scans are then per-formed in a separate experiment. In that implementation, theyused a TR of 108 milliseconds, with a labeling time of 78 mil-liseconds for a total readout time (using EPI) of 30 millisec-onds.56

Using a similar method, Barbier et al57,58 came upwith a dynamic ASL (DASL) technique, in which the re-sponse of the tissue magnetization to a periodically vary-ing labeling pulse is measured. This technique alleviates theuse of a control experiment to observe the effect of the ASLpulse on the tissue; rather, the tissue response function is di-rectly fitted to a theoretical model, providing a simultaneousestimation of both CBF and arterial transit time in a singlemeasurement.

Nonsubtraction MethodsFinally, another class of methods has recently been in-

troduced in which perfusion-weighted images can be obtainedwithout the need of a control acquisition. These nonsubtractionmethods are based on the concept of background suppressionusing multiple inversion pulses,59 first introduced for ASLtechniques by Ye et al for multishot three-dimensional perfu-sion imaging.60 There are mainly two sequences published forsingle-shot nonsubtraction spin labeling techniques:• The first one, dubbed SEEPAGE for “Spin Echo Entrapped

Perfusion image,” starts with a 90° selective pulse, followedby a train of 180° pulses maintaining the tissue magnetiza-tion around zero. The train of pulses is maintained longenough for the non-saturated blood to enter in the slice andprovide a perfusion signal.61

• In the second one, called SSPL for “Single-Shot PerfusionLabeling,” a double-inversion recovery pulse is applied tonull both CSF and tissue inside the volume of interest, whilekeeping the perfusion signal minimally perturbed.62

Both methods have however the drawback of being dif-ficult to quantify.

ABSOLUTE QUANTIFICATION IN ASL

Equations for CBF QuantificationTo get an absolute quantification of CBF from ASL mea-

surements, it is necessary to modify the Bloch equations toinclude the exchange terms between the static tissue magne-tization and the flowing labeled arterial magnetization13:

dMT�t�

dt=

MT0 − MT�t�

T1T+ f�MA�t� −

MT�t�

� � (3)

with MT = tissue magnetization, MT0 = steady state equilibrium

value of MT, T1T = tissue longitudinal relaxation rate, and MA =arterial magnetization per ml of blood and f = CBF in (mL/g/s).This equation is valid under the following assumptions63:• That perfusion is provided by an uniform plug flow to the

tissue.• That exchange between labeled water and tissue magnetiza-

tion can be modeled as a single well-mixed compartment,more particularly that the venous concentration of labeledmagnetization is equal to the tissue concentration divided bythe blood-brain partition coefficient �, ie, Mv(t) = MT(t)/�.

Originally, the first models used for absolute quantifica-tion of ASL did not take into account the effect of arterial tran-sit time �t in the measured signal.10,11,13,24,31 Buxton et al63

demonstrated theoretically that it is one of the most importantparameters needed for the quantification of both PASL andCASL techniques.

Generally, a supplementary condition is also implied toget an analytical solution to Eq. 3: that labeled water is com-pletely and spontaneously extracted from the intravascularspace as it enters the tissue, ie, that labeled blood magnetiza-



tion relaxes with the longitudinal relaxation time of arterialblood T1B until it reaches the tissue, and with T1T thereafter.With these assumptions, the general solution for PASL63 andCASL64 techniques can written as:

For PASL

��MT�t� = �0 0 � t � �t

2�MT

0

�f�t − �t�e−R1BtqPASL�t� �t � t � � + �t

2�MT

0

�f� e−R1BtqPASL�t� � + �t � t

(4)

with

qPASL�t� = �e�R1t�e−�R1�t − e−�R1t�

�R1�t − �t��t � t � � + �t

e�R1t�e−�R1�t − e−�R1��+�t��

�R1�� + �t � t

(5)

and R1B = (T1B)−1 is the longitudinal relaxation rate of arterialblood, �R1 = R1B − R1app is the difference between the relax-ation rate of arterial blood and apparent relaxation rate of thetissue, and R1app = R1T + f/� is the apparent relaxation rate ofthe tissue, which depends on CBF.

For CASL

��MT�t� =

�0 0 � t � �t

2�MT

0

�f T1appe−R1B�tqCASL�t� �t � t � � + �t

2�MT

0

�f T1appe−R1B�t e−R1app�t−�−�t�qCASL�t� � + �t � t

(6)

with

qCASL�t� = �1 − e−R1app�t−�t� �t � t � � + �t

1 − e−R1app� � + �t � t(7)

The main reason to express these two results (Eqs. 4 and6) using a dimensionless q function is to show the similaritybetween the equations governing both CASL and PASL meth-ods. It can be demonstrated that CASL measured in peaksteady state conditions provides a perfusion signal �MT atmost e times larger than PASL, if T1app = T1B.63,65 These qfunctions describe also the effects of the different relaxationtimes as well as the venous clearance of the labeled magneti-zation in PASL, and the approach to steady state in CASL. Inboth cases, and after a sufficiently long time in CASL, bothexpressions are close to 1 and can therefore be neglected in afirst approximation.

The Effect of Magnetization TransferSeveral papers have dealt with the effects of magnetiza-

tion transfer on the absolute quantification.64,66–68 But themost complete theoretical description is depicted in a paper byMcLaughlin et al.67 In this paper, the authors investigated inparticular the questions about both terms in the ratio�M(t)/T1app that appears in the equations providing f in CASLtechniques. They based their observation on the followingequations describing a four-pool model:

�dMT�t�

dt=

MT0 − MT�t�

T1T− kf MT �t� + kbMM�t�

+ f E�f� �MA�t� −MT�t�

� �dMM�t�

dt=

MM0 − MM�t�

T1M+ kf MT �t� − kbMM�t�

(8)

In which E(f) is the water extraction fraction, dependingon the CBF, the index M refers to the magnetization of thebound macromolecular pool,22 and kf and kb are the forwardand backward magnetization transfer rate constants betweenthe free and bound water pool, respectively, obeying the fol-lowing relation at equilibrium: kfMT

0 = kbMM0 . Without entering

too much into the details of this model, the general flavor of itsresults is that under a wide range of conditions, reasonable andconsistent choices for �M(t) and T1app should facilitate calcu-lation of correct values for CBF in CASL experiments.

Restricted Exchange and MultipleCompartment Modeling

Another possible improvement of the original modelbased on the Kety-Schmidt equations for ASL is to include theeffects of restricted water exchange between intravascular andextravascular compartments (ie, finite capillary water per-meability) as well as capillary contributions to the ASL sig-nal.69–72 In these four papers, different additional terms havebeen added to the original equations, and different values ofunderestimation or overestimation of CBF have been found,depending on the assumptions made. Zhou et al70 used a2-compartment model in its expression similar to Eq. 8, withthe exchange terms expressed between intravascular and ex-travascular compartments, and applied their results to measureCBF from a FAIR sequence at 4.7 T. They obtained a goodcorrelation between their FAIR measurements and micro-sphere measurements of perfusion. St. Laurence et al71 workedout similar equations and obtained similar results for CASL,showing that generally, the single-compartment model pro-vides relatively accurate CBF values at 1.5 T (within 5%) forgray matter. Parkes and Tofts69 found an overestimation ofCBF in gray matter at 1.5 T of more than 60% (and 17% inwhite matter) in FAIR-like sequences. They attributed the dif-ferences of their model with the previous ones to differences in



the choice of T1B and some differences in the modeling of thevenous outflow.

ARTIFACTS AND PROBLEMS

SNR and Imaging ProblemsOne of the major problems of ASL techniques is their

intrinsic low signal-to-noise ratio (SNR). Indeed, the measured�M/M0 is often on the order of 1% or less, and subtractionerrors are frequent, often caused by improper selection of theinversion width73 or motion artifacts in clinical studies of dif-ficult patient populations.74 For this reason, several back-ground-suppression schemes have been developed based onthe possibility to cancel the signal of more than one tissue us-ing multiple inversion pulses.59 This background-suppressionmethod has been used for nonsubtraction ASL,61,62 as well asfor multishot three-dimensional PASL60 and CASL.9 Otherways to increase the SNR in perfusion-weighted images havebeen proposed, such as maximizing the SNR in both controland labeling acquisitions, for example, by carefully choosingthe minimal usable B1 field in CASL to minimize MT effectsas much as possible.75 But the best way to ensure a propersubtraction of labeled from control scans is to use fast (single-shot) imaging techniques. In most applications of ASL in thebrain, single-shot EPI10,19,20 or spiral imaging34,60,76,77 hasbeen used. However, strong susceptibility artifacts, especiallyat the basis of the brain, render such readouts less than perfect.Therefore, other readout techniques have been proposed to al-leviate this problem, based on line-scan,78,79 or fast-spinecho80–82 imaging. In applications of ASL outside the brain,very few studies have used single-shot EPI methods, mainly inkidney perfusion imaging83 or musculoskeletal perfusion im-aging.84–86 Instead, simple gradient-echo or fast spin-echo-based sequences, both of which are susceptibility insensitivetechniques, have been used for measuring cardiac perfu-sion,87–91 skeletal muscle perfusion,92–94 pulmonary perfu-sion,42,43,95–98 or perfusion in the urogenital system.39,80,99–102

Transit Time and Vascular ArtifactsAnother main problem of ASL techniques comes from

the remaining labeled signal in the vasculature. Upon subtrac-tion of labeled from control experiments, remaining invertedblood will result in spuriously high perfusion values, espe-cially in presence of a low resolution imaging technique. Forthis reason, the first CASL sequences had already imple-mented small diffusion gradients (diffusion b-value = 1–10seconds/mm2) to eliminate any remaining arterial signal.13,103

Based on a similar idea, Pell et al104 implemented a scheme inwhich a DEFT diffusion preparation sequence is performedprior to a Turbo-FLASH readout. Recently, by comparing thesignal obtained with and without crusher gradients, Wang et alcould obtain an estimate of the arterial transit time.105

Another solution to strongly reduce the effects of intra-vascular contamination is to adjust the predelay between con-

tinuous labeling and readout.28,64 This latter method has theadvantage of not reducing the already small SNR available andis easier to implement. Furthermore, it renders the method in-sensitive to variable transit times. In PASL, the QUIPSS-basedtechniques are effective in a similar way.51–53

