Contrast-Enhanced MR Angiography CATEGORY 1Contrast-Enhanced MR Angiography Jeffrey H. Maki, MD, PhD; Michael V. Knopp, MD, PhD; Martin Prince, MD, PhD Dr. Maki is an Associate Professor

182 � SUPPLEMENT TO APPLIED RADIOLOGY© www.appliedradiology.com April 2003

Rapid advances in the implemen-tation and understanding of three-dimensional (3D) contrast-

enhanced magnetic resonance angiogra-phy (CE-MRA) are making minimallyinvasive vascular imaging safer andmore accurate. For this reason, CE-MRA is increasingly utilized as the pri-mary method of vascular imaging. Thisarticle describes the basic principlesunderlying 3D CE-MRA and demon-strates applications throughout the body.Inherent advantages of 3D contrastMRA over “conventional” MRA tech-niques, such as time-of-flight (TOF) andphase contrast (PC), will be discussed.Furthermore, we will explore techniquesfor improving spatial and temporal reso-lution, reducing respiratory motion arti-facts, decreasing contrast dosing, andmore accurately timing the contrastbolus. Finally, new contrast agents willbe reviewed briefly.

CONVENTIONAL MRAMR pulse sequences can exploit bloodmotion to visualize vascular structuresdirectly without intravascular contrast

material. One of the earliest MRA tech-niques, TOF angiography, is performedusing a flow-compensated, gradient-refocused sequence. Data can beacquired as multiple two-dimensional(2D) slices, or as a single 3D volume. Inthis technique, stationary tissues in theslice or volume of interest are “satu-rated” and, therefore, have low signalintensity.1-3 Blood upstream of the imag-ing volume, however, remains “unsatu-rated.” When this blood flows into theimaging volume, it is bright comparedwith the stationary background tissues.This “in-flow” or TOF approach oftenworks particularly well in relativelynormal arteries and veins, including theinferior vena cava (IVC), iliac veins,carotid arteries, and cerebral vascula-ture.4-12 It also has the benefit of allow-ing selective saturation of bloodadjacent to the imaging volume suchthat either arterial or venous signal canbe suppressed. One major disadvantageof TOF angiography is in-plane satura-tion, which can be a problem withslowly flowing blood in tortuous arter-ies or when the long axis of the vesselcoincides with the scan plane. A secondlimitation is turbulence-induced signalloss in and distal to a stenosis.13,14 Inaddition, the relatively long echo timesrequired for gradient moment nullingmake TOF sensitive to susceptibilityartifacts from metallic clips, metalimplants, and bowel gas or other air-tis-sue interfaces. These artifacts have beenimplicated as a primary reason for inac-curacy in TOF MRA. As a further

Contrast-Enhanced MR AngiographyJeffrey H. Maki, MD, PhD; Michael V. Knopp, MD, PhD; Martin Prince, MD, PhD

Dr. Maki is an Associate Professor in theDepartment of Radiology, University ofWashington, Seattle, WA. Dr. Knopp is aProfessor in the Department of Radiology,The Ohio State University Hospitals, Colum-bus, OH. Dr. Prince is a Professor in theDepartment of Radiology, Weill MedicalCollege of Cornell University, and ColumbiaCollege of Physicians and Surgeons, NewYork, NY.

This CME program is sponsored by an edu-

cational grant grom Bracco Diagnostics Inc.

The Institute for Advanced Medical Education

designates this continuing medical education

activity for a maximum of 3 Category 1 credits

toward the AMA Physician’s Recognition

Award. Each physician should claim only those

credits that he/she actually spent in the activity.

This CME article consists of text and related

images appearing in this journal article and

online at www.appliedradiology.com. You

should read the articles and accompanying

images, refer to the references, and complete

the self-evaluation quiz available online

at www.appliedradiology.com. There is no

charge for participating in this program.

Estimated time for completion: Three hours

Date of release: April 2003

Expiration date: April 2005

Program: MH-003

LEARNING OBJECTIVES

After completing this activity, the reader will:

• Be familiar with general parameters/tech-

niques for performing 3D CE-MRA in the

carotid, aortic, pulmonary, pelvic, renal,

mesenteric, and peripheral arteries, as well

as the portal and peripheral veins

• Have a general understanding of k-space and

artifacts, and understand the significance of

sequential vs. centric acquisition order

• Understand the importance of proper bolus

timing as related to 3D CE-MRA, and know

the different ways to achieve that timing

• Be familiar with some of the new contrast

agents currently under investigation, and

understand how they may apply to future

MRA applications

CMECATEGORY 1

www.appliedradiology.com SUPPLEMENT TO APPLIED RADIOLOGY©� 183April 2003

downside, TOF imaging times tend tobe lengthy. This is due to the need toimage perpendicular to the vessel axisin order to avoid in-plane saturation.Long acquisition times can lead tomotion artifacts, including slice misreg-istration, particularly for those patientswho have difficulty remaining still.

Phase-contrast (PC) angiographymakes use of phase shifts as blood flowsin the presence of flow-encoding gradi-ents.1,15,16 Using phase-difference images(created by subtracting two images cre-ated with opposite flow-encoding gradi-ents), residual phase is proportional tovelocity. Stationary background tissue,however, is suppressed because it has nophase shift on either image, and hencesubtracts completely. The flow-encod-ing gradients can be applied in anydirection, or in multiple directions,depending on the desired flow sensitiv-ity. Phase-contrast angiography can beimplemented as a 2D or 3D gradient-refocused sequence, and is substantiallyimproved when performed after contrastadministration.17 It has proven useful forevaluation of the renal arteries, carotidarteries, and the portal vein.8,12,16-20 In nor-mal vessels without stenoses or turbu-lence, the results are often spectacular.

Another strength of PC angiography isits ability to measure flow velocities. Incombination with cine (synchronizationwith cardiac gating), a time-resolvedvelocity profile can be generated similar

to an ultrasound Doppler waveform.21,22

This provides physiologic flow data andallows for the quantitative measurementof flow rates. Unfortunately, it is thissame velocity sensitivity that limits PCangiography. Strong flow-encoding gra-dients make the sequence susceptible todegradation from bulk (cardiac, respira-tory, translational) motion. In addition,PC sequences only quantify a specifiedrange of velocities.19 In occlusive dis-ease, turbulence causes a broad spectrumof rapidly varying velocities, which inturn causes intravoxel phase dispersionand signal loss. Thus, artifactual loss ofvessel visualization at a stenosis is com-mon.13 Interestingly, this apparent defi-ciency can be useful as an adjunct to 3DCE-MRA for characterizing the hemo-dynamic significance of a stenosis, aspressure gradients are known to occur inregions of high turbulence and jet-flow.

CONTRAST-ENHANCED MRA:THEORYThree-dimensional CE-MRA is per-formed in a manner analogous to con-ventional contrast angiography orhelical computed tomography (CT).Rather than relying on blood motion tocreate intravascular signal, a contrastagent (traditionally a gadoliniumchelate) is introduced to shorten the T1(spin lattice) relaxation time of bloodsuch that it is significantly less than the

surrounding tissues. Blood can then beimaged directly (using a T1-weightedsequence), irrespective of flow effects.This alleviates many of the problemsinherent to TOF and PC angiography.In particular, sensitivity to turbulenceis dramatically reduced, and in-planesaturation effects are eliminated.13,14

The technique allows a small numberof slices to be oriented in the plane ofthe target vessels to quickly image anextensive region of vascular anatomy(equal to the field-of-view) at high in-plane resolution. Thus, 3D CE-MRA isintrinsically fast, allowing high-qualitybreath-hold MR angiograms.

Gadolinium Chelates At present, most CE-MRA examina-tions are conducted using gadolinium-based MR contrast agents. Gadolinium(Gd3+) is a paramagnetic metal ion thatdecreases both the spin-lattice (T1) and spin-spin (T2) relaxation times.23

Because Gd3+ itself is biologically toxic,it is chelated with ligands such asgadopentetate dimeglumine (DTPA),gadoteridol (HP-DO3A), or gadodi-amide (DTPA-BMA) to form small-molecular-weight contrast agents.24

These “extracellular” agents diffusefrom the intravascular compartment intothe interstitial space in a matter of min-utes, meaning selective imaging of thevascular structures must be performedrapidly.13,24 As compared with iodinatedcontrast agents, gadolinium chelateshave a very low rate of adverse eventsand no nephrotoxicity, which is a signifi-cant advantage when evaluating patientswith impaired renal function.25-28

Each gadolinium chelate decreasesthe T1 of blood according to the fol-lowing equation14,29:

Equation #11/T1 = 1/1200 + R1[Gd]

in which R1 is the field-dependent T1relaxivity of the gadolinium chelate,[Gd] is the concentration of gadolin-ium chelate, and 1200 (msec) is the T1of blood without gadolinium (at 1.5 T).

Figure 1. Relative signal intensity (SI) for different T1 values versus flip angle (a) for a spoiled gradientsequence. TR = 5 ms, TE <<T2*.


A similar equation exists for T2, whichis always shorter than T1.

Pulse SequenceDynamic CE-MRA exploits the tran-sient shortening in blood T1 followingintravenous (IV) administration of acontrast agent using a fast 3D spoiledgradient-echo sequence.14,30,31 We usethe term transient because this tech-nique exploits the “first pass” of thecontrast agent, as 80% of a typicalgadolinium chelate leaks into theintravascular space within 5 minutes.32

Typical repetition times (TR) for the 3D T1-weighted gradient-echo(SPGR, T1 fast-field echo [T1-FFE],or fast low-angle shot [FLASH])sequence used for CE-MRA are lessthan 5 to 6 msec, with echo times (TE)of 1 to 2 msec and total scan times inthe 10- to 30-sec range. This type of3D sequence is ideally suited to MRA,as it has high spatial resolution and anintrinsically high signal-to-noise ratio(SNR). In addition, it is fast (can beperformed in a breath-hold), and canbe oriented and reformatted in anydesired plane.33-37 In addition, becauseintravascular signal is dependent on T1relaxation, rather than on inflow orphase accumulation, in-plane satura-tion and turbulence-induced signal lossare not problems.14

The relative signal intensity (SI) fora 3D gradient-echo acquisition is givenas38:

Equation # 2SI = [N(H) (1 – e–TR/T1)/(1-cos(α)e–TR/T1)] sin(α) e–TE/T2*

in which N(H), TR, TE, and α are theproton density, repetition time, echotime, and flip angle, respectively. Asshown in Equation #2, SI is maximizedwhen TR is long or T1 is short, since as(TR/T1) becomes large, e-TR/T1

approaches zero. Of these options,shortening T1 is the primary methodfor increasing intravascular signalintensity, as lengthening TR andthereby increasing scan time is notdesirable. The Ernst angle (αE), which

is important to understand to properlyoptimize parameters for a particularTR and expected T1, is defined as theflip angle maximizing the SI in Equa-tion #2, and can be expressed as:

Equation #3αE = cos –1(e–TR/T1)

Figure 1 shows an example of rela-tive signal intensity versus flip angle αfor different T1 values (assuming a TRof 5 msec with TE <<T2*). Note thesignificant increase in signal intensityas T1 approaches zero.

Contrast Dose and T1In order for blood to appear bright withrespect to background tissues during thearterial phase of dynamic 3D CE-MRA,sufficient contrast material must beadministered to transiently reduce thearterial blood T1 to less than that of thebrightest background tissue (ie, fat),which has a T1 of approximately 270msec. During the first-pass arterialphase, blood T1 is more related to infu-sion rate and contrast agent relaxivitythan to total contrast dose, as arterialphase imaging occurs well beforeintravascular contrast equilibration.This dynamic approach has the advan-tage of nearly eliminating steady-statebackground signal but at the same time

(considering non–blood-pool agents) isessentially a “one-shot” technique,meaning that once contrast is adminis-tered and dynamic imaging performed,the process cannot be repeated withouta second contrast bolus. If a secondbolus is administered, not only are therethe issues of expense and total patientdose, but residual soft-tissue enhance-ment can obscure vascular detail (as thebiologic half-life of gadolinium chelatesis approximately 90 minutes).39 Whenusing more than one injection, subtrac-tion of a precontrast mask is desirable,and in some instances mandatory, forthe second and third injections.40

Timing is extremely important indynamic CE-MRA. In order to appre-ciate timing issues, the relationshipamong injection rate, blood T1, andresultant signal intensity must first bereviewed. For times shorter than therecirculation time, contrast concentra-tion of the bolus in blood can beapproximated as13,14,41:

Equation #4[Contrast] = (rate of contrast injection in mol/sec)

(cardiac output in L/sec)

Although this simple equationignores bolus dilution at leading andtrailing edges, as well as extravasationof contrast in the lungs, a complexmodel originally developed for IV digi-

Figure 2. Blood T1 versus gadolinium infusion rate for changing cardiac output (CO), assuming no recirculation.


tal subtraction angiography arrives at thesame value for maximum contrast con-centration.42 Combining Equations #1,#2, and #4, and assuming no recircula-tion, blood T1 can be calculated for agiven cardiac output and injection rate.This is demonstrated in Figure 2 for car-diac outputs of 3, 5, and 7 L/min using astandard gadolinium chelate preparationwith a concentration of 0.5 mol/L.