RF Pulse ShapesFinally, one of the major implementation problems en-

countered, especially in PASL techniques, is related to the useof very well-defined inversion pulses, as profiles not sharpenough will result in a reduced labeling efficiency as well aspossible contamination by the labeling pulse of the volume ofinterest.49,106–109 For this reason, very sharp labeling pulsesare needed, mainly adiabatic fast passage47,106 or frequencyoffset-corrected inversion pulses.36,108,110 Finally, the TILTsequence uses self-refocusing concatenated 90° Shinar-Leroux optimized RF pulses111 to maintain the profile effi-ciency of the 90° pulses.49 One drawback of this technique isthat it can only be used under very good shimming condi-tions.50

APPLICATIONS IN NEUROSCIENCE

Functional ASL Versus fMRIOne of the original reasons for the development of ASL

was the drive to develop a fully noninvasive technique allow-ing measuring the local changes in CBF related to a functionaltask.19,20,112 Indeed, ASL techniques developed in parallelwith functional MRI (fMRI) techniques based on the bloodoxygen level-dependent (BOLD) contrast.112–114 Similarly tofMRI, early functional ASL studies measured perfusionchanges in primary motor20,32,35,115–118 or visual46,119–122 ac-tivation tasks. Only few studies have been used in cognitivetasks.123 However, fundamental differences exist betweenboth techniques: while T2*-weighted BOLD fMRI measuresan epiphenomenon secondary to local increases in oxygen-ation, ASL provides a direct absolute measurement of CBFchanges. ASL should be therefore more reproducible amongsubjects and generally over longer periods of time,124–126 de-spite its intrinsic lower SNR.34,127,128 Furthermore, it is lessprone to artifacts coming from large draining veins, as its sig-nal is mainly of arteriolar and microvascular origin129,130 andis therefore usable for very high-resolution functional map-ping, such as columnar resolution functional ASL.131 Then ithas to be noted that at very high field (9.4 T), functional areasdepicted by both spin-echo BOLD and CBF were found to bepretty well co-localized, which can be attributed to the com-plete loss of coherence of blood signal for long echo times atsuch high field strength.132 However, the necessity of the sub-traction procedure makes ASL less favorable for event-relatedfMRI than BOLD, necessitating the use of complex reorderingprocedures133,134 or of no-subtraction ASL techniques.62 Apossible way to increase the temporal resolution in ASL can be



achieved using a so-called stroboscopic acquisition, in whichthe frequency of the activation paradigm is shifted with respectto the frequency at which the data are acquired.135

On the other hand, CASL is not as easy to interpret asBOLD. In particular, changes in arterial transit time have beenmeasured using ASL and can be misinterpreted as elevatedperfusion upon activation.136–140 In particular, Gonzales-At etal showed that depending whether the model used a singletransit time or a transit time distribution, the transit times var-ied from 10% to 25% during motor or visual activation in func-tional CASL experiments.137

Furthermore, these changes in transit time are differentin each voxel and are dependent on the arterial content of thevoxel.139,140 As a demonstration of this effect, Figure 4 depictsthe arterial transit time at which voxels in the visual cortexshow maximum cross-correlation with a simple visual stimu-lus paradigm using the Turbo-TILT sequence (see EmergingTechniques).140 At short delay times, the maximum cross-correlation is located near the areas surrounding the centralsulcus, where large arteries are present. For longer delay times,this area is shifted to the posterior and lateral part of the visualcortex.

Generally, detection of functional ASL changes cannotbe compared directly to BOLD fMRI, as the noise character-istics of ASL are very different from those of BOLD as shown

by Wang et al141; they found that fMRI time series are more orless independent in time and did not show spatial coherencevarying across temporal frequencies. This finding has impli-cations for the spatial smoothing of functional perfusion data.

Metabolic and Oxygen Consumption StudiesOne of the first demonstrations of a correlation between

oxygen metabolism and blood flow was demonstrated in theskeletal muscle by Toussaint et al.94 using 31P spectroscopy asa direct measure of muscle metabolism.

In the brain, Kim and Ugurbil demonstrated using com-bined FAIR and BOLD experiments that changes in oxygenconsumption during activation are very low.127 Later, the samegroup demonstrated a linear relationship between CBF andBOLD changes in a graded visual stimulation paradigm.120

These experiments were repeated and reinterpreted using aCO2-based titration procedure142 to obtain independently ab-solute metabolic rates of oxygen by the same group143 and byHoge et al.144,145 A more detailed study of the effects of hy-peroxia, hypercapnia, and hypoxia on both CBF and interstitialoxygen tension was also published by Duong et al.146 Similarvalidation experiments were performed using CASL on a ratmodel by Silva et al.147

ASL has been also used to further study the temporalcharacteristics of activation-related CBF changes. In particu-

FIGURE 4. Application of Turbo-TILTto functional imaging.140 The colormap highlights the postlabeling de-lay at which the voxels in the visualcortex show maximum cross-correlation with a simple visualstimulus paradigm. At short postla-beling delay times, the maximumcross-correlation is located near theareas surrounding the central sulcus.For longer delay times, this area isshifted to the posterior and lateralpart of the visual cortex.



lar, based on the mismatch between the temporal characteris-tics of the BOLD and ASL signals following brain activation,Buxton et al116 developed a simple biomechanical model ex-plaining these changes, the so-called “balloon model.” Con-currently, ASL was used by Silva et al148 to understand theearly negative changes in the BOLD signal (called the “early-dip”),149 which is thought to be closer to the site of activa-tion.150 However, the authors of this study could not demon-strate any evidence of such early loss in hemoglobin oxygen-ation preceding the rise in CBF and concluded that increases inCBF and oxygen consumption were likely to be dynamicallycoupled in their animal model. Other applications of ASL inthe determination of the functional characteristics of theBOLD signal included, among others, a study on the linearityof the temporal dynamics of ASL.151

Pharmacological ApplicationsASL is also of great interest in the study of various drugs,

in both humans and animals, as it does not require the injectionof another compound that might interfere with the tested phar-macological agent.152 Furthermore, ASL provides absolutequantification of a physiological parameter (CBF), unlikeT2*-weighted BOLD, in which changes in signal can be inter-preted as coming either from changes in oxygen extraction,blood flow, or metabolism. In particular, a twofold to fourfoldincrease in CBF has been demonstrated after administration of2-chloroadenosine, a potent cerebrovasodilator, to rats.153,154

Similarly, elevated CBF was measured in volunteers followinginjection of acetazolamide.155 The same drug was also used inmeasuring the cerebrovascular reserve in patients with ische-mic symptoms referable to large artery cerebrovascular steno-sis of the anterior circulation.156 Drugs with opposite effects,such as indomethacin157,158 or the anesthetic gas isoflurane,159

have also been tested using CBF in both humans and animals,respectively. Finally, Born et al measured the changes in CBFin the visual cortex in children under sedation,160 trying to in-vestigate the origin of the negative BOLD signal observed of-ten in such case and found a slight decrease in CBF upon visualactivation.

CLINICAL APPLICATIONS

Validation of ASL Prior to Clinical StudiesPrior to any clinical application of ASL, several animal

studies reported general behavior of these sequences in termsof reproducibility and comparison with “gold-standard” inva-sive methods. In a study on rats, Zhou et al showed a goodcorrelation between values measured with FAIR and radio-active microsphere CBF measurement methods.70 Ewing et alalso demonstrated a good correlation of local CBF measuredby ASL on direct comparison with quantitative autoradiogra-phy.161 Finally, Pell et al recently used the oxygen clearancemethod and compared it with a PASL technique (FAIR).162

They also found a very good correlation between CBF valuesmeasured by both techniques, although impaired slightly bythe large scattering in their FAIR measurements. Furthermore,human studies of cerebral perfusion using ASL have showngood within-subject variation in normal adults,163 as well as inchildren.164 Lia et al165,166 have compared the results of CBFmeasured independently in health volunteers by contrast re-agent-based techniques to ASL measurements and found goodcorrelation between both techniques (r = 0.86 over 8 subjects).Finally, Ye et al have published a study comparing the resultsof ASL to 15O-labeled PET and found an extremely good con-cordance between both methods in the cortical gray matter(ASL: 64 ± 12 mL/100 g/min and PET: 67 ± 13 mL/100g/min).167 However, perfusion in the white matter was under-estimated, in agreement with what could be expected fromtheoretical studies using extended models.71

As a fully noninvasive method, ASL might be particu-larly advantageous where a technique not necessitating intra-venous injection is required, especially when repeated scansare foreseen. In animal models of transient cardiac arrest, ASLhas been successfully used to assess postresuscitation cerebralreperfusion.168–171 By making repeated ASL measurements,some authors have demonstrated that early resuscitation in-creased reperfusion after cardiac arrest and improved sur-vival.168 This technique has now been used to study the effectsof different treatment strategies to improve cerebral recov-ery.169,171

Pediatric StudiesASL techniques, with better safety and image quality,

are attractive for studying cerebral perfusion in the pediatricpopulation and pose less of an ethical problem than nuclearmedicine studies.74,164 Children have generally a higherblood flow than adults with a peak at around 10 years ofage.164,172,173 This higher physiological CBF will lead to in-creased SNR in ASL and reduce the artifacts caused by slowarterial transit time. Conversely, in dynamic contrast reagentstudies in children, the increased CBF, coupled with the re-duced total dose of contrast media, would lead to technicaldifficulties in sampling the first pass bolus.

Recent studies from Wang et al164 and Oguz et al74 alsofound an increase in the cerebral blood flow in normal childrencompared with adult controls. Furthermore, in the latter report,children with sickle cell disease were identified with reductionin CBF without large arterial disease on MR angiogram, dem-onstrating the potential for noninvasive prospective identifica-tion of children at risk for stroke and other complications.

Cerebral IschemiaIn animal studies, ASL has been successfully applied to

study the effects of complete174,175 or partial176 occlusion of



the middle cerebral artery. Early human studies have shownthat ASL methods could be successfully applied in acutestroke177–180 and could be used in acetazolamide challenge testin patients with vascular stenosis.156,181

The main problem in cerebral ischemia, as well as inpartial or complete arterial occlusion, is the lengthening intransit time due to collateral flow or reduced arterial veloc-ity,177 sometimes producing artificially elevated signalchanges due to remaining labeled magnetization in the largearteries on the affected site. Such effects need to be clearlydistinguished from luxury perfusion occurring in chronicstroke patients, such as depicted in Figure 5, in which a 50-year-old man presenting with right hemiplegia and aphasiawas scanned more than 1 day after symptom onset. In this casediffusion-weighted and T2-weighted MR images show a hy-perintensity in the infracted area, and the CASL images showincreased cerebral blood flow from luxury perfusion.

TumorsMeasurement of tumor blood flow is important for tu-

mor grading and evaluating anticancer treatment effect.182–184

As example of the use of ASL in tumors, Figure 6 showsclearly increased tumor blood flow in a patient with gliomadespite the lack of contrast enhancement on conventional MRimaging. Because of the use of a purely diffusible tracer, ASLtechniques may allow absolute quantification unaffected bydisrupted blood–brain barrier, a common problem using con-trast reagent-based techniques for assessing tumor perfusion,and allows distinction between high- and low-grade glio-mas.184 However, dynamic contrast medium-based techniquesare still the preferred technique today because of superior SNRand increased anatomic coverage. Furthermore, patients withbrain tumors usually undergo contrast-enhanced scans as partof conventional imaging protocols, and the advent of high fieldimaging will soon allow measurement of perfusion without theneed for an increase in contrast reagent dose.185,186 ASL mayalso be useful for repeated studies on the effects of radiationtherapy or pharmacological agents (such as nicotinamide) to

increase tumor blood flow and also to assess different therapiestreatment regimens in both intracranial and head and neck ma-lignancies.183,187,188

Other Cerebral DiseasesASL has been used in many other diseases affecting the

central nervous system, including Alzheimer disease,189,190

epilepsy,82,191 clinical depression,192 rare diseases, such aspostpartum vasculopathy,193 Rett syndrome,29,194 or Fabri dis-ease,195 as well as traumatic injury in rodent models.153,196 Asummary of the clinical findings in patients is presented below.

In both studies on Alzheimer’s disease, either usingPASL190 or using CASL,189 focal areas of hypoperfusion wereconsistently found in patients as compared with the normalcontrols, mainly in the posterior temporoparietal and occipitalareas190 as well as in frontal and posterior cingulate corti-ces.189 These deficits correlated with the severity of dementia.

Two studies have also been reported on changes in theCBF of patients with temporal lobe epilepsy (TLE), using ei-ther CASL (191) or PASL (82). Both studies could find localor lateralized areas of reduced CBF in the mesial temporal loberelated to TLE. Interestingly, both studies found significantcorrelation between ASL results and PET measurements ofCBF82 or glucose metabolic rate.191

A recent PASL study investigated the links between de-pression, cardiac disease, and CBF.192 For this study, the au-thors acquired two axial slices through the upper and lowerhalves of the lateral ventricles. On the upper slice, CBF wassignificantly lower on the left side in the depressed subjectsthan in the control group, suggesting that relative cerebral hy-poperfusion may underlie depression in elderly cardiac pa-tients.

Rett Syndrome is a rare X-linked genetic disease, lead-ing to developmental regression and deceleration of headgrowth accompanied by loss of communication skills, sei-zures, respiratory abnormalities, and peculiar stereotypedhand-wringing behavior.29 In both studies of this disease, ei-

FIGURE 5. CASL (A), isotropicallydiffusion-weighted (DW) (B), andT2-weighted (C) images of a 50-year-old man with recent onset ofright hemiplegia and aphasia. Allimages were acquired at 3.0 T. TheCASL images were acquired in 6minutes 40 seconds. The DW andT2-weighted MR images show hy-perintensity in the infarcted area,while the perfusion-weighted CASLimage shows increased cerebralblood flow from luxury perfusion(arrow). Note that the DW image isnot acquired at the same angle asthe two other images.



ther with PASL29 or with CASL,194 reduced general perfusionwas observed, with a marked reduction (30%) in the frontallobe, probably related to their delayed developmental state. InFabry disease, Moore et al found that CBF was elevated in theposterior circulation, especially in the thalamus.195

Other Clinical Non-Brain IndicationsAlthough the majority of applications have been focused

on cerebral perfusion, ASL methods have been success-fully applied to other organs, with interest in pulmo-nary,42,43,95–98,197–206 cardiac,11,87–91,207 renal,39,83,99,102,208

breast,209 uterine,210 or ovarian100,101 perfusion studies.Pulmonary ventilation or perfusion MR studies, particu-

larly quantitative methods, would be desirable to improve spa-tial resolution and avoid the radiation dose associated withclassic radionuclide studies used to assess pulmonary embo-lism. ASL techniques were successfully applied in studies ofnormal volunteers95,96,98,203 as well as in animals197,201 andhumans with perfusion deficits,97 lung transplantation,202 andcystic fibrosis.206

Myocardial perfusion was studied in isolated207 and in-tact animal hearts90,91 to assess vasodilation induced by phar-macological agents. Human investigations show promisingpotential of ASL to become an important noninvasive and eas-ily repeatable method to study changes in cardiac blood flow inpatients with coronary disease.211

Several studies of ASL in renal diseases have been per-formed. In a study of kidney transplant rejection, Wang et al99

found that reduction in renal cortical perfusion rate determinedby MRI was significantly correlated with histologic rejection.Other reduced renal perfusion due to renal vascular dis-ease39,83 have also been described.

EMERGING TECHNIQUES

High-Field Clinical ASLIn the past few years, moderate and high field MR sys-

tems ( 3.0T) have been introduced in the clinical settings.Advantages of higher field strength have been demonstrated

for magnetic resonance angiography,212 fMRI,213 and mag-netic resonance spectroscopy.214 An expected twofold SNRincrease proportional to the magnetic field strength is the mostappealing feature of 3T MR imaging, but other properties suchas increased T1 relaxation times may also provide great advan-tages for ASL. Indeed, most CASL sequences use a TR in theorder of 5 seconds, with the shortest possible echo times, re-sulting in proton density weighted images. Therefore, both theincrease in T1 and decrease in T2 expected at 3T215 will notaffect the image contrast, and the native perfusion-weightedimages will plainly profit from a twofold increase inSNR.216,217 Furthermore, the increase in arterial blood T1b

from 1400 ms218 at 1.5 T to 1680 ms219 at 3.0 T shall increasethe contrast-to-noise ratio of the images by approximately 20%to 30%, depending on the imaging parameters. In a study byFranke et al,220 both increase in signal as well as in T1b werefound to be the main contributors for the increase in CNR inASL when going to higher field strength. Therefore, a reduc-tion of the scan time by a factor of up to four for identicalimaging parameters is expected when going from 1.5 T to 3.0T; alternatively, this increase in SNR may be traded for in-creased image resolution.

In PASL sequences, usually an inversion (eg, FAIR) orpresaturation (EPISTAR, TILT, QUIPSS) pulse is followed 1to 2 seconds later by the acquisition sequence. These pulseswill therefore be the main determinant of the contrast in thenative (nonsubtracted) image, which will show in most cases aT1 weighting. For this reason, the expected SNR gain is lessthan that of CASL sequences217 at high field strengths. How-ever, the increased T1b will still contribute to an increase in SNR.

ASL at Multiple Inversion TimesIn the majority of PASL or CASL studies, a single delay

time TI between labeling and image acquisition is used to es-timate CBF. In such cases, the effects of arterial transit time onthe CBF estimation are difficult to evaluate and can potentiallycause errors in calculated perfusion values.31,52,64 This is es-pecially true for patients with stroke and steno-occlusive dis-ease of the carotid arteries for whom the labeled blood flowing

FIGURE 6. CASL (A), postcontrastT1-weighted (B), and T2-weighted(C) images of a 32-year-old womanwith seizures on follow-up for rightfrontal glioma. All images were ac-quired at 3.0 T. The CBF-weightedCASL image was acquired in 3 min-utes, and only 19 averages were in-cluded to form this image.74 Al-though the tumor does not enhanceafter intravenous contrast injection(Gd-DTPA), the CASL perfusion-weighted image demonstrates aclear increase in cerebral perfusion.



via collateral vessels will cause an increased transit time in theaffected area.

A possible approach to solve the transit time problem isbased on the observation that, instead of minimizing the effectof transit time, this physiologically important hemodynamicparameter can also be measured in addition to CBF measure-ments by performing multiple ASL experiments at various TIsbetween ASL labeling and MR acquisition. Buxton et al. in-troduced the theoretical basis of this multiple TI method andproved the principles with an EPISTAR-like technique.63

Gonzalez-At et al proposed a similar method based on aCASL sequence and applied it to fMRI studies.137 Severalstudies have investigated the shape of ASL hemodynamic re-sponse curves, with the amount of ASL signal plotted againstTI.34,103,138 Generally, an increase of the perfusion-weightedsignal is found at short TI for PASL sequences caused by thearrival of the labeled blood through the arterial system (Fig. 1),followed by a decrease at longer TI, caused by a combinationof washout of the tracer and T1 relaxation, seen in both PASLand CASL137 (see Absolute Quantification). The main draw-back of ASL at multiple inversion times is the considerablylonger scan time than for single delay experiments, often ren-dering this technique impractical, especially in difficult patientpopulations. Recently, Inflow Turbo Sampling FAIR (ITS-FAIR) and Turbo TILT were introduced, both of which allowabsolute quantification of CBF and arterial transit time in asingle scan.139,140,221 Both techniques acquire a series of im-ages at increasing delay times after one single inversion pulseby combining a pulsed ASL with a repeated small flip anglegradient echo sampling strategy, in a way similar to the Look-Locker technique used for fast T1 mapping.222 When appliedto PASL sequences, the Look-Locker acquisition train allowsmonitoring the dynamics of blood inflow and tissue perfusionat high temporal resolution (50–200 milliseconds). A clinicalexample of transit time and CBF maps obtained with the turbotilt method is demonstrated in Figure 7.

Selective ASL MethodsThus far, most ASL techniques have been used to mea-

sure tissue perfusion of the brain by nonselectively labeling ofall of the brain feeding arteries. Edelman et al. introduced se-lective labeling with a sagittal inversion slab for angiography;however, no flow territory mapping or CBF quantification wasperformed in these studies.223 As mentioned earlier, local sur-face coils in CASL experiments can also be used for selectivelabeling of the right and left common carotid artery sepa-rately.24–27 Because the surface coils only allow for labeling ofsuperficial arteries, this method is restricted to selective label-ing of either internal carotid arteries (ICA), and no separatelabeling of the posterior circulation can be achieved.