Note in Figure 2 that for a “typical”cardiac output of 5 L/min, an injectionrate of 1 mL/sec yields a blood T1 of 36msec, while a 2 mL/sec injection givesa blood T1 of 18 msec. These T1 valuesare, in fact, much shorter than the T1 offat, even though our own experiencesuggests true T1 values are not quitethis short, being instead nearer doublethis estimate. Such a rapid injectionrate, however, can, but need not neces-sarily, be sustained over the entireacquisition time. For a typical 80-kgpatient receiving a “single dose” (0.1mmol/kg) of gadolinium, this equatesto 16 mL of a 0.5 mol/L gadoliniumchelate preparation. Thus a 1-mL/secinjection can be sustained for 16 sec, ora 2-mL/sec injection sustained for 8sec. Even though the acquisition timemay be considerably greater than this,by properly timing the bolus withrespect to image acquisition, as well asby realizing that a contrast boluslengthens as it propagates from thevenous to the arterial circulation, fulladvantage of the dynamic blood T1

shortening can be obtained. This islargely due to the unique way in whichMR data are acquired.

Fourier (k-space) ConsiderationsMR imaging does not map spatial datalinearly with respect to time as does con-ventional CT.43 With 3D MR imaging,the entire 3D Fourier or k-space data setis collected before reconstructing indi-vidual slices. Because k-space maps spa-tial frequencies rather than spatial data,k-space data does not correspond toimage space directly. Instead, differentregions of k-space data determine differ-ent image features. For example, the cen-ter of k-space, or “low” spatialfrequencies, dominate image contrast,whereas the periphery of k-space, or“high” spatial frequencies, contributemore to fine details such as edges.44,45

This means that the state of the intravas-cular T1 in large vessels is essentially“captured” at the instant correspondingto acquisition of central k-space.44 If cen-tral k-space is collected when arterialcontrast concentration is peaking but hasnot yet reached the venous structures(arterial phase), only the arteries willenhance. The fact that intravascular con-trast concentration may not be uniformover the entire acquisition time may gen-erate artifacts, but these have beenshown to be relatively minor providedthat the contrast is not rapidly changingover the critical central portion of k-

space and at least some “tail” of the con-trast bolus is present during collection ofhigh spatial frequency data at the periph-ery of k-space.41,44 Thus with proper tim-ing, gadolinium injection duration neednot be as long as the entire acquisitiontime, and in fact many protocols useinjection durations much shorter than theacquisition time with quite acceptableresults.46,47 For a given gadolinium dose,the injection strategy then becomes atrade-off between a fast injection (shorterT1—more intravascular signal) and longinjection (more uniform T1—fewer arti-facts). We have a general rule of thumbthat advocates injecting the total contrastdose over a duration of approximately50% to 60% of that of the acquisition forsingle station examinations.

Another factor that must be consid-ered is phase-encoding order. Tradi-tionally, phase encoding is performed“sequentially” so that central k-spaceis acquired at the mid-point of the scan.Alternatively, phase encoding can beperformed in a “centric” fashion suchthat central k-space is acquired at thebeginning of the scan.48,49 Althoughcentric phase-encoding order has beenshown to be somewhat more prone toartifacts if the contrast concentrationchanges rapidly during acquisition ofcentral k-space, it greatly simplifiesbolus timing (which will be discussedbelow) and is less susceptible to arti-fact from incomplete breath-holds.50,51

CONTRAST MATERIALINJECTION RATEAs shown in Figure 2, injection ratesgreater than approximately 2 mL/secbring a diminishing return with respectto T1 shortening. But as shown in Fig-ure 1, even small decreases in T1 leadto tremendous increases in signalintensity. By combining Equations #1through #4, signal intensity (for TR =5, TE = 2) can be plotted versus injec-tion rate for a fixed cardiac output (5L/min) and flip angle (Ernst angle) asshown in Figure 3. This is done bothdisregarding T2* effects, and using theapproximation that T2* = (0.66)T1.41

Figure 3. Relative signal intensity (SI) versus gadolinium injection rate for a spoiled gradient-echosequence. Flip angle (α) = Ernst angle, cardiac output = 5 L/min.


This graph demonstrates that signalintensity increases asymptotically asinjection rate increases, particularlywhen taking into account T2* effects,with negligible increases seen beyonda rate of 4 to 5 mL/sec.52,53

Different authors have taken differentapproaches to optimizing injection rate.Before the advent of high-speed gradientsystems, which permitted total acquisi-tion times of <1 minute (in the realm of abreath-hold), imaging times were on theorder of 3 to 5 minutes, and injectionswere typically performed so that a “dou-ble dose” (0.2 mmol/kg) was adminis-tered at a uniform rate over the entire scanduration.30,54 For a 4-minute scan on a 70-kg patient, this amounted to an injectionrate of just over 0.1 mL/sec. With themuch shorter acquisition times currentlypossible, many investigators advocate adouble dose (0.2 mmol/kg, or alterna-tively a set volume of 20 or 30 mL) ofgadolinium chelate with injection rates of 1 to 2 mL/sec. Others use up to a“quadruple dose” (0.4 mmol/kg) andrates of 4 to 5 mL/sec, while a new trend

now advocates much smaller doses.35,37,55-

57 We personally believe that using a sin-gle 20-mL vial of gadolinium chelatewith an injection rate of 1.5 to 2.5 mL/secis generally a good compromise betweenmaximizing arterial signal and minimiz-ing artifacts for breath-hold scans.58

Multi-station exams, large aneurysms ordissections, and cases in which venous

evaluation is important are different, andthese will be discussed shortly.

CONTRAST-ENHANCED MRA: BOLUS TIMING CONSIDERATIONSAs discussed above, the timing of theGd bolus acquisition of central k-space

Figure 4. Relative signal intensity (SI) in the aorta versus time following a 9.5-mL injection of gadoliniumchelate at a rate of 2 mL/sec.

A B C

Figure 5. Maximum intensity projection (MIP) images from different MRA studies demonstrating: (A) good arterial phase timing in an abdominal aortic study,(B) slow flow through an abdominal aortic aneurysm with resultant leading edge artifact (mild ringing) and decreased signal-to-noise ratio (SNR) in the iliacarteries, and (C) a carotid MRA with slightly delayed triggering leading to mild to moderate enhancement of the right jugular vein.


is extremely important. The contrasttravel time, defined as the time requiredfor contrast to travel from the injectionsite to the vascular territory of interest,is highly variable, as is the degree towhich the bolus “broadens” as it transitsthe heart and pulmonary circulation. Amean transit time to the abdominal aortaof 23 ± 5 sec, with a range of 13 to 37sec has been reported.59 An example oftime resolved abdominal aortic SI(which is proportional to [Gd]) after IVinjection of 9.5 mL gadolinium chelateat 2 mL/sec is shown in Figure 4.

As shown in Figure 4, note first thatthe arterial signal peaks relativelysharply (if the bolus were of longerduration, ie, larger total dose, the arter-ial peak would be somewhat length-ened). This observation has importantimplications. Because it has beenshown that intravascular signal inten-sity is determined by the Gd concentra-tion at the time the center of k-space iscollected,44 whether central k-space isacquired at time t1, t2, or t3 (Figure 4)has a dramatic influence over the resul-tant image. Perfect timing (t2) yieldsmaximum arterial signal with minimalvenous signal, as shown in Figure 5A.If, however, central k-space data isacquired too early (t1—while arterialGd is still increasing rapidly), “ring-ing” or “banding” artifacts of variableseverity can be generated, as subtlydemonstrated in the iliac arteries (Fig-ure 5B).44 These artifacts are alsoreferred to as “leading-edge” artifactssince they result from imaging duringthe leading edge of bolus arrival.Acquiring central k-space data too late (t3) leads to submaximal arterialsignal intensity and associated venousenhancement. These two factors can bedetrimental to evaluation of the arterialstructures. A mild example of suchdelayed timing is shown in Figure 5C.Hence, if one knows, or can predict ormeasure the time course of intravascu-lar Gd, the placement of central k-space can be tailored to achieve thedesired image characteristics, be itmaximum arterial signal, maximumportal venous signal, etc.

“Best-Guess” TechniqueSeveral solutions to proper bolus tim-ing have been offered. Perhaps thesimplest solution is to just make aneducated best guess. This involvesestimating the contrast travel timefrom the site of injection to the vascu-lar structure of interest. While this timeis highly variable, an experienced MRangiographer can achieve good resultsreliably by taking factors such as injec-tion site, age, cardiac output, and vas-cular anatomy into consideration.Sequential phase encoding is preferredwhen using a best guess, as this encod-ing strategy is more tolerant of timingerrors (see the discussion of CE-MRALogistic Considerations below).44

Using this technique, timing can becalculated as follows35:

Equation #5Imaging Delay = (Estimated Contrast Travel Time)+ (Injection Time/2) – (Imaging Time/2)

where Imaging Delay is the delaybetween initiating the bolus and startingthe scan (if negative, it refers to the delaybetween initiating the scan and startingthe bolus). Equation #5 times the injec-tion such that the mid-point of the bolusarrives at the desired vascular territory atthe mid-point (center of k-space) of thescan. For example, a scan time of 18 secwith an estimated contrast travel time of20 sec and a 20 mL gadolinium bolus at2 mL/sec (10-sec injection), the imagingdelay would be 20 + 10/2 – 18/2 = 16 sec(ie, start the acquisition 16 sec after initi-ating contrast injection). This timing isdesigned so the midpoint of the bolusarrives as the center of k-space is col-lected, meaning that depending on injec-tion duration, there may already bevenous enhancement by the time centralk-space is collected. A potentially bettertiming strategy is:

Equation #6Imaging Delay = (Estimated Contrast Travel Time)– (Imaging Time/2) + (Rise Time)

in which Rise Time is the expectedtime from arterial arrival to arterial

peak, typically 3 to 5 sec. This timing formula places the center of k-space at the approximate peak of contrastconcentration.

The most difficult aspect of the best-guess technique is estimating the con-trast travel time. As previously alludedto, it is better to image slightly late(increased venous phase) than slightlyearly (ringing/banding artifact, Figure5B). With this in mind, as a generalguideline for travel time from an ante-cubital vein to the abdominal aorta, westart with approximately 15 sec for ayoung, healthy patient. A young hyper-tensive or athletic patient would requirea few seconds less. A healthy, elderlypatient would require approximately 20to 25 sec. Patients with cardiac diseaseor an aortic aneurysm would be in the25- to 35-sec range. Patients withsevere cardiac failure in conjunctionwith aortic pathology might require upto 40 to 50 sec. We add 3 to 4 extra sec-onds if the IV line is in the hand.