Using a PASL technique similar to that of Edelman et al,223

Eastwood et al selectively labeled either the right or left carotidartery systems by alternating the application of spatially selec-tive inversion pulses in sagittal orientation.224 However, theywere not able to get an absolute quantification using this tech-nique. Recently, another PASL method for selective labelingof ICAs and basilar artery was implemented using two-dimensional labeling pulses forming a pencil beam profile.225

By planning this pencil beam with a gaussian profile parallel tothe slices of interest and below the circle of Willis, 2 to 4 cm ofthe left or right ICA or basilar artery could be labeled. How-ever, a significant contamination of the perfusion territories ofthe nonlabeled arteries was still present and no CBF valueswere obtained, due to the complex labeling scheme used andthe absence of global control for magnetization transfer effects.

Hendrikse et al226 recently developed another techniquebased on TILT,48 called regional perfusion imaging (RPI).With RPI, selective labeling is achieved by using the sharplabeling profiles of the concatenated TILT labeling pulses49

and by interactively planning the spatially selective inversionslabs.226 RPI allows CBF quantification because of its inherentglobal control for MT effects independent of the angulation ofthe labeling slab.48,49 As a potential clinical use of this tech-

FIGURE 7. Absolute CBF (A) and �D= time for the lower edge of the bo-lus to reach the tissue (B) of a pa-tient with right carotid occlusionand small right side infarction (ar-row). It can be seen that the wholeright hemisphere, although not suf-fering from a reduced blood flow,presents with a significant increasein �D, indicating delayed cerebralperfusion through co-lateral circula-tion.



nique, RPI images in a control subject and a patient after highflow extracranial to intracranial bypass surgery are demon-strated in Figure 8.

Velocity-Encoded ASLIn general, most ASL techniques described so far are

based on the proximal labeling of arterial magnetization fol-lowed by its measurement in the tissue of interest after a certaindelay time. As expressed earlier, the main problem of suchtechniques is the finite arterial transit time between the label-ing slab and the volume of interest. This transit time can beconsiderably longer in patients suffering from arterial occlu-sion, stroke, or arteriovenous malformations, for example.Furthermore, the transit time also depends on the actual rela-tive position of the tissue to the labeling plane or slab. That is,the delay will generally be longer for more distal slices.

In an attempt to render ASL techniques independentfrom these arterial transit delays, another class of arterialspin labeling techniques has recently been introduced, inwhich arterial spins are labeled everywhere according to theirvelocity227,228 and termed later velocity-selective ASL byWong et al.229 In this method, a series of nonselective 90°-180°-90° pulse sequence is used (DEFT sequence), with inter-leaved gradients to select spins of a particular velocity, in away similar to MRI sequences used to obtain quantitative flowmeasurements in large vessels.230 For the control experiment,the same sequence is repeated with very low gradients. Themain difference between this method and other ASL tech-niques is that flowing spins are labeled everywhere, includingin the volume of interest, therefore minimizing the transit de-lay time necessary for the blood to reach any region of interestin the organ.

CONCLUSIONTremendous improvements have appeared in the field of

ASL over the last 12 years. Interestingly, however, most up-

to-date methods are still based on the original scheme pro-posed by Williams et al,13 except maybe some techniquesbased on velocity-selective ASL.227,229 This new way of mea-suring blood flow has been used widely in animal imaging,small clinical studies, as well as a potential surrogate to BOLDfMRI in neuroscience. From the original brain technique haveemerged various methods specifically designed to measureblood flow in a large number of organs.

On the technical side, improvements in the techniquesand control for the numerous potential artifacts have permittedthe development of reliable multislice perfusion imaging, witha resolution often as good as, or even better than, most nuclearmedicine approaches. Despite all these improvements and thecontinuous research done in this domain, ASL is still anemerging technique per se and has not replaced more invasiveprocedures for the assessment of blood flow in patients.Among the possible explanations for this are probably the rela-tive complexity of the method and its intrinsic low SNR andrelative high sensitivity to motion artifacts. As summarized inthis review, some of the latest developments and the advent ofhigh field clinical imagers might change this perception, how-ever, and future advances ahead will definitely bring the ro-bustness needed by this technique to be widely used in the clin-ics.

REFERENCES1. Kety SS, Schmidt CF. The determination of cerebral blood flow in man

by use of nitrous oxide in low concentrations. Am J Physiol. 1945;143:53–66.

2. Kety SS, Schmidt CF. The nitrous oxide method for the determination ofcerebral blood flow in man: theory, procedure and normal values. J ClinInvest. 1948;27:467–483.

3. Roberts DA, Rizi R, Lenkinski RE, et al. Magnetic resonance imaging ofthe brain: blood partition coefficient for water: application to spin-tagging measurement of perfusion. J Magn Reson Imaging. 1996;6:363–366.

4. Powers WJ, Grubb RL Jr, Darriet D, et al. Cerebral blood flow and ce-rebral metabolic rate of oxygen requirements for cerebral function andviability in humans. J Cereb Blood Flow Metab. 1985;5:600–608.

FIGURE 8. Regional perfusion imag-ing (RPI) of a normal volunteer (A)and of a patient after EC-IC bypasssurgery (B). In this picture, CBFcoming from the right ICA (for thevolunteer) and from the bypass (forthe patient) are color-coded in red,and CBF coming from the left ICA iscolor-coded in green. Notice thatthe posterior circulation is comingfrom both sides in the volunteer,while both territories are perfectlydistinct in the patient. This probablyreflects a mixing of flow from bothICAs through the circle of Willis(CoW) in the volunteer, while thepatient’s bypass graft is attached ata location distal to the CoW.



5. Raichle ME. Measurement of local cerebral blood flow and metabolismin man with positron emission tomography. Fed Proc. 1981;40:2331–2334.

6. Maeda T, Matsuda H, Hisada K, et al. Three-dimensional regional cere-bral blood perfusion images with single-photon emission computed to-mography. Radiology. 1981;140:817–822.

7. Barbier EL, Lamalle L, Decorps M. Methodology of brain perfusionimaging. J Magn Reson Imaging. 2001;13:496–520.

8. Calamante F, Thomas DL, Pell GS, et al. Measuring cerebral blood flowusing magnetic resonance imaging techniques. J Cereb Blood FlowMetab. 1999;19:701–735.

9. Alsop DC. Perfusion magnetic resonance imaging. In: Atlas SW, ed.Magnetic Resonance Imaging of the Brain and Spine, 3rd ed, vol. 1.Philadelphia: Lippincott Williams & Wilkins, 2002:215–238.

10. Kwong KK, Chesler DA, Weisskoff RM, et al. MR perfusion studieswith T1-weighted echo planar imaging. Magn Reson Med. 1995;34:878–887.

11. Detre JA, Zhang W, Roberts DA, et al. Tissue specific perfusion imagingusing arterial spin labeling. NMR Biomed. 1994;7:75–82.

12. Silva AC, Williams DS, Koretsky AP. Evidence for the exchange ofarterial spin-labeled water with tissue water in rat brain from diffusion-sensitized measurements of perfusion. Magn Reson Med. 1997;38:232–237.

13. Williams DS, Detre JA, Leigh JS, et al. Magnetic resonance imaging ofperfusion using spin inversion of arterial water. Proc Natl Acad Sci USA.1992;89:212–216.

14. Dixon WT, Du LN, Faul DD, et al. Projection angiograms of blood la-beled by adiabatic fast passage. Magn Reson Med. 1986;3:454–462.

15. Zhang W, Williams DS, Koretsky AP. Measurement of rat brain perfu-sion by NMR using spin labeling of arterial water: in vivo determinationof the degree of spin labeling. Magn Reson Med. 1993;29:416–421.

16. Maccotta L, Detre JA, Alsop DC. The efficiency of adiabatic inversionfor perfusion imaging by arterial spin labeling. NMR Biomed. 1997;10:216–221.

17. Utting JF, Thomas DL, Gadian DG, et al. Velocity-driven adiabatic fastpassage for arterial spin labeling: results from a computer model. MagnReson Med. 2003;49:398–401.

18. Gach HM, Kam AW, Reid ED, et al. Quantitative analysis of adiabaticfast passage for steady laminar and turbulent flows. Magn Reson Med.2002;47:709–719.

19. Edelman RR, Siewert B, Darby DG, et al. Qualitative mapping of cere-bral blood flow and functional localization with echo-planar MR imag-ing and signal targeting with alternating radio frequency. Radiology.1994;192:513–520.

20. Kim SG. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application tofunctional mapping. Magn Reson Med. 1995;34:293–301.

21. Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tis-sue water proton relaxation in vivo. Magn Reson Med. 1989;10:135–144.

22. Henkelman RM, Huang X, Xiang QS, et al. Quantitative interpretationof magnetization transfer. Magn Reson Med. 1993;29:759–766.

23. Silva AC, Zhang W, Williams DS, et al. Multi-slice MRI of rat brainperfusion during amphetamine stimulation using arterial spin labeling.Magn Reson Med. 1995;33:209–214.

24. Zhang W, Silva AC, Williams DS, et al. NMR measurement of perfusionusing arterial spin labeling without saturation of macromolecular spins.Magn Reson Med. 1995;33:370–376.

25. Trampel R, Mildner T, Goerke U, et al. Continuous arterial spin labelingusing a local magnetic field gradient coil. Magn Reson Med. 2002;48:543–546.

26. Zaharchuk G, Ledden PJ, Kwong KK, et al. Multislice perfusion andperfusion territory imaging in humans with separate label and imagecoils. Magn Reson Med. 1999;41:1093–1098.

27. Mildner T, Trampel R, Moller HE, et al. Functional perfusion imagingusing continuous arterial spin labeling with separate labeling and imag-ing coils at 3 T. Magn Reson Med. 2003;49:791–795.

28. Alsop DC, Detre JA. Multisection cerebral blood flow MR imaging withcontinuous arterial spin labeling. Radiology. 1998;208:410–416.

29. Naidu S, Kaufmann WE, Abrams MT, et al. Neuroimaging studies inRett syndrome. Brain Dev. 2001;23(suppl 1):62–71.

30. Schwarzbauer C, Morrissey SP, Haase A. Quantitative magnetic reso-nance imaging of perfusion using magnetic labeling of water protonspins within the detection slice. Magn Reson Med. 1996;35:540–546.

31. Calamante F, Williams SR, van Bruggen N, et al. A model for quantifi-cation of perfusion in pulsed labelling techniques. NMR Biomed. 1996;9:79–83.