Test-Bolus TechniqueA better, and only slightly more time-consuming way to estimate contrasttravel time is through use of a testbolus.59-61 With this technique, 1 to 2 mLof gadolinium (followed by a 10- to 15-mL saline flush) is injected at the samerate as planned for the actual injection.Multiple single-slice fast-gradient-echoimages of the appropriate vascularregion are then obtained as rapidly aspossible (at least every 1 to 2 sec) forapproximately 1 minute. To minimizeTOF effects, and thereby maximize con-trast conspicuity, the 2D test bolusimage should either be oriented alongthe vessel of interest (ie, sagittal or coro-nal for the aorta), or alternatively, be rel-atively thick (>1 cm) with a superiorsaturation band or a blood-nulling inver-sion pre-pulse. The time of peak arterialenhancement (contrast travel time) isthen determined visually or by using aregion-of-interest (ROI) analysis.Acquisition timing is then determinedwith Equation #6, keeping in mind that alonger bolus will take longer to peak


than a test bolus, and, therefore, a 3- to4-sec Rise Time should be included.Alternatively, if centric encoding isdesired (which is beneficial if the patientcannot breath-hold well50), the followingequation should be used, again anticipat-ing a Rise Time in the 3- to 4-sec range:

Equation #7Imaging Delay = (Contrast Travel Time) + (Rise Time)

Using the test bolus technique forabdominal aortic imaging, Earls et al59

demonstrated a mean aortic SNRincrease of 902% (15 patients) follow-ing a 1-mL test bolus of gadopentetatedimeglumine. By utilizing the resultantcontrast travel time (24.0 ± 12.6 sec),they were 100% successful in obtaininga “pure” arterial-phase study (as com-pared with 80% with best-guess tim-ing), and saw significantly increasedaortic SNR (29.8 versus 20.5).

There are three main drawbacks to thetest-bolus technique. First, setting up,performing, and analyzing the test boluslengthens the overall examination time.In experienced hands, the average timepenalty reported was <5 minutes.59 Sec-ond, the test bolus rapidly redistributesinto the interstitial space, therebyincreasing background signal. For suchsmall test doses, however, these effectsare negligible. Finally, there is noabsolute guarantee that the imagingbolus will behave identically to the testbolus. This is because of moment-to-moment patient variables such asvenous return and cardiac output, as wellas the different total volume of injection.For example, a test bolus administeredwith the arm at the patient’s side may notbe accurate if the arms are above thehead for the actual scan.

MR FluoroscopyA third method of contrast bolus timinguses extremely rapid “fluoroscopic”MR imaging.45,62-64 With this technique,2D sagittal gradient refocused imagesare rapidly (<1 sec/image) obtainedthrough the vascular structure of inter-

est, ideally using complex subtractionto improve the contrast-to-noise ratioand decrease artifacts. In this fashion,images are generated in near real-timeand are updated at >1 image/sec. Theoperator can watch the contrast bolusarrive, then switch over to the centric3D MRA sequence when the desiredenhancement is detected.

This technique has the advantage ofallowing real-time operator–dependentdecision making. This may be particu-larly advantageous in the evaluation ofcases with unusual or asymmetric flowpatterns, such as unilateral stenoses andslow filling aneurysms. As an example,an abdominal aortic aneurysm can bemonitored such that triggering is some-what delayed in order to allow for com-plete enhancement of the iliac vessels,which may fill slowly (Figure 5B). Theability to assess the vasculature “on thefly” under these circumstances is agreat asset, as is the time and contrastsavings associated with not performinga test injection.

A recent refinement is to electroni-cally shift the MR fluoroscopic monitor-ing to a proximal region of vascularanatomy to get more advance warning ofcontrast arrival.65 Using carotid CE-MRA as an example, MR fluoroscopicmonitoring in the chest shows contrastarriving in the subclavian vein, then rightheart, pulmonary arteries, left heart, arch,and finally, proximal great vessels.Tracking the bolus over such a longperiod makes precise triggering easier. Ifthe flow is very slow, the operator cancompensate to allow for greater targetvessel enhancement before triggering.This same work also demonstrates thatfluoroscopic triggering is improved byrecessing the absolute center of k-space afew seconds from the beginning of the3D acquisition, thus avoiding early trig-gering, which can cause the previouslydescribed leading-edge artifact.

Temporally Resolved 3D Contrast MRABecause proper bolus timing is so diffi-cult and a timing mistake can ruin the

study, some authors advocate “tempo-rally” resolved techniques wherebymultiple 3D datasets are acquired (orsynthesized) extremely rapidly (typi-cally 2 to 8 sec).66-70 With such tech-niques, bolus timing is no longer afactor, as multiple vascular phases areobtained without any predeterminedtiming (ie, inject and begin scanningsimultaneously). The operator simplyselects the desired image set, be it purearterial, maximum venous, etc. This isparticularly useful in the carotid andpulmonary arteries, where the venousphase is extremely rapid, and in thecase of vessels such as in the calf,where variable rate filling may occurdue to stenoses, occlusions, or rapidarteriovenous shunting due to soft-tis-sue inflammatory disease.66,70,71

The most straightforward way toaccelerate acquisition time is simply byscanning faster. This can be done usingsome combination of limiting the imag-ing volume, decreasing the resolution,decreasing TR, or using a parallelimaging technique, such as sensitivityencoding (SENSE).72-75 Recent devel-opments in gradient systems allow TRsof well under 2 msec on many systems,which is a three-fold improvementcompared with just a couple years ago.In the pulmonary arteries, Goyen et al73

achieved 4-sec temporal resolutionusing partial k-space filling (TR = 1.6msec), although slice thickness was rel-atively thick at 5.5 mm reconstructed to2.75 mm. In another approach, taken byFinn et al,74 the aorta is imaged in twoalternating planes (oblique sagittal andcoronal, TR = 1.6 msec) at approxi-mately 1 sec resolution per plane. Thisis in part possible due to a thick-slabacquisition (approximately 20 mminterpolated to 10 mm), meaning thedata is quasi-projectional. While off-plane reformats in this case are poor, in-plane maximum intensity projections(MIPs) (two planes) are of high resolu-tion. Finn et al has also used a 3D radialsampling scheme to further increaseresolution.76 Once high-resolution tem-poral data sets are obtained, even if thetemporal resolution is insufficient for


isolating the desired vascular phase,postprocessing techniques, such as cor-relation analysis can be used to synthe-size images of a particular type, in thiscase pure pulmonary arterial versuspure pulmonary venous.77

Simultaneous high spatial resolution,high temporal resolution, and increasedvolume coverage can be obtained usingthe 3D time-resolved imaging of con-trast kinetics (TRICKS) technique andits variants.67,71,78-80 In the most basicimplementation of this technique, k-space is divided into multiple blocks.The central block of k-space (whichcontributes most to overall image con-trast) is collected repeatedly every 2 to8 sec in an alternating fashion with theother blocks of more peripheral k-space. Images are then “synthesized” athigh temporal resolution by piecingtogether each unique central block of k-space with a linear interpolation of theremaining k-space blocks acquired inclosest temporal proximity. This tech-nique allows greater temporal and spa-tial resolution/volume coverage than isavailable with a streamlined conven-tional sequence. The technique lendsitself to “video” format, which allowspassage of contrast material to beviewed directly. The 3D TRICKS-typesequences are now finding their wayinto commercial packages. The biggestdrawback, other than the tremendouslylarge number of images generated, isthe large amount of processing time

required to perform the multiple recon-structions, although this is rapidlyimproving. Also, k-space discontinu-ities, in conjunction with varyingintravascular gadolinium concentra-tion, can potentially lead to artifacts.These artifacts, however, have beenshown to be small provided thegadolinium bolus is not too compact.78

A recent refinement in this strategy isto traverse k-space using radial or spiralprojections, which allow for slidingwindow reconstructions at very hightemporal and spatial resolutions.81 Onevariation, known as vastly undersam-pled isotropic projection reconstruction(VIPR) is particularly promising.82

CONTRAST-ENHANCED MRA: LOGISTIC CONSIDERATIONSThe following logistic considerationsapply to 3D contrast MRA in general,whether using best-guess timing, a testbolus, or triggering.

Artifacts: k-space–RelatedAs previously discussed, the key tomaximizing arterial enhancement is toimage such that the center of k-space iscollected when arterial contrast con-centration is at a maximum. Acquiringcentral k-space data after arterial con-trast has peaked leads to suboptimalarterial signal and increased venoussignal (Figures 4 and 5C). Acquiring

central k-space during the rapid ups-lope in arterial contrast arrival causes“ringing”-type leading-edge artifacts(Figure 5B). Even with optimal timing,similar artifacts can be generated withthe use of very compact boluses suchthat contrast falls rapidly while stillacquiring relatively central k-space.78

If the bolus is too short, such that thecontrast concentration tapers toorapidly toward the end of the acquisi-tion (high k-space frequencies), blur-ring of the edges results.44 Theseartifacts are not as severe if k-space ismapped in a sequential, rather than acentric, manner.44

Thus, when using best-guess timing,sequential phase encoding is more for-giving. Also, if using this technique,we suggest slightly overestimating the contrast travel time so that the(inevitable) timing errors are morelikely to cause increased venousenhancement than “ringing” artifacts.When using the test bolus techniquesuch that contrast travel time is known,either centric or sequential phaseencoding can be used. The advantageto centric phase encoding is twofold:less degradation with an incompletebreath-hold (see the discussion of res-piratory artifacts below); and simpli-fied operator logistics with respect totiming (no need to count backward tothe middle of the scan to figure outwhen to inject; just inject, wait theproper delay, initiate breath-hold, andstart imaging). For automated bolusdetection, centric phase encoding istypically used. Because contrast risetimes are not instantaneous, the delayfrom triggering to initiation of dataacquisition becomes crucial in order toavoid artifacts. Examining Figure 4, ifthe 3D MRA triggers at time t1, theappropriate delay is (t2 - t1). This delay,of course, varies by patient and vascu-lar territory. A delay of 3 to 6 sec typi-cally works well for abdominal aorticimaging. In patients expected to haveparticularly slow flow (eg, in patientswith congestive heart failure or ananeurysm), this time should be length-ened to 7 to 10 sec. In the carotid arter-

A B

Figure 6. Coronal subvolume maximum intensity projections from renal MRA studies demonstrating (A) good breath-holding, and (B) poor breath-holding.


ies, however, which have a more rapidupslope and a very short arteriovenouswindow, shorter trigger delays of 2 to 3sec are more appropriate in order toavoid jugular venous enhancement.63,83

Recessing the absolute center of k-space from the beginning of the scanimproves consistency by providinghigh arterial-phase contrast even if thetrigger is premature.

Artifacts—RespiratoryRespiratory motion causes ghosting,blurring, and signal loss.84,85 In 3D imag-ing, blurring occurs in the direction ofthe motion, whereas ghosting is mostpronounced in the slow phase-encodingdirection. For elliptical centric phase-encoding order, the ghosting is spreadout in both phase-encoding directions.86

Before the advent of fast imaging sys-tems, 3D CE-MRA acquisition timeswere on the order of 3 minutes, makingbreath-holding an impossibility. Thus,early CE-MRA images suffered fromsignificant image degradation, particu-larly when evaluating small vessels,such as renal arteries.14 Figures 6A andB illustrate the difference betweengood- and poor-quality breath-holding.

There is no disagreement amongresearchers regarding the beneficialeffects of breath-holding on abdominaland thoracic 3D CE-MRA.35-37,50,87 Hol-land et al87 evaluated the same 6patients using both breath-hold andnon–breath-hold renal 3D CE-MRA.They found that the non–breath-holdstudies failed to identify 3 of 4 periph-eral renal artery abnormalities, missed1 of 2 accessory renal arteries, andwere unable to visualize any of therenal artery bifurcations at the hilum.In contrast, breath-hold 3D CE-MRAidentified all abnormalities, all acces-sory renal arteries, and renal arterybifurcations at the hilum in all patients.In the carotid arteries, breath-holdingdoes not appear essential, although itdoes improve visualization of the archvessels.88

Maki et al50 looked at SNR loss andblur associated with incomplete breath-

holds. Their work demonstrated that fortypical MRA acquisition matrices andscan times, breath-holding during theacquisition of the central 50% of k-space was the most important for mini-mizing blur. This means that for centricphase encoding, the majority of thebenefit from breath-holding can beobtained even if the patient beginsbreathing at the midpoint of the scan.Data from Wilman et al51 shows that themore fully “centric” the phase-encod-ing order is, the less the artifact frompartial breath-holds, and thus the pref-erence for an elliptical versus a linearcentric phase encoding order. Thesedata suggest that for centric phase-encoding orders, an optimistic assess-ment of a patient's breath-hold ability isin order, as there is little to lose if they“don't quite make it.” Recent data,however, has suggested that a largenumber of patients have a relativelylarge degree of diaphragmatic motioneven during an apparently successfulbreath-hold.86 This causes blurring andringing artifacts within the visceralarteries, obscuring small vessels andsubtle abnormalities (such as those thatoccur with fibromuscular dysplasia), ineffect counteracting the spatial resolu-tion inherent in the sequence. Thisargues for shortening abdominal MRAacquisition times to well below poten-tial breath-hold times, making rapidacquisition techniques, such as SENSEand the ultra-short TR discussed above,even more attractive. In anatomicregions that can remain truly motion-less, such as the calves and feet, longacquisitions work well.47

We have found that patients are bet-ter able to breath-hold if the procedureand expectations are described well inadvance of the actual scan. The vastmajority of ambulatory patients canbreath-hold for at least 20 to 30 sec,while many have endurance wellbeyond this. While performing a testbreath-hold before the scan is always agood idea, watching a patient's respira-tory pattern often gives you a clue totheir breath-hold potential. In ourexperience, patients who take slow

deep breaths with relatively long expi-ratory pauses have no problem com-pleting a 30- to 40-sec breath-hold.Patients who breath rapidly (>25breaths per minute) with no pausebetween expiration and subsequentinspiration often have trouble with abreath-hold of this duration. Sick orpostoperative inpatients often cannot,or will not, hold their breath at all,again making the high temporal resolu-tion scanning more attractive.

We recommend the followingbreathing strategy. For best-guess ortest-dose bolus timing techniques, pre-breathe the patient for 3 or 4 cycles(saying, “deep breath in... and breathout...”) timed so that the last inspira-tion is held (“deep breath in andhold...”) as the scan starts. Someauthors advocate the use of supple-mental oxygen during this pre-breath-ing.35,86 For fluoroscopic techniques,pre-breathe the patient 3 or 4 cycles,then have them relax, start the fluoro-scopic sequence, and wait for bolusarrival. Once the MRA sequence istriggered, use some pre-arranged sig-nal, such as squeezing the patient’s armin addition to a loud verbal “hold yourbreath” to initiate the breath-hold. Thisneeds to occur during the delay time,so that the patient is motionless oncethe actual data collection starts and thecritical center of k-space is collected.

Determining Imaging TimeThe duration of a 3D CE -MRA acqui-sition (ts) depends on several variables,and can be approximated as:

Equation #8ts ~ TR * YRES * ZRES * (fractional y FOV)/(SF)

in which TR is the repetition time,YRES and ZRES are the number ofphase-encoding steps in y and z, frac-tional y FOV is the asymmetric y field-of-view fraction, and SF is the speedupfactor due to parallel imaging tech-niques, such as SENSE.89 Repetitiontime is itself a function of echo time(TE), echo type (ie, full or fractional),


and bandwidth. Spatial resolution isdetermined by voxel size, which isdetermined by the following equation:

Equation #9voxel size = (y FOV/YRES) by (z FOV/ZRES) by (x FOV/XRES)

in which x, y, and z FOV refer to the x,y, and z fields of view, and XRES is thenumber of data points digitized duringreadout. Using fractional y field-of-view decreases the number of y phaseencodings, shortening scan time whilemaintaining resolution but collecting asmaller (or “rectangular”) FOV in they direction. This is particularly usefulfor large FOV studies where structuresat the periphery of the y FOV are notimportant, and wrap around (aliasing)can be tolerated. Parallel imaging tech-niques, such as SENSE, require the useof multiple (phased-array) coils. Thesetechniques collect fewer phase encod-ings than are prescribed (phase and/orslice direction), effectively reducingthe FOV(s) with resultant aliasing.75,89,90

This can decrease scan time by up to afactor of the number of coils in thephased array, depending on geometry.SENSE has, to date, been most usefulfor cardiac imaging, although itappears to have great potential withMRA as well.91,92 Using sensitivity datafor each phased-array coil obtainedfrom a quick “reference” scan, a math-ematical algorithm can then “unfold”the aliased data set, restoring the origi-nal prescribed FOV with only a smallgeometry-related loss in SNR whencoils are properly optimized. Oneimportant concept to remember whenusing SENSE is that the prescribedFOV must be such that there is noaliasing, or else the reconstructionalgorithm breaks down and causessevere aliasing artifacts. Thus, SENSEis most advantageous when a full rec-tangular FOV is desired.

All of these parameters must be jug-gled to achieve the required volumecoverage, spatial resolution, and SNRwhile keeping the scan time (ts) withinan acceptable range, which should be

as short as possible and certainlywithin the patient’s anticipated breath-hold duration (tb). Thus, there are mul-tiple trade-offs. Increasing resolution(YRES or ZRES with same overallslab thickness), increasing volumecoverage (ZRES with same slice thick-ness), or increasing SNR (decreasingbandwidth or changing from fractionalto full echo), all increase ts. Decreasingthe rectangular FOV or using a parallelimaging technique will decrease ts. TheSNR must also be considered. In gen-eral, for all these techniques, SNRscales as the voxel size times thesquare root of scan time ts.

38,41 Oneinteresting property unique to contrastMRA, however, is that all other thingsbeing equal (eg, not changing band-width), if the contrast injection rate isincreased proportionally with thedecrease in scan duration (ie, injectsame total contrast dose faster), theSNR is unchanged and CNR increases.Maki et al41 describe a mathematicalanalysis specific to breath-hold CE-MRA to optimize SNR by consideringthese variables.

A useful technique with no associ-ated time penalty is to zero-fill k-spacein multiple directions.93 Most MRscanners have an option to zero-fillboth frequency and phase to 256 or512. In addition, a 2x zero-fill in slicecan be specified, yielding twice thenumber of slices. While the slice thick-ness is unchanged, the slices now over-lap by one-half a slice thickness. Whilezero filling may not increase resolu-tion, it permits smoother reformats byreducing partial volume effects.

The art, then, of 3D breath-hold CE-MRA boils down to a careful choice ofthe parameters in Equations #8 and #9and is often an iterative process. As ageneral rule for non–time-resolvedabdominal MRA, we start with a fre-quency resolution (XRES) of approxi-mately 400, with zero-filling to 512 ×512 in frequency and phase, and times2 in slice. We then consider the maxi-mum estimated breath-hold duration(tb) and desired slice thickness. Weincrease ZRES to cover the required

volume. If an asymmetric FOV can beused, we do so. Alternatively, we useSENSE with a SF of 2 to 3.47,89 We thenincrease YRES until tb is reached.Always keep in mind the voxel dimen-sion in z versus that in y. They shouldbe roughly similar, as one fundamentalpurpose of a 3D data set is to generaterelatively isotropic resolution so thatmultiplaner reformats (MPRs) main-tain resolution in all planes.

If ts cannot be made less than tb afterthese manipulations, then sacrificesmust be made. Partial Fourier tech-niques can be used (ie, collect only afraction of y phase encodes) at the costof a decreased SNR. Alternatively,bandwidth can be increased, again atthe cost of a decreased SNR, or slicethickness can be increased at the costof decreased spatial resolution. HigherSENSE factors (>2.5 to 3.0) can beemployed, although this also decreasesSNR. Finally, incomplete breath-hold-ing can be anticipated at the cost ofincreased ghosting and blurring.

In cases where breath-holding is notas important, such as in the extremitiesor carotids, there is more flexibility withregard to choosing imaging parameters.In these circumstances, the limiting factor becomes contrast bolus duration.As previously described, a “compact”bolus (with respect to imaging duration)can cause artifacts such as ringing andblurring. So, assuming a double dose(0.2 mmol/kg) of a gadolinium chelate(0.5 mol/L) for a 80-kg patient adminis-tered at 2 mL/sec, the injection durationis only 16 sec. Assuming the bolus willbroaden by several seconds beforereaching the target vessels, this is stillless than approximately 20 to 22 sec. Toavoid artifacts, this needs to be a sub-stantial fraction of the scan time ts, andscan times much greater than approxi-mately 60 sec are likely counterproduc-tive. Nonetheless, this increase inimaging time allows for increased SNRand spatial resolution. When consider-ing carotids or peripheral vessels withsoft-tissue disease, the rapidity ofvenous return must also be considered.This limits scan duration, as the longer


the scan, the closer the proximity ofphase encoding to central k-space whenvenous signal arrives. This leads togreater venous signal and leading-edgevenous ringing artifact. With truly cen-tric sequences, however, carotid scantimes of up to 40 sec and lower stationperipheral scan times of well over 1minute have been shown to workextremely well, in part due to recircula-tion of contrast in the latter part of thescan.47,63

Patient PreparationThe more relaxed and informed apatient is, the more easily they canremain still and perform a long breath-hold. Therefore, reassurance and abrief description of the scan can reallyhelp. In patients who are particularlyanxious, premedication with sedatives,such as diazepam, may be useful. Thishelps the patient to relax and lie still,and also decreases cardiac output.14

This latter effect helps enhance imagequality, as arterial phase [Gd] increaseswith decreasing cardiac output (Equa-tion #4) (Figure 2).

Since most 3D CE-MRA is per-formed in the coronal plane, arm posi-tion is important to allow for a smallerFOV without aliasing, particularly ifusing SENSE. For carotid, large FOVaortic, or peripheral studies, the armscan remain by the patient’s side, asaliasing is not a factor. For pulmonary,small FOV aortic, renal, and mesen-teric studies, the arms must either beelevated out of the imaging plane usingcushions, or extended over the head.35

Both of these positions in which thearms are elevated have the added bene-fit of gravity-aided venous return fromthe IV injection site.

We recommend placing the IVbefore the patient is in the magnet. Thisalleviates anxiety and possible shifts inpatient position that may occur if it isplaced while in the magnet. If the MRAis performed with the arms by thepatient’s side, placing the IV in theantecubital vein is adequate. If the armsare positioned over the head, it is

preferable to place the IV below theantecubital fossa so it will not kink andobstruct should the patient’s elbowbend. A 22-gauge catheter is usuallyadequate, although for smaller diametercatheters (>22-gauge), warming thecontrast to body temperature maydecrease the viscosity enough to allowfor adequate injection rates.

Contrast InjectionAs previously described, the total Gddose and rate of delivery is a trade-offbetween maximizing intravascular sig-nal and minimizing artifacts. In theideal situation, a large dose at a highrate would be optimal. This, however,must be weighed against safety, practi-cality, and cost. We feel that for breath-hold CE-MRA with accurate bolustiming (test bolus or fluoroscopy), asingle dose (0.1 mmol/kg) adminis-tered at a rate of approximately 1.5 to 2mL/sec represents a good balancebetween these factors, depending ontotal scan time. In practice, however,once a bottle (typically 20 mL) ofgadolinium is opened, it is best to injectthe entire bottle, and hence we typicallyuse 20 mL. For best-guess techniques,we prefer closer to 0.2 mmol/kg (eg, 40mL) at the same approximate rate.These recommendations can, undercertain circumstances, be modified. Forexample, in a particularly small patientfor whom the volume of a single dose issmall, artifact from a compact bolusmay be generated. To remedy this,either the injection rate can bedecreased (thereby lengthening thebolus at the expense of intravascularsignal), or the dose can be increased (atthe expense of increased cost). If cost isan overwhelming concern, the dose andinjection rate can be decreased at theexpense of intravascular signal, or thedose alone can be decreased at the riskof compact bolus artifact.

In experienced hands, we find thatmanual contrast injection works quitewell. Since many smaller MR centersdo not have a power injector, this isoften the only choice. Manual injec-

tions are also simpler in some ways, asthere is less “plumbing” and lessopportunity for mechanical difficul-ties. A power injector, however, offersa more constant and standardized bolusdelivery. Either way, the contrast injec-tion must be followed rapidly by ade-quate saline flush to complete deliveryof the bolus and help flush the armveins. Suggested flush volumes rangefrom 15 to 50 mL, while most authorsreport using 15 to 20 mL.35,59 We sug-gest 25 mL of saline flush at the samerate as the Gd bolus. For hand injec-tions, a standardized tubing set thatallows rapid switch-over between Gdand saline injection works well (Smart-set, TopSpins, Plymouth, MI).