32. Kim SG, Tsekos NV. Perfusion imaging by a flow-sensitive alternatinginversion recovery (FAIR) technique: application to functional brain im-aging. Magn Reson Med. 1997;37:425–435.

33. Kao YH, Wan X, MacFall JR. Simultaneous multislice acquisition witharterial-flow tagging (SMART) using echo planar imaging (EPI). MagnReson Med. 1998;39:662–665.

34. Yang Y, Frank JA, Hou L, et al. Multislice imaging of quantitative ce-rebral perfusion with pulsed arterial spin labeling. Magn Reson Med.1998;39:825–832.

35. Wong EC, Buxton RB, Frank LR. Implementation of quantitative perfu-sion imaging techniques for functional brain mapping using pulsed arte-rial spin labeling. NMR Biomed. 1997;10:237–249.

36. Yongbi MN, Branch CA, Helpern JA. Perfusion imaging using FOCI RFpulses. Magn Reson Med. 1998;40:938–943.

37. Tanabe JL, Yongbi M, Branch C, et al. MR perfusion imaging in humanbrain using the UNFAIR technique: un-inverted flow-sensitive alternat-ing inversion recovery. J Magn Reson Imaging. 1999;9:761–767.

38. Berr SS, Mai VM. Extraslice spin tagging (EST) magnetic resonanceimaging for the determination of perfusion. J Magn Reson Imaging.1999;9:146–150.

39. Berr SS, Hagspiel KD, Mai VM, et al. Perfusion of the kidney usingextraslice spin tagging (EST) magnetic resonance imaging. J Magn Re-son Imaging. 1999;10:886–891.

40. Zhou J, Mori S, van Zijl PC. FAIR excluding radiation damping(FAIRER). Magn Reson Med. 1998;40:712–719.

41. Zhou J, van Zijl PC. Perfusion imaging using FAIR with a short prede-lay. Magn Reson Med. 1999;41:1099–1107.

42. Mai VM, Berr SS. MR perfusion imaging of pulmonary parenchymausing pulsed arterial spin labeling techniques: FAIRER and FAIR. JMagn Reson Imaging. 1999;9:483–487.

43. Mai VM, Hagspiel KD, Christopher JM, et al. Perfusion imaging of thehuman lung using flow-sensitive alternating inversion recovery with anextra radiofrequency pulse (FAIRER). Magn Reson Imaging. 1999;17:355–361.

44. Lai S, Wang J, Jahng GH. FAIR exempting separate T (1) measurement(FAIREST): a novel technique for online quantitative perfusion imagingand multi-contrast fMRI. NMR Biomed. 2001;14:507–516.

45. Pell GS, Thomas DL, Lythgoe MF, et al. Implementation of quantitativeFAIR perfusion imaging with a short repetition time in time-course stud-ies. Magn Reson Med. 1999;41:829–840.

46. Schwarzbauer C, Heinke W. BASE imaging: a new spin labeling tech-nique for measuring absolute perfusion changes. Magn Reson Med.1998;39:717–722.

47. Edelman RR, Chen Q. EPISTAR MRI: multislice mapping of cerebralblood flow. Magn Reson Med. 1998;40:800–805.

48. Golay X, Stuber M, Pruessmann KP, et al. Transfer insensitive labelingtechnique (TILT): application to multislice functional perfusion imag-ing. J Magn Reson Imaging. 1999;9:454–461.

49. Pruessmann KP, Golay X, Stuber M, et al. RF pulse concatenation forspatially selective inversion. J Magn Reson. 2000;146:58–65.

50. Jahng GH, Zhu XP, Matson GB, et al. Improved perfusion-weightedMRI by a novel double inversion with proximal labeling of both taggedand control acquisitions. Magn Reson Med. 2003;49:307–314.

51. Wong EC, Buxton RB, Frank LR. Quantitative imaging of perfusionusing a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med.1998;39:702–708.

52. Wong EC, Buxton RB, Frank LR. A theoretical and experimental com-parison of continuous and pulsed arterial spin labeling techniques forquantitative perfusion imaging. Magn Reson Med. 1998;40:348–355.

53. Luh WM, Wong EC, Bandettini PA, et al. QUIPSS II with thin-slice TI1periodic saturation: a method for improving accuracy of quantitative per-



fusion imaging using pulsed arterial spin labeling. Magn Reson Med.1999;41:1246–1254.

54. Luh WM, Wong EC, Bandettini PA, et al. Comparison of simultaneouslymeasured perfusion and BOLD signal increases during brain activationwith T(1)-based tissue identification. Magn Reson Med. 2000;44:137–143.

55. Wong EC, Luh WM, Liu TT. Turbo ASL: arterial spin labeling withhigher SNR and temporal resolution. Magn Reson Med. 2000;44:511–515.

56. Silva AC, Kim SG. Pseudo-continuous arterial spin labeling techniquefor measuring CBF dynamics with high temporal resolution. Magn Re-son Med. 1999;42:425–429.

57. Barbier EL, Silva AC, Kim HJ, et al. Perfusion analysis using dynamicarterial spin labeling (DASL). Magn Reson Med. 1999;41:299–308.

58. Barbier EL, Silva AC, Kim SG, et al. Perfusion imaging using dynamicarterial spin labeling (DASL). Magn Reson Med. 2001;45:1021–1029.

59. Mani S, Pauly J, Conolly S, et al. Background suppression with multipleinversion recovery nulling: applications to projective angiography.Magn Reson Med. 1997;37:898–905.

60. Ye FQ, Frank JA, Weinberger DR, et al. Noise reduction in 3D perfusionimaging by attenuating the static signal in arterial spin tagging(ASSIST). Magn Reson Med. 2000;44:92–100.

61. Blamire AM, Styles P. Spin echo entrapped perfusion image (SEEPAGE):a nonsubtraction method for direct imaging of perfusion. Magn ResonMed. 2000;43:701–704.

62. Duyn JH, Tan CX, van Gelderen P, et al. High-sensitivity single-shotperfusion-weighted fMRI. Magn Reson Med. 2001;46:88–94.

63. Buxton RB, Frank LR, Wong EC, et al. A general kinetic model forquantitative perfusion imaging with arterial spin labeling. Magn ResonMed. 1998;40:383–396.

64. Alsop DC, Detre JA. Reduced transit-time sensitivity in noninvasivemagnetic resonance imaging of human cerebral blood flow. J CerebBlood Flow Metab. 1996;16:1236–1249.

65. Chesler DA, Kwong KK. An intuitive guide to the T1 based perfusionmodel. Int J Imaging Syst Technol. 1995;6:171–174.</jrn>

66. Pekar J, Jezzard P, Roberts DA, et al. Perfusion imaging with compen-sation for asymmetric magnetization transfer effects. Magn Reson Med.1996;35:70–79.

67. McLaughlin AC, Ye FQ, Pekar JJ, et al. Effect of magnetization transferon the measurement of cerebral blood flow using steady-state arterialspin tagging approaches: a theoretical investigation. Magn Reson Med.1997;37:501–510.

68. Silva AC, Zhang W, Williams DS, et al. Estimation of water extractionfractions in rat brain using magnetic resonance measurement of perfu-sion with arterial spin labeling. Magn Reson Med. 1997;37:58–68.

69. Parkes LM, Tofts PS. Improved accuracy of human cerebral blood per-fusion measurements using arterial spin labeling: accounting for capil-lary water permeability. Magn Reson Med. 2002;48:27–41.

70. Zhou J, Wilson DA, Ulatowski JA, et al. Two-compartment exchangemodel for perfusion quantification using arterial spin tagging. J CerebBlood Flow Metab. 2001;21:440–455.

71. St. Lawrence KS, Frank JA, McLaughlin AC. Effect of restricted waterexchange on cerebral blood flow values calculated with arterial spin tag-ging: a theoretical investigation. Magn Reson Med. 2000;44:440–449.

72. Ewing JR, Cao Y, Fenstermacher J. Single-coil arterial spin-tagging forestimating cerebral blood flow as viewed from the capillary: relativecontributions of intra- and extravascular signal. Magn Reson Med. 2001;46:465–475.

73. Yongbi MN, Tan CX, Frank JA, et al. A protocol for assessing subtrac-tion errors of arterial spin-tagging perfusion techniques in human brain.Magn Reson Med. 2000;43:896–900.

74. Oguz KK, Golay X, Pizzini FB, et al. Sickle cell disease: continuousarterial spin-labeling perfusion MR imaging in children. Radiology.2003;227:567–574.

75. Lei H, Peeling J. A strategy to optimize the signal-to-noise ratio in one-coil arterial spin tagging perfusion imaging. Magn Reson Med. 1999;41:563–568.

76. Li TQ, Moseley ME, Glover G. A FAIR study of motor cortex activationunder normo- and hypercapnia induced by breath challenge. Neuroim-age. 1999;10:562–569.

77. Yang Y, Gu H, Zhan W, et al. Simultaneous perfusion and BOLD im-aging using reverse spiral scanning at 3T: characterization of functionalcontrast and susceptibility artifacts. Magn Reson Med. 2002;48:278–289.

78. Zhang W. A quantitative analysis of alternated line scanning in k spaceand its application in MRI of regional tissue perfusion by arterial spinlabeling. J Magn Reson B. 1995;107:165–171.

79. Branch CA, Hernandez L, Yongbi M, et al. Rapid and continuous moni-toring of cerebral perfusion by magnetic resonance line scan assessmentwith arterial spin tagging. NMR Biomed. 1999;12:15–25.

80. Chen Q, Siewert B, Bly BM, et al. STAR-HASTE: perfusion imagingwithout magnetic susceptibility artifact. Magn Reson Med. 1997;38:404–408.

81. Crelier GR, Hoge RD, Munger P, et al. Perfusion-based functional mag-netic resonance imaging with single-shot RARE and GRASE acquisi-tions. Magn Reson Med. 1999;41:132–136.

82. Liu HL, Kochunov P, Hou J, et al. Perfusion-weighted imaging of in-terictal hypoperfusion in temporal lobe epilepsy using FAIR-HASTE:comparison with H(2)(15)O PET measurements. Magn Reson Med.2001;45:431–435.

83. Prasad PV, Kim D, Kaiser AM, et al. Noninvasive comprehensive char-acterization of renal artery stenosis by combination of STAR angiogra-phy and EPISTAR perfusion imaging. Magn Reson Med. 1997;38:776–787.