Imaging Different Vascular PhasesWith most 3D CE-MRA studies, exten-sive efforts are made to image the arter-ial phase optimally, as has beendiscussed extensively. Once the arterialphase data set is collected, however, thesequence can (and should) be repeatedto obtain venous and equilibriumphases. Later phases are useful in eval-uating unsuspected dissections, portalvenous or venous structures, parenchy-mal enhancement patterns, and perhapseven renal glomerular function.94,95

Temporal resolution is determined by acombination of scan time and the timerequired for the patient to prepare foranother breath-hold. Aside from tech-niques such as 3D TRICKS, the fasterthe scan, the better the potential tempo-ral resolution.67 In cases in which hightemporal resolution is desired (such asin the carotids or in cases of arteriove-nous shunts), the previously describedstreamlined techniques can be used in asingle breath-hold, although SNR andspatial resolution decrease.66 In moreconventional breath-hold 3D CE-MRA, with a scan time of approxi-mately 20 to 25 sec, we advocate givingthe patient approximately 8 to 10 sec to“catch their breath,” then scanningagain. In this manner, the second arteri-ovenous phase occurs approximately


35 sec after peak arterial phase. This isusually well timed for the portal venousphase, and often adequate for evaluat-ing venous structures and parenchymalphases in the visceral organs. Some-times, a third phase may be desired,usually to evaluate slow venous returnor the renal collecting system.

Because these later phase examina-tions occur more in the equilibriumphase of Gd distribution, intravascularT1 is increased, and therefore signal isreduced as compared with the arterialphase (Figure 5C). In order to maxi-mize signal in these later phases, theflip angle should, if possible, bereduced (Figure 1). While the optimumflip angle for the arterial phase may be30° to 40° at a TR of 4 to 6 msec, the

optimal flip angle decreases to 15° to25° in later phases. To image Gdexcreted into the collecting system andureter, use a high flip angle (45° to 60°)with the widest bandwidth and shortestpossible echo time.

Postproccessing and DisplayThree-dimensional CE-MRA producesa contiguous volume of image data. Fora typical body MRA study, this volumeis asymmetric, perhaps 400 × 300 × 64mm. For a representative acquisitionmatrix size of 400 × 150 × 32, thisyields a true voxel size of 1.0 × 2.0 × 2.0mm. For diagnostic purposes, this dataset is best viewed interactively using acomputer workstation allowing for thin

multiplanar reformation (MPR).1,58 Inthis manner, thin (1.0 to 2.0 mm in thiscase) slices can be viewed in axial,sagittal, coronal, or oblique planes, thuseliminating overlapping structures andunfolding tortuous vessels. Becausethese reformatted slices remain thin andare not projections (see below), they notonly provide the best achievable con-trast, but also minimize the chance ofdiagnostic error.1,58,96

Because thin reformations showonly short vascular segments, it isadvantageous to utilize the MIP post-processing technique.97,98 Using thisalgorithm, the user first specifies thesubvolume thickness and desired view-ing plane. The algorithm then generatesrays perpendicular to the viewing

plane, takes the maximumvalue of any voxelencountered along thatray, and assigns that maxi-mum value to the corre-sponding pixel in theoutput image. This tech-nique is fast, and worksextremely well with tech-niques such as CE-MRA,since vascular signal istypically much greater

than background signal. It providesprojection images that are quite similarin appearance to conventionalangiograms. The MIP algorithm isdemonstrated in Figure 7, which showsa full thickness coronal MIP of the tho-racic aorta (Figure 7A), a thin-sliceaxial reformation through the aorta atthe level of the pulmonary arteries (Fig-ure 7B), and an oblique subvolumeMIP taken from the axial reformatdemonstrating the origins of the greatvessels and a large Stanford B aorticdissection (Figure 7C). As can be seen,this technique is very helpful for dis-playing complex vascular anatomy,particularly when vessels are not ori-ented along a single plane.

Despite its usefulness, the MIP algo-rithm is subject to artifacts. Perhaps themost common artifact arises when sta-tionary tissue has greater signal inten-sity than the vascular structures of

A C

Figure 7. Stanford type-B aortic dissection. (A) Full-volume oblique sagittal maximum intensity projections(MIP). Note the bovine arch. (B) Axial multiplanar reformat showing the two lumens with mural thrombusin the false lumen. (C) Subvolume MIP demonstrating the intimal flap with paired true and false lumens.

B


interest, as can occur in the presence ofother vessel segments, fat, hemorrhage,metallic susceptibility artifacts, ormotion artifacts. This, in turn, leads tothe mapping of nonvascular signal intothe projection image and causes a dis-continuity in vessel signal, potentiallymimicking a stenosis or occlusion (Fig-ures 8A and B).1 This type of artifact isbest overcome by minimizing the thick-ness of the MIP subvolume, therebyexcluding as much extraneous data as possible. Other artifacts inherent tothe MIP technique have beendescribed.99 These mainly consist ofunderestimating vessel lumen, and aremore of a problem with TOF or phase-contrast techniques. Because of thesepotential pitfalls, most authors agreethat the MIP images should be used as aroadmap, utilizing the source images fordefinitive diagnosis.1,58,99 A new studysuggests that volume rendering is moreaccurate than the MIP and is similar tousing MPR, but should still not replacecareful evaluation of the source imagesand MPRs.100

Subtraction techniques are also use-ful in image evaluation, particularly invascular regions not subjected to sig-nificant respiratory motion. Theseareas include the extremities, pelvis,and carotids. A digital subtraction

MRA can be accomplished most easilyby subtracting precontrast from post-contrast magnitude (reconstructed)images. Provided the patient maintainsthe same position on both studies, sub-traction will eliminate background sig-nal and improve vessel conspicuity.This technique has been used success-fully in the extremities.40,101 An improve-ment on this technique, particularly for thicker-slice or 2D examinations,involves complex subtraction of theprecontrast and postcontrast “raw” (k-space) datasets. This overcomesphase differences in voxels that containboth stationary and moving spins (ie,very small vessels), allowing forincreased vascular signal.102,103

CONTRAST-ENHANCED MRA:CLINICAL APPLICATIONSThis section reviews techniques andimaging parameters for selective bodyCE-MRA studies. The parameters pro-vided here were largely derived fromclinical experience with the PhilipsIntera 1.5-T system (Philips MedicalSystems, Bothell, WA). The generaltechniques described here are easilyapplicable to other systems, althoughthe parameters may need to be extrapo-lated as appropriate.

Based on the previous discussion,several general principles apply whenperforming CE-MRA. First, TR and TEneed to be as short as practical withoutexcessively increasing the bandwidth.Echo time can, and should, be mini-mized using a partial (asymmetric)echo. For the Intera system using anXRES of 400 with a water-fat shift of approximately 0.8 (bandwidth of ± 36.8 kHz), minimum TR/TEs ofapproximately 4.0 to 5.0/1.2 to 2.0msec are possible (somewhat depen-dant on choice of coil). Some systemsare capable of even shorter TRs,although because SNR decreases as thesquare root of bandwidth, care shouldbe taken not to increase bandwidthexcessively to achieve minimaldecreases in TR. Second, phase (y) andslice (z) resolution should be relativelybalanced and ideally no greater thanapproximately one-third the smallestvessel of interest.104,105 In general, mostMRA practitioners tend to let the z res-olution (slice thickness) drift somewhatlower than the phase (y) resolution.Note that the resolution in the fre-quency (x) direction is typically some-what greater than in the phase and slicedirections. For most abdominal appli-cations, y and z resolutions of 1.8 to 2.4 mm are adequate, although for

A

Figure 8. Demonstration of a potential maximum intensity projection (MIP)artifact. (A) Subvolume axial MIP through the right renal artery looks normal.(B) Careful review of the source coronal images demonstrates how the MIPvolume (white lines) incorporates part of the ectatic aorta (black arrow),masking the high-grade right renal artery stenosis. B


imaging large structures, such as theportal vein or thoracic aorta, where finedetail is not of great concern, increas-ing resolution to 2.5 to 3.0 mm is oftenacceptable,106 and for vessels, such asperipheral below-the-knee vessels or carotids, isotropic resolutions of 1.0 mm are appropriate.47,107 Third, tobenefit from breath-holding, which isessential in the abdomen, the number ofslices and phase-encoding steps mustbe balanced between adequate volumecoverage, adequate resolution, and estimated breath-hold time. For a typi-cal field-of-view of 400 × 300 mm, 192 phase encodings corresponds to a yresolution of 2.1 mm. For 192 phaseencodings and a TR of 5 msec, approxi-mately 35 slices can be obtained in a 25-sec breath-hold. These imagingtimes can be further reduced using par-tial Fourier imaging (halfscan, 1/2 num-ber of excitations) or SENSE. Forexample, a SENSE factor of 2.5 with anFOV of 400 × 400 and the same TR andYRES, the same volume could beobtained in just over 13 sec. Alterna-tively, 65 slices could be obtainedinstead of 35 in the same 25 sec. Finally,for most arterial phase studies, it is bestto use an automated bolus detection

technique. This is particularly importantfor renal and carotid studies, wheredelayed timing causes artifacts due toadjacent venous opacification. If auto-mated bolus detection is not available, atest bolus can be used. If neither of thesetechniques is feasible, educated best-guess timing can be used as a fallback.

Abdominal Aorta/Iliac ArteriesThree-dimensional CE-MRA of theabdominal aorta and iliac arteries ismost often performed to evaluateaneurysmal disease (Figures 5A andB), dissection (Figure 9), or as part of aclaudication work-up. The acquisitionis performed in the coronal plane tofacilitate evaluation of the iliac andrenal vessels, and a phased-array bodycoil should be used whenever possible.The imaging volume can be prescribedfrom a breath-hold sagittal 2D black-blood gradient-echo or true fast imag-ing with steady state precession (trueFISP) “bright blood” localizer, or alter-natively from an MIP reformat of a fast low-resolution axial 2D TOFsequence. A breath-hold localizer ispreferred, as it captures the anatomy inthe same position in which the CE-

MRA will be performed, thus decreas-ing the chances of a vital structure(such as the anterior aspect of ananeurysm) falling outside the FOV. Italso gives the patient a chance to prac-tice a breath-hold when the outcome isnot crucial.

A 400 × 300 mm FOV with a 400matrix and y resolution of approxi-mately 2 mm is a good starting pointfor CE-MRA of the abdominal aorta(approximately 400 × 200 matrix). Ifthe patient is large or has iliac/thoracicaortic disease, the FOV and slice thick-ness can be increased as needed. Aspreviously described, slice thicknessand number of slices are chosen basedon desired resolution and breath-hold-ing capability. Ideally, true slice thick-ness should be approximately ≤2.4mm. For very large aneurysms, how-ever, it may be necessary to increaseslice thickness to 3 or 4 mm. Breath-holding is important for aorto-iliacevaluation, particularly if good evalua-tion of the renal arteries is desired.Often, however, surprisingly good aor-tic and iliac image quality can beobtained even for patients who breath-hold poorly.

Because large aneurysms have aslow, swirling flow, extra time isrequired for the aneurysm to opacifymaximally and to allow for filling ofthe iliac arteries (Figure 5B). Whenusing fluoroscopic bolus detection, wefind it useful to trigger approximately 2to 3 sec beyond where we normallywould (assuming a flouroscopy to 3DCE-MRA delay of 3 to 4 sec). With alarge aneurysm or known dissection,we typically use a full 40 mL of Gd;otherwise, 20 mL works well. Obtain-ing a delayed-phase image immedi-ately following the arterial phase isrecommended, particularly for dissec-tions or very-slow-filling aneurysms.70

Thoracic AortaAs with the abdominal aorta, thoracicaortic MRA is most often performedfor aneurysm or dissection (Figures 7and 10). When evaluating the thoracic

B

Figure 9. (A) Coronal maximum intensity projec-tion from an abdominal aortic MRA demonstratinga focal infrarenal dissection. (B) Coronal sourceimage showing the site of intimal disruption.

A


aorta, an oblique sagittal 3D volumeoriented with the arch typically worksbest, minimizing the total number ofslices. Obtaining two dynamic phases

is extremely important, particularlywith dissections, where slow-fillinglumens may opacify much better onthe venous phase (Figures 10A and B).

We typically use 40 mL of Gd in thepresence of a known dissection.

When targeting only the thoracicaorta, a phased-array body coil and anFOV of approximately 380 × 266 mmshould be used, as a moderate amountof aliasing can be tolerated. Whenevaluating the entire thoraco-abdomi-nal aorta, the built-in body coil is typi-cally used, as most phased-array coilsdo not have sufficient craniocaudalcoverage. Using an FOV of approxi-mately 440 × 290 mm, the entire aortato the bifurcation can usually be cov-ered, including the renal artery origins(Figure 10D). In either case, the ori-gins of the great vessels should beincluded in the imaging volume. Trueslice thickness can be as great as 3 mm,although if renal artery evaluation isimportant, thinner slices (approxi-mately 2.4 mm) are desirable. Totalnumber of true slices depends on thesize and tortuosity of the aorta, as wellas slice thickness, but is typically in therange of 30 to 40.

A complete evaluation of the thoracicaorta includes multiplanar cardiac-gated black-blood imaging of the aorticwall to detect intramural hematoma, an important prognostic indicator,108

although a complete description of thisis beyond the scope of this review.

Renal ArteriesContrast-enhanced MRA of the renalarteries is similar to an abdominal aor-tic study, with the exception thatincluding the entire iliac vasculature isless important (although useful, if pos-sible). This allows for a thinner coronalslab, and therefore decreased slicethickness and/or acquisition time. Typ-ically, approximately 30 slices with atrue slice thickness no greater than 2.2to 2.4 mm, a matrix of 400 × 230, andan FOV of 400 × 300 mm is sufficient.This allows for high resolution (1 × 1.7× 2.2 mm) with relatively short scantimes (under 20 to 25 sec). As dis-cussed previously, breath-holding isextremely critical, as distal branchesand accessory renal arteries are diffi-

A B

C D

Figure 10. (A) Arterial and (B) immediate delayed phase full-volume maximum intensity projections(MIPs) from a thoraco-abdominal MRA. The more posterior false lumen opacifies more rapidly andextends into the iliac arteries, with the delayed phase better demonstrating the more anterior true lumen.(C) Sagittal and (D) coronal arterial phase subvolume MIPs help define the extremely complicatedanatomy in this case. A stenotic right renal artery and superior mesenteric artery originate from the truelumen, whereas the celiac axis and a proximally occluded left renal artery originate from the false lumen.


cult to evaluate in the presence ofdiaphragmatic motion.86 This and otherrecent studies suggest that minimizingacquisition time (and thereby motionartifacts) is extremely beneficial. Tech-niques as simple as limiting the vol-ume coverage or as advanced asSENSE are good choices for the renalarteries.91

Figure 11 demonstrates a mild tomoderate right stenosis and a moderateto severe proximal left renal arterystenosis (Figure 11A), with angio-

graphic correlation (Figures 11B andC). Note that 3D phase-contrast imag-ing (Figure 11D) can be a usefuladjunct in the evaluation of renal arterystenosis. While the CE-MRA imagesdemonstrate anatomy, the PC imagesshow signal loss in regions of turbu-lence or extremely slow flow, suggest-ing which lesions are of hemodynamicsignificance.109,110 In this example, thereis no right-sided signal loss, but mod-erate signal drop-out on the left, sug-gesting the left-sided lesion ishemodynamically significant, whichwas confirmed by X-ray angiography(Figure 11B). Note also the left-sidedcortical thinning (Figure 11E), a sec-ondary sign of significant renal arterystenosis, as well as the incidentallynoted small left renal artery aneurysm(Figure 11A), which was nicely cor-roborated on a later phase X-rayangiogram (Figure 11C). Figures 12Aand 12B demonstrate an MIP and avolume-rendered image from a patientwith proven fibromuscular dysplasia.

Transplant renal arteries can be eval-uated easily as well.111,112 An example isshown in Figure 13, in which the trans-plant artery has thrombosed with subse-quent infarction of the renal graft. Sincetransplants are in the pelvis, respiratorymotion is not as problematic, andbreath-holding is less critical. Nonethe-less, it is best to have the patient breath-hold if possible. While coronal imagingwith a 320- to 360-mm FOV is usuallyadequate, oblique sagittal imaging canbe performed with a smaller FOV. Thisgives higher resolution, but coversmuch less vascular territory. Transplantrenal veins are easily seen on the venousphase of the study.

Mesenteric ArteriesThe proximal mesenteric vessels areincluded on renal or abdominal aorticCE-MRA studies, and are usually wellvisualized, even in the presence of respi-ratory motion (Figure 14). The inferiormesenteric artery is often seen as well.

Dedicated mesenteric artery studies,typically for the evaluation of mesen-

C

A

B

D

Figure 11. (A) Full-volume maximum intensityprojection (MIP) from a renal artery study withangiographic correlation (B and C) demonstratinghigh-grade left, and moderate- to high-grade rightrenal artery stensosis, as well as a small left renalartery aneurysm (arrows). (D) Axial MIP from aphase-contrast study demonstrates no signaldrop-out on the right, with mild signal drop-out onthe left, indicating probable hemodynamic signifi-cance of the left renal artery stensosis. (E) Exam-ining the source images can provide addition cluesto the hemodynamic significance of a lesion, as inthis case where smooth left renal cortical thinningis seen.

E


teric ischemia or tumor encasement, areperformed in either the coronal or sagit-tal plane, and are better suited to evalu-ating the more proximal vessels.113,114

Slice thickness should be similar to thatof a renal artery study (less thanapproximately 2.4 mm), particularlywhen using the coronal plane, as theproximal mesenteric arteries travel pri-marily anteriorly. Keep in mind that the hepatic, gastroduodenal, and distalsuperior mesenteric arteries may be fur-ther anterior than is usually included ina renal study, and, therefore, volumecoverage will need to be increased (orshifted anteriorly) in order to see thesestructures.

Portal/Hepatic VeinsPortal venous or hepatic vein studiesare performed in a coronal plane with aphased-array body coil (Figure 15).115

Slice thickness can be increased toapproximately 3 mm to adequatelycover the relevant portal venous struc-tures. Axial bright-blood scout images,such as true FISP, are useful for deter-mining the anterior and posteriorextent of the portal/hepatic veins, andmay even be diagnostic.

With portal venous and hepaticvenous studies, timing is not as crucialas with arterial studies, since the portalvenous phase is longer and thus easierto isolate. After obtaining the arterialphase images, which allow for arterialevaluation, an approximately 10-secpause allows the patient to recoverfrom the breath-hold. Imaging is thenrepeated, followed by a second pauseand a third scan. In this manner, thesecond set of images usually providesa nice portal venous and hepaticvenous phase. The third set may showcollaterals to better advantage.

A B

Figure 12. (A) Maximum intensity projection and (B) volume-rendered image from a renal MRA in apatient with proven fibromuscular dysplasia.

A B

Figure 13. (A) Oblique full-thickness maximum intensity projection from a renal artery MRA performed ina renal transplant patient with abdominal pain and rising creatinine 2 weeks following a motor vehicle acci-dent. The arrow shows the occluded transplant renal artery. (B) A delayed phase source image shows nearcomplete absence of renal blood flow. The graft could not be salvaged.

Figure 14. (A) Full-thickness coronal maximumintensity projection (MIP) from an abdominalMRA. (B) The sagittal subvolume MIP demon-strates incidental complete occlusion of the proxi-mal celiac axis. Note the long moderate main rightrenal artery stenosis with poststenotic dilatation. A B


This protocol has the added advantageof essentially being a dynamic liver study.Hence if there is concern over a hepatic

mass, this can be evaluated dynamically,provided it is included in the imaging vol-ume.116 Under these circumstances, fur-

ther delayed phases should be obtainedout to 5 or 10 minutes.

Pulmonary ArteriesThree-dimensional CE- MRA of thepulmonary arteries is gaining accep-tance (Figure 16). A recent large-scalestudy comparing MRA with conven-tional digital subtraction angiogrphy(DSA) suggests that pulmonary MRAis similar in accuracy to that of helicalCT, being sensitive and specific foremboli in segmental and largervessels.117 Creating high-quality pul-monary MR angiograms, however,requires considerable care. Goodbreath-holding or extremely shortacquisition times are extremely impor-tant, and, unfortunately, many patientswho require pulmonary angiographyare significantly respiratory compro-mised and therefore cannot breath-hold well. A phased-array body coilshould be used whenever possible.

The exam must be tailored to theindividual patient. Two approaches canbe used—single-acquisition coronal,and double-acquisition sagittal (sepa-

Figure 15. Coronal 3D Gd MRA of portal vein with 20 mL Gd-BOPTA. (A) Coronal MIP of entire volume. (B) Axial reformation through portal vein with dotted linesindicating optimal volume for performing subvolume MIP. (C) Subvolume MIP shows portal vein to better advantage by excluding more posterior structures.Note anomalous posterior division of right portal vein (small arrow) arising prior to left portal vein (large arrow).

(B)

Figure 16. Coronal MIP of a pulmonary MRA(A) shows right-sided aortic arch with mirror-image branching and no normal pulmonary arterydue to congenital atresia. Subvolume MIPs (B)and (C) show right pulmonary artery (arrow) aris-ing from dilated ascending aorta (Waterstonshunt) and left pulmonary artery (curved arrow)arising from descending aorta (Potts shunt).

A

B

C

Main Portal Vein

Main Portal Vein

SMV

(A)

(C)


rate injections and acquisitions for each lung).117,118 Which approach to use is a matter of preference, breath-hold capability, and machine speed. For the double-acquisition sagittalapproach, the volume coverage foreach acquisition is smaller, and a high

degree of rectangular FOV can be used. This reduces acquisition time, oralternatively, allows for increased spa-tial resolution (as compared with a sin-gle coronal acquisition). In addition,the arms can remain by the patient’sside, as the fold-over direction is ante-rior-posterior. On the down side, twoseparate acquisitions require morescanner and postprocessing time andmore contrast, and the pulmonary trunkand proximal pulmonary arteries maybe excluded (although this is not usu-ally clinically relevant).

For either approach, a total slab thick-ness of 100 to 150 mm is typical, with trueslice thickness optimally <2.6 mm.118 Fre-quency FOV is typically 300 to 360 mm,depending on patient size, with a higherdegree of rectangular FOV possible forthe sagittal approach (approximately 45%to 70% versus approximately 65% to70%). Matrix size should be on the orderof 320 × 160, such that overall resolutionis <1.0 × 2.0 × 2.6 mm. We always per-form at least one practice run to evaluatethe patient’s breath-holding capability,ensure the patient will comply with thebreathing instructions, and check that anyfold-over is acceptable.

For a TR of 5 msec, acquisitiontimes for coronal and sagittal acquisi-

tions will be approximately 25 and 20sec, respectively. Shorter scan timesare certainly desirable, especiallygiven the difficulty these patients oftenhave with breath-holding and the car-diac motion within the chest. Weexpect increased use of high temporalresolution techniques in the immediatefuture.73,82 Correct bolus timing isimportant, but not as essential as it isfor other vascular territories. It is diffi-cult to isolate the pulmonary arterialphase from the pulmonary venousphase, although this is possible withextremely high temporal resolution orcorrelation analysis.77 Fortunately,however, this is not a big problem, aspulmonary arteries and veins are easilydistinguished on coronal images. Themain pulmonary artery should be usedfor fluoroscopic triggering or calibrat-ing a timing bolus. If these techniquesare unavailable, transit time is typically5 to 15 sec. Gadolinium doses areapproximately double dose for coronalimaging and single dose per side forsagittal imaging.

Extracranial Carotid ArteriesCarotid CE-MRA is typically per-formed using a single contrast injec-

D

Figure 17. Carotid MRA. (A) The full-thicknessmaximum intensity projection (MIP) demon-strates the great vessels extending from the arch,although there is some signal drop-off due to thecoil. (B) Right and (C) left sagittal subvolumeMIPs demonstrate occlusion and high-gradestenosis of the internal carotid arteries, respec-tively. (D) Note also the high-grade left vertebralartery “tandem” stenosis in the coronal subvol-ume MIP.