84. Frank LR, Wong EC, Haseler LJ, et al. Dynamic imaging of perfusion inhuman skeletal muscle during exercise with arterial spin labeling. MagnReson Med. 1999;42:258–267.

85. Raynaud JS, Duteil S, Vaughan JT, et al. Determination of skeletalmuscle perfusion using arterial spin labeling NMRI: validation by com-parison with venous occlusion plethysmography. Magn Reson Med.2001;46:305–311.

86. Richardson RS, Haseler LJ, Nygren AT, et al. Local perfusion and meta-bolic demand during exercise: a noninvasive MRI method of assess-ment. J Appl Physiol. 2001;91:1845–1853.

87. Reeder SB, Atalay MK, McVeigh ER, et al. Quantitative cardiac perfu-sion: a noninvasive spin-labeling method that exploits coronary vesselgeometry. Radiology. 1996;200:177–184.

88. Belle V, Kahler E, Waller C, et al. In vivo quantitative mapping of car-diac perfusion in rats using a noninvasive MR spin-labeling method. JMagn Reson Imaging. 1998;8:1240–1245.

89. Reeder SB, Holmes AA, McVeigh ER, et al. Simultaneous noninvasivedetermination of regional myocardial perfusion and oxygen content inrabbits: toward direct measurement of myocardial oxygen consumptionat MR imaging. Radiology. 1999;212:739–747.

90. Waller C, Kahler E, Hiller KH, et al. Myocardial perfusion and intracap-illary blood volume in rats at rest and with coronary dilatation: MR im-aging in vivo with use of a spin-labeling technique. Radiology. 2000;215:189–197.

91. Waller C, Hiller KH, Voll S, et al. Myocardial perfusion imaging usinga non-contrast agent MR imaging technique. Int J Cardiovasc Imaging.2001;17:123–132.

92. Streif JU, Hiller KH, Waller C, et al. In vivo assessment of absoluteperfusion in the murine skeletal muscle with spin labeling MRI: mag-netic resonance imaging. J Magn Reson Imaging. 2003;17:147–152.

93. Marro KI, Kushmerick MJ. Skeletal muscle perfusion measurements us-ing adiabatic inversion of arterial water. Magn Reson Med. 1997;38:40–47.

94. Toussaint JF, Kwong KK. M’Kparu F, et al. Interrelationship of oxida-tive metabolism and local perfusion demonstrated by NMR in humanskeletal muscle. J Appl Physiol. 1996;81:2221–2228.

95. Roberts DA, Gefter WB, Hirsch JA, et al. Pulmonary perfusion: respi-ratory-triggered three-dimensional MR imaging with arterial spin tag-ging—preliminary results in healthy volunteers. Radiology. 1999;212:890–895.

96. Hatabu H, Tadamura E, Prasad PV, et al. Noninvasive pulmonary per-fusion imaging by STAR-HASTE sequence. Magn Reson Med. 2000;44:808–812.

97. Mai VM, Chen Q, Bankier AA, et al. Effect of lung inflation on arterialspin labeling signal in MR perfusion imaging of human lung. J MagnReson Imaging. 2001;13:954–959.



98. Mai VM, Bankier AA, Prasad PV, et al. MR ventilation-perfusion im-aging of human lung using oxygen-enhanced and arterial spin labelingtechniques. J Magn Reson Imaging. 2001;14:574–579.

99. Wang JJ, Hendrich KS, Jackson EK, et al. Perfusion quantitation intransplanted rat kidney by MRI with arterial spin labeling. Kidney Int.1998;53:1783–1791.

100. Tempel C, Neeman M. Spatial and temporal modulation of perfusion inthe rat ovary measured by arterial spin labeling MRI. J Magn ResonImaging. 1999;9:794–803.

101. Tempel C, Neeman M. Perfusion of the rat ovary: application of pulsedarterial spin labeling MRI. Magn Reson Med. 1999;41:113–123.

102. Karger N, Biederer J, Lusse S, et al. Quantitation of renal perfusion usingarterial spin labeling with FAIR-UFLARE. Magn Reson Imaging. 2000;18:641–647.

103. Ye FQ, Mattay VS, Jezzard P, et al. Correction for vascular artifacts incerebral blood flow values measured by using arterial spin tagging tech-niques. Magn Reson Med. 1997;37:226–235.

104. Pell GS, Lewis DP, Branch CA. Pulsed arterial spin labeling using Tur-boFLASH with suppression of intravascular signal. Magn Reson Med.2003;49:341–350.

105. Wang J, Alsop DC, Song HK, et al. Arterial transit time imaging withflow encoding arterial spin tagging (FEAST). Magn Reson Med. 2003;50:599–607.

106. Frank LR, Wong EC, Buxton RB. Slice profile effects in adiabatic in-version: application to multislice perfusion imaging. Magn Reson Med.1997;38:558–564.

107. Schepers J, Garwood M, van der Sanden B, et al. Improved subtractionby adiabatic FAIR perfusion imaging. Magn Reson Med. 2002;47:330–336.

108. Zhan W, Gu H, Silbersweig DA, et al. Inversion profiles of adiabaticinversion pulses for flowing spins: the effects on labeling efficiency andlabeling accuracy in perfusion imaging with pulsed arterial spin-labeling. Magn Reson Imaging. 2002;20:487–494.

109. Keilholz-George SD, Knight-Scott J, Berr SS. Theoretical analysis ofthe effect of imperfect slice profiles on tagging schemes for pulsed arte-rial spin labeling MRI. Magn Reson Med. 2001;46:141–148.

110. Yongbi MN, Yang Y, Frank JA, et al. Multislice perfusion imaging inhuman brain using the C-FOCI inversion pulse: comparison with hyper-bolic secant. Magn Reson Med. 1999;42:1098–1105.

111. Shinnar M, Leigh JS. The application of spinors to pulse synthesis andanalysis. Magn Reson Med. 1989;12:93–98.

112. Kwong KK, Belliveau JW, Chesler DA, et al. Dynamic magnetic reso-nance imaging of human brain activity during primary sensory stimula-tion. Proc Natl Acad Sci USA. 1992;89:5675–5679.

113. Bandettini PA, Wong EC, Hinks RS, et al. Time course EPI of humanbrain function during task activation. Magn Reson Med. 1992;25:390–397.

114. Ogawa S, Lee TM, Kay AR, et al. Brain magnetic resonance imagingwith contrast dependent on blood oxygenation. Proc Natl Acad Sci USA.1990;87:9868–9872.

115. Kim SG, Tsekos NV, Ashe J. Multi-slice perfusion-based functionalMRI using the FAIR technique: comparison of CBF and BOLD effects.NMR Biomed. 1997;10:191–196.

116. Buxton RB, Wong EC, Frank LR. Dynamics of blood flow and oxygen-ation changes during brain activation: the balloon model. Magn ResonMed. 1998;39:855–864.

117. Ye FQ, Smith AM, Yang Y, et al. Quantitation of regional cerebral bloodflow increases during motor activation: a steady-state arterial spin tag-ging study. Neuroimage. 1997;6:104–112.

118. Ye FQ, Yang Y, Duyn J, et al. Quantitation of regional cerebral bloodflow increases during motor activation: a multislice, steady-state, arterialspin tagging study. Magn Reson Med. 1999;42:404–407.

119. Schulte AC, Speck O, Oesterle C, et al. Separation and quantification ofperfusion and BOLD effects by simultaneous acquisition of functionalI(0)- and T2(*)-parameter maps. Magn Reson Med. 2001;45:811–816.

120. Zhu XH, Kim SG, Andersen P, et al. Simultaneous oxygenation andperfusion imaging study of functional activity in primary visual cortex atdifferent visual stimulation frequency: quantitative correlation betweenBOLD and CBF changes. Magn Reson Med. 1998;40:703–711.

121. Talagala SL, Noll DC. Functional MRI using steady-state arterial waterlabeling. Magn Reson Med. 1998;39:179–183.

122. Darby DG, Nobre AC, Thangaraj V, et al. Cortical activation in the hu-man brain during lateral saccades using EPISTAR functional magneticresonance imaging. Neuroimage. 1996;3:53–62.

123. Ye FQ, Smith AM, Mattay VS, et al. Quantitation of regional cerebralblood flow increases in prefrontal cortex during a working memory task:a steady-state arterial spin-tagging study. Neuroimage. 1998;8:44–49.

124. Aguirre GK, Detre JA, Zarahn E, et al. Experimental design and therelative sensitivity of BOLD and perfusion fMRI. Neuroimage. 2002;15:488–500.

125. Detre JA, Wang J. Technical aspects and utility of fMRI using BOLDand ASL. Clin Neurophysiol. 2002;113:621–634.

126. Wang J, Aguirre GK, Kimberg DY, et al. Arterial spin labeling perfusionfMRI with very low task frequency. Magn Reson Med. 2003;49:796–802.

127. Kim SG, Ugurbil K. Comparison of blood oxygenation and cerebralblood flow effects in fMRI: estimation of relative oxygen consumptionchange. Magn Reson Med. 1997;38:59–65.

128. Preibisch C, Haase A. Perfusion imaging using spin-labeling methods:contrast-to-noise comparison in functional MRI applications. Magn Re-son Med. 2001;46:172–182.

129. Tsekos NV, Zhang F, Merkle H, et al. Quantitative measurements ofcerebral blood flow in rats using the FAIR technique: correlation withprevious iodoantipyrine autoradiographic studies. Magn Reson Med.1998;39:564–573.

130. Duong TQ, Kim SG. In vivo MR measurements of regional arterial andvenous blood volume fractions in intact rat brain. Magn Reson Med.2000;43:393–402.

131. Duong TQ, Kim DS, Ugurbil K, et al. Localized cerebral blood flowresponse at submillimeter columnar resolution. Proc Natl Acad Sci USA.2001;98:10904–10909.

132. Lee SP, Silva AC, Kim SG. Comparison of diffusion-weighted high-resolution CBF and spin-echo BOLD fMRI at 9.4 T. Magn Reson Med.2002;47:736–741.

133. Liu HL, Pu Y, Nickerson LD, et al. Comparison of the temporal responsein perfusion and BOLD-based event-related functional MRI. Magn Re-son Med. 2000;43:768–772.