A B C


tion in the coronal plane (Figure 17).More so than in any other vascular ter-ritory, elliptical centric phase-encodingorder and nearly perfect timing isrequired to isolate the arterial phase, asjugular venous enhancement canseverely degrade arterial evaluation.63

This is due to the unique characteris-tics of the intracranial circulation, withits short arteriovenous window (4 to 6 sec), and the blood brain barrier,which prevents significant extractionof contrast (Figure 5C). While onewould expect shorter acquisition times(high temporal resolution) to increasethe likelihood of isolating a pure arter-ial phase, excellent results have beenachieved with long imaging times (40 sec to > 2 min) and high spatial res-olution in combination with ellipticalcentric phase encoding.63,79,83,119

To accurately characterize a carotidstenosis according to North AmericanSymptomatic Carotid EndarterectomyTrial (NASCET) criteria, particularlyone with tortuous anatomy and high-speed turbulent jets, an accurate mea-surement of stenotic diameter isrequired.120 Given this, we suggest true1-mm isotropic or better resolution,using as short a TE as practical todecrease signal dephasing in the oftenvery turbulent stenotic lumen. Tomaintain adequate SNR at these highresolutions, a birdcage or phased-array

B C

A

D

Figure 18. (A) Composite coronal maximumintensity projection (MIP) from a three-stationperipheral moving-table MRA performed for bilat-eral calf claudication. The upper and middle sta-tions were performed in a coronal plane, eachrequiring <12 sec. The lower station was acquiredas 2 sagittal slabs with true 1-mm isotropic reso-lution and required 70 sec. This allowed for inclu-sion of the pedal arteries. The upper and lowerstations were performed using SENSE. Note theextensive disease throughout the arterial struc-tures, with bilateral high-grade stenoses of theexternal iliac arteries. (B) The 1-mm isotropiclower station resolution provides excellentanatomic detail of the trifurcation (coronal subvol-ume MIP) and (C) right and (D) left pedal arteries.These subvolume MIPs show the dorsalis pedisarteries are widely patent bilaterally, but the leftposterior tibial artery is occluded distally, with theright lateral plantar artery occluded at its origin.


coil is necessary. Unfortunately,depending on patient habitus, olderneck coils suffer severe sensitivitydrop-off near the arch, making evalua-tion of the important great vessel ori-gins a problem (Figure 17A). Recentcoil advances (combination birdcageand anterior-posterior surface coils),however, make reliable evaluation ofthis important region possible as partof a routine carotid examination.83

An FOV of 360 × 180 mm typicallyworks well, allowing visualization downto the arch. The problem of arm aliasinginto the chest is minimized due to inher-ent coil drop-off laterally, although thismust be considered when using thebuilt-in body coil (which is not sug-gested). A total slab thickness of 50 to65 mm is usually adequate. Two-dimen-sional TOF or true FISP images can beused to evaluate the extent of the verte-bral and carotid arteries for localizationto ensure the CE-MRA volume is mini-mized. Typical scan parameters mightbe 384 × 384 × 60 matrix with 1-mmslice thickness (reconstructed to 512 ×512 × 120), for a true resolution of 0.9 ×0.9 × 1.0 mm. For a TR of 5 msec, acqui-sition time for such a sequence would beapproximately 50 sec. Carotid arteryMRA is an exception to our general rulethat injection time should be on the orderof 50% to 60% of acquisition time, asrapid recirculation of contrast seems tosupport the longer acquisition times forthe high-resolution sequences. We sug-gest 20 to 30 mL of Gd at 1.5 to 3.0mL/sec. As stated previously, breath-holding may help for evaluation of thearch vessels but is not necessary for thecarotid arteries.88

Peripheral MRAThe restricted FOV inherent to MRscanners has limited CE-MRA to ana-lyzing only 400- to 500-mm chunks ofvascular anatomy at one time. Whilethis is acceptable for carotid, aortic,renal, pulmonary, and mesenteric stud-ies, it has made peripheral CE-MRAstudies challenging. While multipleinjections and multiple acquisitions

with intervening table movement canbe performed,121 a recent revolutionaryadvance has eliminated the FOV bar-rier. By moving the table while scan-ning, a single bolus of contrast can beimaged multiple times as it flows downthe legs (Figure 18).122-124 This “mov-ing-table” or “bolus-chase” approachallows arterial imaging from theabdominal aorta to the mid-foot in justa few minutes, and is available in dif-ferent variations commercially. Evenwhole-body MRA using a single injec-tion has been described.125

Once the FOV issue for peripheralMRA has been addressed, two furtherobstacles must be overcome. First, spa-tial resolution must be adequate foroperative planning, as below-the-kneearteries are on the order of 2 to 3 mm,and grafting to an inadequate vessel canbe a clinical disaster. In general, theupper- (abdominal) and middle- (thigh)station vessels are relatively large, andresolution requirements are not as strictas in the lower- (calves/feet) station,where vessels may be only 2 to 3 mm.Thus, spatial resolutions on the order ofthose previously discussed for the aortaand visceral arteries, with true 2.5 to3.0 mm through plane resolution, arelikely adequate in the upper and middlestations. The lower station, however,requires spatial resolution more inkeeping with a carotid examination,with true 1-mm isotropic resolution anattainable goal.92 The craniocaudalFOV for each station must be approxi-mately 420 to 480 mm with at least 30mm of overlap between stations.

The second obstacle is that of venousenhancement, which must be absent orminimal, as the numerous paired veins inthe leg can easily obscure and confusethe arterial anatomy. It seems surprisingthat numerous early works demonstratedthe complete lack of venous enhance-ment in the majority of cases, eventhough acquisition in the lower stationdid not begin until approximately 60 secafter initiating upper-station acquisi-tion.123,124,126 Several investigators havebegun to explore this issue, which seemsto be a complicated function of imaging

delay, contrast injection rate, and patient-specific parameters (particularly thepresence/absence of claudication versusinflammatory conditions).127,128 Of these,it is likely patient pathology that playsthe largest role. In general, patients withclaudication do not demonstrate signifi-cant venous enhancement until quite late.Patients with rest pain or inflammatoryconditions, such as ulcers, cellulitis, orosteomyelitis, tend to have early venousenhancement, likely secondary to arteri-ovenous shunting and increased perfu-sion. The trick then is to tailor theexamination to the patient. We havebegun using a 2-mL timing bolus in thecalf to look for the presence of venousenhancement, and if present, to deter-mine the length of the arteriovenouswindow.

We suggest the following protocol. Ifthe patient has claudication or a timingstudy that indicates late venousenhancement, perform a moving-tablestudy with coronal upper and middlestations, and either a coronal or sagittal(two slabs—one for each calf/foot—allows for easier coverage and easilyincludes the pedal vessels129) lower sta-tion study. Every attempt should bemade to image the upper and middlestations quickly (short TR, SENSE,rapid table movement, etc.) so thatacquisition of the lower station beginsas soon as possible, decreasing chancesof venous enhancement.47 Lower sta-tion acquisition can, and should, be rel-atively lengthy, with spatial resolutionat or near 1 mm isotropic and scantimes > 1 minute very acceptable.47,130

Phase-encoding orders should bereverse linear, linear, and elliptical cen-tric, respectively. This ensures ade-quate filling of the distal upper andmiddle stations (external iliac and prox-imal femoral arteries, popliteal arter-ies), and suppresses lower-stationvenous enhancement. We typically use40 mL of Gd with a biphasic bolus andflouroscopic triggering at the upper sta-tion (although a timing bolus will workequally well). Depending on acquisi-tion time and table movement delay, aninitial gadolinium infusion rate of 1.2 to


1.8 mL/sec (20 mL) followed by 0.8 to1.2 mL/sec (remaining 20 mL) workswell. When determining bolus injectionrates, consider that arterial flow downthe diseased leg averages approxi-mately 6 cm/sec, such that travel timefrom aorta to foot is of the order of 15to 20 sec.47,131 Also consider that the

bolus broadens significantly, and likethe carotid, recirculation of contrastappears to be quite adequate for sup-porting high-resolution, long acquisi-tions in the lower station.

If, on the other hand, the patient hasdistal rest pain or inflammatory disease,or demonstrates rapid venous enhance-

ment on a lower extremity test bolus, wesuggest two separate injections. The firstinjection should target only the lowerstation, with parameters similar to thoseof a moving-table examination, timingfrom the test bolus (flouroscopic trigger-ing difficult in the lower leg), and use of 15 mL of Gd at a rate of 0.8 to

Table 1. MRI Vascular Contrast Agent Classification*

Paramagnetic Superparamagnetic

No protein Weak protein Strong protein interaction interaction interaction Macromolecular USPIO

Gadopentetate dimeglumine Gadobenate dimeglumine MS-325 P792 SH U 555A (Resovist®)(Gd-DTPA; Magnevist®) (Gd-BOPTA; MultiHance®) AMI 227

Gadoteridol B-22956 Gadomer-17 NC100150(Gd-HP-DO3A; ProHance®)

Gadodiamide (Gd-DTPA-BMA; Omniscan®)

Gadoversetamide (Gd-DTPA-BMEA; Optimark®)

Gadoterate meglumine (Gd-DOTA; Dotarem®)

Gadobutrol (Gd-BT-DO3A; Gadovist®)

*Classification scheme for “vascular” magnetic resonance imaging (MRI) contrast agents. The paramagnetic gadolinium chelates can be classified according to their degree of proteininteraction. The ultra-small iron oxide particles (USPIO) are “blood pool agents” that demonstrate long intravascular enhancement. Table based on data from Knopp M, von Tengg-Kobligk H, Floemer F, Schoenberg S. Contrast agents for MRA: Future directions. J Magn Reson Imaging. 1999;10:314-316.136

Manufacturer information: MS-325, EPIX Medical, Cambridge, MA; Resovist, Schering AG, Berlin, Germany; Magnevist, Schering AG/Berlex Imaging, Wayne, NJ; Multihance, BraccoImaging, Milano, Italy/Bracco Diagnostics, Princeton, NJ; Prohance, Bracco Imaging /Bracco Diagnostics; NC100150, Amersham Health, Oslo, Norway; Omniscan, Amersham Health,Oslo, Norway; Optimark, Mallinckrodt, St. Louis, MO; Dotarem, Guerbet, Aulnay-Sous-Bois, France; Schering AG, Berlin, Germany.

A B

Figure 19. (A) Early and (B) delayed subvolume maximum intensity projections from a CE-MRV study where dilute (1:12.5) gadolinium was infused bilaterallyinto the antecubital veins. Note the complete occlusion of the left innominate vein, with high-grade stenosis at the confluence of the right innominate vein andSVC. The left subclavian vein signal void is secondary to slab positioning.


1.0 mL/sec. A second injection of 25 mL(1.2 to 1.8 mL/sec) can then be per-formed to evaluate the upper and middlestations as a two-station moving-tableexamination. Regardless of the tech-nique used, subtraction is extremelyimportant, particularly in the lowerextremities, and a precontrast maskshould always be obtained.40 For thetwo-injection technique, subtractionbecomes an absolute necessity follow-ing the second injection.

MR VENOGRAMSGadolinium-enhanced MR venogramscan be performed in a manner similar to

a conventional contrast venogram.132,133

By injecting a dilute Gd solutiondirectly into the peripheral veins (armor leg) during 3D image acquisition,structures such as the subclavian veins,superior vena cava (SVC), and IVC canbe evaluated (Figure 19). This providesmuch greater venous enhancement thanthe venous phase of an arterial study,since the Gd is delivered to the veins ina much higher concentration (less dilu-tion of the bolus and no extravasationinto extravascular structures). The arte-rial enhancement is also less than onthe venous phase of a typical contrastMRA study, although such delayedphases can be effective as well.134

We find that diluting one bottle (20 mL) of Gd into a 250 mL bag ofsaline (1:12.5) works well wheninjected into a peripheral vein at 2 to 3mL/sec. Multiple veins can be injectedsimultaneously, as in Figure 19. Imag-ing delay is not terribly important, asthe dilute Gd solution can be deliveredin large volumes. Volume coverageand slice thickness must be tailored to the individual case, and multiplephases should be obtained.

FUTURE DIRECTIONS—CONTRAST AGENTS

First-Pass Gadolinium AgentsCurrently, the most widely used contrastagents in the United States, Europe, andelsewhere for first-pass CE-MRA arethe “conventional” gadolinium chelates(Table 1). These are available as 0.5Molar formulations and possess T1relaxivities (1.5 T) between 4.3 and 5.0sec-1 • mmol-1.135 The similar concentra-tions and relaxation properties of theseagents generally translate into similarvascular imaging performance wheninjected at equal doses.