134. Liu HL, Gao JH. Perfusion-based event-related functional MRI. MagnReson Med. 1999;42:1011–1013.

135. Yang Y, Engelien W, Pan H, et al. A CBF-based event-related brainactivation paradigm: characterization of impulse-response function andcomparison to BOLD. Neuroimage. 2000;12:287–297.

136. Zhou J, van Zijl PC. Effect of transit times on quantification of cerebralblood flow by the FAIR T(1)-difference approach. Magn Reson Med.1999;42:890–894.

137. Gonzalez-At JB, Alsop DC, Detre JA. Cerebral perfusion and arterialtransit time changes during task activation determined with continuousarterial spin labeling. Magn Reson Med. 2000;43:739–746.

138. Yang Y, Engelien W, Xu S, et al. Transit time, trailing time, and cerebralblood flow during brain activation: measurement using multislice,pulsed spin-labeling perfusion imaging. Magn Reson Med. 2000;44:680–685.

139. Gunther M, Bock M, Schad LR. Arterial spin labeling in combinationwith a look-locker sampling strategy: inflow turbo-sampling EPI-FAIR(ITS-FAIR). Magn Reson Med. 2001;46:974–984.

140. Hendrikse J, Lu H, van der Grond J, et al. Measurements of cerebralperfusion and arterial hemodynamics during visual stimulation usingTURBO-TILT. Magn Reson Med. 2003;50:429–433.

141. Wang J, Aguirre GK, Kimberg DY, et al. Empirical analyses of null-hypothesis perfusion FMRI data at 1.5 and 4 T. Neuroimage. 2003;19:1449–1462.

142. Davis TL, Kwong KK, Weisskoff RM, et al. Calibrated functional MRI:mapping the dynamics of oxidative metabolism. Proc Natl Acad SciUSA. 1998;95:1834–1839.

143. Kim SG, Rostrup E, Larsson HB, et al. Determination of relativeCMRO2 from CBF and BOLD changes: significant increase of oxygenconsumption rate during visual stimulation. Magn Reson Med. 1999;41:1152–1161.

144. Hoge RD, Atkinson J, Gill B, et al. Linear coupling between cerebral



blood flow and oxygen consumption in activated human cortex. ProcNatl Acad Sci USA. 1999;96:9403–9408.

145. Hoge RD, Atkinson J, Gill B, et al. Stimulus-dependent BOLD and per-fusion dynamics in human V1. Neuroimage. 1999;9:573–585.

146. Duong TQ, Iadecola C, Kim SG. Effect of hyperoxia, hypercapnia, andhypoxia on cerebral interstitial oxygen tension and cerebral blood flow.Magn Reson Med. 2001;45:61–70.

147. Silva AC, Lee SP, Yang G, et al. Simultaneous blood oxygenation level-dependent and cerebral blood flow functional magnetic resonance imag-ing during forepaw stimulation in the rat. J Cereb Blood Flow Metab.1999;19:871–879.

148. Silva AC, Lee SP, Iadecola C, et al. Early temporal characteristics ofcerebral blood flow and deoxyhemoglobin changes during somatosen-sory stimulation. J Cereb Blood Flow Metab. 2000;20:201–206.

149. Ernst T, Hennig J. Observation of a fast response in functional MR.Magn Reson Med. 1994;32:146–149.

150. Kim SG, Duong TQ. Mapping cortical columnar structures using fMRI.Physiol Behav. 2002;77:641–644.

151. Miller KL, Luh WM, Liu TT, et al. Nonlinear temporal dynamics of thecerebral blood flow response. Hum Brain Mapp. 2001;13:1–12.

152. Salmeron BJ, Stein EA. Pharmacological applications of magnetic reso-nance imaging. Psychopharmacol Bull. 2002;36:102–129.

153. Robertson CL, Hendrich KS, Kochanek PM, et al. Assessment of 2-chlo-roadenosine treatment after experimental traumatic brain injury in the ratusing arterial spin-labeled MRI: a preliminary report. Acta NeurochirSuppl. 2000;76:187–189.

154. Kochanek PM, Hendrich KS, Robertson CL, et al. Assessment of theeffect of 2-chloroadenosine in normal rat brain using spin-labeled MRImeasurement of perfusion. Magn Reson Med. 2001;45:924–929.

155. Brown GG, Eyler Zorrilla LT, Georgy B, et al. BOLD and perfusionresponse to finger-thumb apposition after acetazolamide administration:differential relationship to global perfusion. J Cereb Blood Flow Metab.2003;23:829–837.

156. Detre JA, Samuels OB, Alsop DC, et al. Noninvasive magnetic reso-nance imaging evaluation of cerebral blood flow with acetazolamidechallenge in patients with cerebrovascular stenosis. J Magn Reson Im-aging. 1999;10:870–875.

157. St. Lawrence KS, Ye FQ, Lewis BK, et al. Effects of indomethacin oncerebral blood flow at rest and during hypercapnia: an arterial spin tag-ging study in humans. J Magn Reson Imaging. 2002;15:628–635.

158. St. Lawrence KS, Ye FQ, Lewis BK, et al. Measuring the effects ofindomethacin on changes in cerebral oxidative metabolism and cerebralblood flow during sensorimotor activation. Magn Reson Med. 2003;50:99–106.

159. Sicard K, Shen Q, Brevard ME, et al. Regional cerebral blood flow andBOLD responses in conscious and anesthetized rats under basal and hy-percapnic conditions: implications for functional MRI studies. J CerebBlood Flow Metab. 2003;23:472–481.

160. Born AP, Rostrup E, Miranda MJ, et al. Visual cortex reactivity in se-dated children examined with perfusion MRI (FAIR). Magn Reson Im-aging. 2002;20:199–205.

161. Ewing JR, Wei L, Knight RA, et al. Direct comparison of local cerebralblood flow rates measured by MRI arterial spin-tagging and quantitativeautoradiography in a rat model of experimental cerebral ischemia. JCereb Blood Flow Metab. 2003;23:198–209.

162. Pell GS, King MD, Proctor E, et al. Comparative study of the FAIRtechnique of perfusion quantification with the hydrogen clearancemethod. J Cereb Blood Flow Metab. 2003;23:689–699.

163. Floyd TF, Ratcliffe SJ, Wang J, et al. Precision of the CASL-perfusionMRI technique for the measurement of cerebral blood flow in wholebrain and vascular territories. J Magn Reson Imaging. 2003;18:649–655.

164. Wang J, Licht DJ, Jahng GH, et al. Pediatric perfusion imaging usingpulsed arterial spin labeling. J Magn Reson Imaging. 2003;18:404–413.

165. Lia TQ, Guang Chen Z, Ostergaard L, et al. Quantification of cerebralblood flow by bolus tracking and artery spin tagging methods. MagnReson Imaging. 2000;18:503–512.

166. Lia TQ, Haefelin TN, Chan B, et al. Assessment of hemodynamic re-sponse during focal neural activity in human using bolus tracking, arte-rial spin labeling and BOLD techniques. Neuroimage. 2000;12:442–451.

167. Ye FQ, Berman KF, Ellmore T, et al. H(2)(15)O PET validation ofsteady-state arterial spin tagging cerebral blood flow measurements inhumans. Magn Reson Med. 2000;44:450–456.

168. Liachenko S, Tang P, Hamilton RL, et al. Regional dependence of cere-bral reperfusion after circulatory arrest in rats. J Cereb Blood FlowMetab. 2001;21:1320–1329.

169. Liachenko S, Tang P, Xu Y. Deferoxamine improves early postresusci-tation reperfusion after prolonged cardiac arrest in rats. J Cereb BloodFlow Metab. 2003;23:574–581.

170. Xu Y, Liachenko S, Tang P. Dependence of early cerebral reperfusionand long-term outcome on resuscitation efficiency after cardiac arrest inrats. Stroke. 2002;33:837–843.

171. Krep H, Bottiger BW, Bock C, et al. Time course of circulatory andmetabolic recovery of cat brain after cardiac arrest assessed by perfu-sion- and diffusion-weighted imaging and MR-spectroscopy. Resuscita-tion. 2003;58:337–348.

172. Schoning M, Niemann G, Hartig B. Transcranial color duplex sonogra-phy of basal cerebral arteries: reference data of flow velocities fromchildhood to adulthood. Neuropediatrics. 1996;27:249–255.

173. Schoning M, Hartig B. Age dependence of total cerebral blood flow vol-ume from childhood to adulthood. J Cereb Blood Flow Metab. 1996;16:827–833.

174. Calamante F, Lythgoe MF, Pell GS, et al. Early changes in water diffu-sion, perfusion, T1, and T2 during focal cerebral ischemia in the rat stud-ied at 8.5 T. Magn Reson Med. 1999;41:479–485.

175. Zaharchuk G, Mandeville JB, Bogdanov AA Jr, et al. Cerebrovasculardynamics of autoregulation and hypoperfusion: an MRI study of CBFand changes in total and microvascular cerebral blood volume duringhemorrhagic hypotension. Stroke. 1999;30:2197–2204; discussion2204–2195.

176. Lythgoe MF, Thomas DL, Calamante F, et al. Acute changes in MRIdiffusion, perfusion, T(1), and T(2) in a rat model of oligemia producedby partial occlusion of the middle cerebral artery. Magn Reson Med.2000;44:706–712.

177. Chalela JA, Alsop DC, Gonzalez-Atavales JB, et al. Magnetic resonanceperfusion imaging in acute ischemic stroke using continuous arterial spinlabeling. Stroke. 2000;31:680–687.

178. Detre JA, Alsop DC. Perfusion magnetic resonance imaging with con-tinuous arterial spin labeling: methods and clinical applications in thecentral nervous system. Eur J Radiol. 1999;30:115–124.

179. Tsuchiya K, Katase S, Hachiya J, et al. Cerebral perfusion MRI witharterial spin labeling technique at 0.5 Tesla. J Comput Assist Tomogr.2000;24:124–127.

180. Siewert B, Schlaug G, Edelman RR, et al. Comparison of EPISTAR andT2*-weighted gadolinium-enhanced perfusion imaging in patients withacute cerebral ischemia. Neurology. 1997;48:673–679.

181. Yen YF, Field AS, Martin EM, et al. Test-retest reproducibility of quan-titative CBF measurements using FAIR perfusion MRI and acetazol-amide challenge. Magn Reson Med. 2002;47:921–928.

182. Gaa J, Warach S, Wen P, et al. Noninvasive perfusion imaging of humanbrain tumors with EPISTAR. Eur Radiol. 1996;6:518–522.

183. Silva AC, Kim SG, Garwood M. Imaging blood flow in brain tumorsusing arterial spin labeling. Magn Reson Med. 2000;44:169–173.

184. Warmuth C, Gunther M, Zimmer C. Quantification of blood flow inbrain tumors: comparison of arterial spin labeling and dynamic suscep-tibility-weighted contrast-enhanced MR imaging. Radiology. 2003;228:523–532.

185. Nobauer-Huhmann IM, Ba-Ssalamah A, Mlynarik V, et al. Magneticresonance imaging contrast enhancement of brain tumors at 3 tesla ver-sus 1.5 tesla. Invest Radiol. 2002;37:114–119.

186. Ba-Ssalamah A, Nobauer-Huhmann IM, Pinker K, et al. Effect of con-trast dose and field strength in the magnetic resonance detection of brainmetastases. Invest Radiol. 2003;38:415–422.

187. Brown SL, Ewing JR, Kolozsvary A, et al. Magnetic resonance imagingof perfusion in rat cerebral 9L tumor after nicotinamide administration.Int J Radiat Oncol Biol Phys. 1999;43:627–633.

188. Schmitt P, Kotas M, Tobermann A, et al. Quantitative tissue perfusionmeasurements in head and neck carcinoma patients before and duringradiation therapy with a non-invasive MR imaging spin-labeling tech-nique. Radiother Oncol. 2003;67:27–34.



189. Alsop DC, Detre JA, Grossman M. Assessment of cerebral blood flow inAlzheimer’s disease by spin-labeled magnetic resonance imaging. AnnNeurol. 2000;47:93–100.

190. Sandson TA, O’Connor M, Sperling RA, et al. Noninvasive perfusionMRI in Alzheimer’s disease: a preliminary report. Neurology. 1996;47:1339–1342.

191. Wolf RL, Alsop DC, Levy-Reis I, et al. Detection of mesial temporallobe hypoperfusion in patients with temporal lobe epilepsy by use ofarterial spin labeled perfusion MR imaging. AJNR Am J Neuroradiol.2001;22:1334–1341.

192. Doraiswamy PM, MacFall J, Krishnan KR, et al. Magnetic resonanceassessment of cerebral perfusion in depressed cardiac patients: prelimi-nary findings. Am J Psychiatry. 1999;156:1641–1643.

193. Chalela JA, Kasner SE, McGarvey M, et al. Continuous arterial spinlabeling perfusion magnetic resonance imaging findings in postpartumvasculopathy. J Neuroimaging. 2001;11:444–446.

194. Golay X, Eichler F, Barker PB, et al. Cerebral blood flow in Rett syn-drome: evaluation with continuous arterial spin labeling. Glasgow. Pro-ceedings of the 9th ISMRM. 2001:1575.

195. Moore DF, Ye F, Schiffmann R, et al. Increased signal intensity in thepulvinar on T1-weighted images: a pathognomonic MR imaging sign ofFabry disease. AJNR Am J Neuroradiol. 2003;24:1096–1101.

196. Forbes ML, Hendrich KS, Kochanek PM, et al. Assessment of cerebralblood flow and CO2 reactivity after controlled cortical impact by perfu-sion magnetic resonance imaging using arterial spin-labeling in rats. JCereb Blood Flow Metab. 1997;17:865–874.

197. Mai VM, Hagspiel KD, Altes T, et al. Detection of regional pulmonaryperfusion deficit of the occluded lung using arterial spin labeling in mag-netic resonance imaging. J Magn Reson Imaging. 2000;11:97–102.

198. Keilholz SD, Knight-Scott J, Christopher JM, et al. Gravity-dependentperfusion of the lung demonstrated with the FAIRER arterial spin tag-ging method. Magn Reson Imaging. 2001;19:929–935.

199. Roberts DA, Rizi RR, Lipson DA, et al. Dynamic observation of pulmo-nary perfusion using continuous arterial spin-labeling in a pig model. JMagn Reson Imaging. 2001;14:175–180.

200. Uematsu H, Levin DL, Hatabu H. Quantification of pulmonary perfu-sion with MR imaging: recent advances. Eur J Radiol. 2001;37:155–163.

201. Keilholz SD, Mai VM, Berr SS, et al. Comparison of first-pass Gd-DOTA and FAIRER MR perfusion imaging in a rabbit model of pulmo-nary embolism. J Magn Reson Imaging. 2002;16:168–171.

202. Lipson DA, Roberts DA, Hansen-Flaschen J, et al. Pulmonary ventila-tion and perfusion scanning using hyperpolarized helium-3 MRI and ar-terial spin tagging in healthy normal subjects and in pulmonary embo-lism and orthotopic lung transplant patients. Magn Reson Med. 2002;47:1073–1076.

203. Mai VM, Liu B, Polzin JA, et al. Ventilation-perfusion ratio of signalintensity in human lung using oxygen-enhanced and arterial spin label-ing techniques. Magn Reson Med. 2002;48:341–350.

204. Rizi RR, Lipson DA, Dimitrov IE, et al. Operating characteristics ofhyperpolarized 3He and arterial spin tagging in MR imaging of ventila-tion and perfusion in healthy subjects. Acad Radiol. 2003;10:502–508.

205. Uematsu H, Ohno Y, Hatabu H. Recent advances in magnetic resonanceperfusion imaging of the lung. Top Magn Reson Imaging. 2003;14:245–251.

206. Wang T, Schultz G, Hebestreit H, et al. Quantitative perfusion mappingof the human lung using 1H spin labeling. J Magn Reson Imaging. 2003;18:260–265.

207. Hiller KH, Waller C, Voll S, et al. Combined high-speed NMR imagingof perfusion and microscopic coronary conductance vessels in the iso-lated rat heart. Microvasc Res. 2001;62:327–334.

208. Roberts DA, Detre JA, Bolinger L, et al. Renal perfusion in humans: MRimaging with spin tagging of arterial water. Radiology. 1995;196:281–286.

209. Zhu DC, Buonocore MH. Breast tissue differentiation using arterial spintagging. Magn Reson Med. 2003;50:966–975.

210. Hagspiel KD, Matsumoto AH, Berr SS. Uterine fibroid embolization:assessment of treatment response using perfusion-weighted extraslicespin tagging (EST) magnetic resonance imaging. J Magn Reson Imag-ing. 2001;13:982–986.

211. Wacker CM, Fidler F, Dueren C, et al. Quantitative assessment of myo-cardial perfusion with a spin-labeling technique: preliminary results inpatients with coronary artery disease. J Magn Reson Imaging. 2003;18:555–560.

212. Bernstein MA, Huston J 3rd, Lin C, et al. High-resolution intracranialand cervical MRA at 3.0T: technical considerations and initial experi-ence. Magn Reson Med. 2001;46:955–962.

213. Gati JS, Menon RS, Ugurbil K, et al. Experimental determination of theBOLD field strength dependence in vessels and tissue. Magn ResonMed. 1997;38:296–302.

214. Barker PB, Hearshen DO, Boska MD. Single-voxel proton MRS of thehuman brain at 1.5T and 3.0T. Magn Reson Med. 2001;45:765–769.

215. Wansapura JP, Holland SK, Dunn RS, et al. NMR relaxation times in thehuman brain at 3.0 tesla. J Magn Reson Imaging. 1999;9:531–538.

216. Yongbi MN, Fera F, Yang Y, et al. Pulsed arterial spin labeling: com-parison of multisection baseline and functional MR imaging perfusionsignal at 1.5 and 3.0 T: initial results in six subjects. Radiology. 2002;222:569–575.

217. Wang J, Alsop DC, Li L, et al. Comparison of quantitative perfusionimaging using arterial spin labeling at 1.5 and 4.0 Tesla. Magn ResonMed. 2002;48:242–254.

218. Lu H, Golay X, van Zijl PC. Intervoxel heterogeneity of event-relatedfunctional magnetic resonance imaging responses as a function of T(1)weighting. Neuroimage. 2002;17:943–955.

219. Lu H, Clingman C, Golay X, et al. What is the longitudinal relaxationtime (T1) of blood at 3.0 Tesla? Toronto. Proceedings of the 11th ISMRM.2003:669.

220. Franke C, van Dorsten FA, Olah L, et al. Arterial spin tagging perfusionimaging of rat brain: dependency on magnetic field strength. Magn Re-son Imaging. 2000;18:1109–1113.

221. Weber MA, Gunther M, Lichy MP, et al. Comparison of arterial spin-labeling techniques and dynamic susceptibility-weighted contrast-enhanced MRI in perfusion imaging of normal brain tissue. Invest Ra-diol. 2003;38:712–718.

222. Look DC, Locker DR. Time saving in measurement of NMR and EPRrelaxation times. Rev Sci Instrum. 1970;41:250–251.

223. Edelman RR, Mattle HP, O’Reilly GV, et al. Magnetic resonance imag-ing of flow dynamics in the circle of Willis. Stroke. 1990;21:56–65.

224. Eastwood JD, Holder CA, Hudgins PA, et al. Magnetic resonance imag-ing with lateralized arterial spin labeling. Magn Reson Imaging. 2002;20:583–586.

225. Davies NP, Jezzard P. Selective arterial spin labeling (SASL): perfusionterritory mapping of selected feeding arteries tagged using two-dimensional radiofrequency pulses. Magn Reson Med. 2003;49:1133–1142.

226. Hendrikse J, van der Grond J, Lu H, et al. Flow territory mapping of thecerebral arteries with regional Perfusion MRI (RPI). Stroke. 2004; (inpress).

227. Duhamel G, de Bazelaire C, Alsop DC. Evaluation of systematic quan-tification errors in velocity-selective arterial spin labeling of the brain.Magn Reson Med. 2003;50:145–153.

228. Norris DG, Schwarzbauer C. Velocity selective radiofrequency pulsetrains. J Magn Reson. 1999;137:231–236.

229. Wong EC, Liu TT, Sidaros K, et al. Velocity selective arterial spin la-beling, Honolulu. Proceedings of the 10th ISMRM. 2002:621.

230. Guilfoyle DN, Gibbs P, Ordidge RJ, et al. Real-time flow measurementsusing echo-planar imaging. Magn Reson Med. 1991;18:1–8.



Perfusion Imaging Using Arterial Spin Labeling

Documents

Transcript of Perfusion Imaging Using Arterial Spin Labeling