Recently, a number of gadoliniumchelates have been developed that possess properties advantageous forfirst-pass CE-MRA.136 These includeconventional Gd agents formulated at1.0 Molar rather than 0.5 Molar (ie,gadobutrol), and agents with higher invivo relaxivity due to a capacity forweak protein interaction (ie, gadobenatedimeglumine [Gd-BOPTA]). In thecase of gadobutrol, studies have shownthat the higher concentration of Gd

Figure 20. The full-thickness maximum intensityprojection (0.1 mmol/kg Gd-BOPTA) showsocclusion of the left subclavian artery (arrow)with collateralization from the left vertebralartery. The patient presented with typical clinicalsymptoms, and interventional therapy was per-formed. Due to the clear depiction of the size ofthe narrowed lumen, an interventional procedurecould be accurately planned. (Images courtesy ofGünther Schneider, Department of DiagnosticRadiology, University Hospital, Homburg/Saar,Germany.)

A B

Figure 21. A 64-year-old patient after left-side nephrectomy with newly developed arterial hypertension.(A) The full-thickness maximum intensity projection reveals a proximal renal artery stenosis with slightpoststenotic dilatation of the vessel. (B) The corresponding DSA examination performed during interven-tional therapy confirms the stenosis. Note the excellent depiction of smaller vessels on the MRA imageand the good correlation with DSA. (Images courtesy of Günther Schneider, Department of DiagnosticRadiology, University Hospital, Homburg/Saar, Germany.)


combined with a more rapid bolusenables greater intravascular signal thanis achievable with equivalent doses oftraditional agents.137-139 This may proveuseful for the improved visualization ofsmaller arteries in vascular regions inwhich the SNR is limited, such as in the

calf and distal renal arteries. Improvedvisualization of smaller pelvic vesselshas already been reported.139

Gadobenate dimeglumine alsodemonstrates increased intravascularsignal as compared with conventionalgadolinium chelates (Table 1).139-141 This

agent differs from the other availableGd agents in that it possesses a higherin vivo T1 relaxivity (9.7 sec-1 • mmol-1)due to weak and transient interactionsbetween the Gd-BOPTA chelate andserum proteins, particularly albu-min.135,142 Clinical benefits of theincreased relaxivity have been demon-strated for all vascular territories,including the peripheral vasculature.143-

145 Specifically, there is better visualiza-tion of lesions in larger vessels (Figure20), as well as better depiction ofsmaller vessels themselves (Figure 21).Like the conventional nonprotein-inter-acting gadolinium chelates, gadobenatedimeglumine demonstrates an excel-lent safety profile.146

“Blood Pool” Gadolinium AgentsBecause of the rapid redistribution ofthe first-pass gadolinium chelates intothe extracellular compartment, variousattempts have been made to formulateintravascular blood-pool contrastagents for 3D CE-MRA. Currently,two main types of intravascular agentsbased on Gd are under development:Agents that differ from the purelyextracellular and weakly protein-inter-acting agents in having strong affinityfor serum albumin (Table 1), andmacromolecular agents whose sizeprecludes their rapid extravasation(Table 1). Although no agents are yetapproved for clinical use, two exam-ples of agents with strong affinity forserum proteins are currently undergo-ing clinical trials: MS-325147-149 (Figure22) and B22956150,151 (Figure 23). Bothof these agents appear very promising,with excellent safety profiles and thecapacity to perform conventional high-quality first-pass dynamic imaging inaddition to delayed steady-state vascu-lar imaging. This may be particularlyuseful for coronary studies.150,152,153

Examples of Gd-based blood poolagents with macromolecular structuresare gadomer-17154 and P792.155 LikeMS-325 and B22956, preliminaryresults suggest these agents may provebeneficial for coronary MRA.156,157

A B

Figure 22. Contrast-enhanced MRA (CE-MRA) of the foot using the blood-pool agent MS-325 in a patientwith ischemic rest pain. (A) Arterial and (B) equilibrium sagittal full-volume maximum intensity projec-tions, MS-325 dose 0.05 mmol/kg. The arterial study is subtracted, 46 sec in duration, with true voxelsize of 0.9 × 0.9 × 1.8 mm3. The equilibrium study is fat suppressed at a delay of 5 minutes, 4:57 in dura-tion, with a true voxel size 0.9 × 0.9 × 0.8 mm3.

A B

Figure 23. B22956 in imaging of healthy right coronary artery. (A) Unenhanced (T2Prep acquisition) and(B) B22956-enhanced (0.075 mmol/kg) image acquired at 25-min postcontrast using 3D Navigator gatedand corrected inversion recovery-fast field echo (IR-FFE) sequence. (Images courtesy of Eckart Fleck andIngo Paetsch, German Heart Institute, Berlin, Germany; Matthias Stuber, Beth Israel Deaconess MedicalCenter, Cardiovascular Division, Boston, MA; and Friedrich Cavagna, Bracco Imaging SpA, Milan, Italy.)


Superparamagnetic Iron Oxide AgentsThe second major category of potentialcontrast agents for MRA consists of thesuperparamagnetic group, which isbased on ultra-small particles of ironoxide (USPIO) (Table 1). Currently, SH U 555A is the only clinicallyapproved USPIO agent available forCE-MRA, although few studies havebeen conducted in vascular territoriesother than the upper abdomen.158 AMI227 is another potentially useful repre-sentative of this group, although, again,few studies have been conducted.159

Perhaps the most promising of theUSPIO agents currently under devel-opment is the blood-pool agentNC100150 (Figure 24).160,161 As withthe Gd blood-pool agents, the potentialadvantages of this agent include a long

intravascular half-life with minimalleakage into the interstitial space,thereby permitting the steady-statevascular imaging. Preliminary studieson the use of this agent for coronaryMRA have already been reported,161

and work is ongoing to establish itspotential for other applications such asrenal perfusion.162

As it stands today, the imaging qual-ity achievable with first-pass agentshas, to some extent, eliminated the ini-tial target role for intravascular agents.It has become apparent that real bene-fits will most likely only be clinicallyachievable if the intravascular agentscan be used both for first-pass andsteady-state imaging. Nevertheless, itis doubtful that this type of blood-poolagent will ever have any advantageover the currently available Gd agentsfor dynamic, arterial-phase MRA.Conceivably, this type of agent mayhave benefits for ultra-high–resolutionimaging, for assessment of vasculatureduring intervention, and for real-timetherapy monitoring. Other opportuni-ties that may present themselvesinclude perfusion applications com-bined with first-pass imaging.

While MR angiography usingblood-pool agents may ultimatelyprove beneficial, all current clinicalwork utilizes extracellular gadoliniumchelates.

FUTURE DIRECTIONS—HARDWAREAmong the most important recentdevelopments in MR hardware has beenthe introduction of whole body 3.0-Tscanners. The principal advantage ofincreasing the magnetic field strength isimproved SNR, resulting in better reso-lution and conspicuity of vessels usingsimilar acquisition times.163 Althoughclinical experience with CE-MRA at3.0 T is somewhat limited at present,early studies have indicated markedlyimproved performance of unenhancedTOF techniques, particularly in imagingof the intracranial and cervical vascula-ture.163,164 Similarly, imaging at 3.0 T has

been reported to improve the resolutionand delineation of small venous vesselsin the brain,165 and appears advanta-geous for contrast-enhanced vascularand coronary imaging.166-168 It remains tobe seen just how the availability of 3.0 Tmachines will impact CE-MRA. Sev-eral challenges remain to be overcome,including developing new coils, over-coming specific absorption rate(radiofrequency power deposition) lim-itations, and optimizing protocols. Thisaside, however, it appears likely theSNR-derived improvement in spatialand temporal resolution at 3.0 T willlead to significant improvements in CE-MRA image quality.

CONCLUSIONContrast-enhanced MRA provides asafe, fast, and cost-efficient alternativeto conventional diagnostic angiogra-phy throughout the vasculature.Numerous recent advances allow forhigh-resolution breath-hold and multi-station imaging with optimal arterialphase contrast bolus timing. As thetechnique is further refined, bettercoils, better contrast agents, and per-haps higher magnetic field strengthswill allow for greater spatial and tem-poral resolution, further increasing theeffectiveness and utility of CE-MRA.

REFERENCES1. Saloner D. MRA: Principles and display. In:Higgins C, Hricak H, Helms C, eds. Magnetic Res-onance Imaging of the Body. Philadelphia: Lip-pencott-Raven; 1997:1345-1368.2. Gullberg G, Wherli F, Shimakawa A, SimmonsM. MR vascular imaging with a fast gradient refo-cusing pulse sequence and reformatted imagesfrom transaxial sections. Radiology. 1997;165:241-246.3. Doumolin C, Cline H, Souza S, et al. Three-dimensional time-of-flight magnetic resonanceangiography using spin saturation. Magn ResonImaging. 1989;11:35-46.4. Owen R, Baum R, Carpenter J, et al. Sympto-matic peripheral vascular disease: Selection ofimaging parameters and clinical evaluation withMR angiography. Radiology. 1993;187:627-635.5. Cortell E, Kaufman J, Geller S, et al. MR angiog-raphy of the tibial runoff vessels: Imaging with thehead coil compared to conventional arteriography.AJR Am J Roentgenol. 1996;167:147-151.

Figure 24. NC100150 (5.0 mg Fe/kg bodyweight,slow hand injection, third injection of an accumu-lating dose scheme) reveals occlusion of the leftrenal artery and a 60% to 70% stenosis of theright renal artery.


6. Glickerman D, Obregon G, Schmiedl U, et al.Cardiac-gated MR angiography of the entire lowerextremity: A prospective comparison with con-ventional angiography. AJR Am J Roentgenol.1996;167:445-451.7. Patel P, Kuntz K, Klufas R, et al. Carotid bifur-cation. Can magnetic resonance angiography andduplex ultrasonography replace contrast arteriog-raphy? Stroke. 1995;26:1753-1758. 8. Anderson C. MRA of the aortic arch andextracranial carotid arteries. In: Higgins C, HricakH, Helms C, eds. Magnetic Resonance Imaging ofthe Body. Philadelphia, Pa: Lippencott-Raven;1997:1369-1381.9. Blatter D, Parker D, Robinson R. Cerebral MRangiography with multiple overlapping thin slabacquisition. Radiology. 1991;179:805-811.10. Kim D, Edelman R, Kent K, et al. Abdominalaorta and renal artery stenosis: Evaluation with MR angiography. Radiology. 1990;174:727-731.11. Yucel E, Kaufman J, Prince M, et al. Time-of-flight renal MR arteriography: Utility in patientswith renal insufficeincy. Magn Reson Imaging.1993;11:925-930.12. Loubeyre P, Cahan R, Grozel F, et. al. Trans-plant renal artery stenosis. Evaluation of diagnosiswith magnetic resonance angiography comparedwith color duplex sonography and arteriography.Transplantation 1996; 62:446-450.13. Prince M. Body MR angiography with gado-linium contrast agents. MRI Clin N Am. 1996;4:11-24. 14. Prince M. Gadolinium-enhanced MR aortogra-phy. Radiology. 1994; 191:155-164.15. Bryant D, Payne J, Firmin D, Longmore D.Measurement of flow with NMR imaging using agradient pulse and phase difference technique.Comput Assist Tomogr. 1984;8:588-593.16. Gedroyc W. Magnetic resonance angiographyof renal arteries. Urol Clin North Am. 1994;21:201-214.17. Bass J, Prince M, Londy F, Chenevert T. Effectof gadolinium on phase-contrast MR angiographyof the renal arteries. AJR Am J Roentgenol. 1997;168:261-266.18. Gedroyc W, Negus R, Al-Kutoubi A, et al.Magnetic resonance angiography of renal trans-plants. Lancet. 1992;339:789-791.19. Yucel E. Magnetic resonance angiography ofthe lower extremity and renal arteries. Sem Ultra-sound CT MRI. 1992;13:291-302.20. Ngheim H, Winter R, Mountford M, et al.Evaluation of the portal venous system before livertransplantation: value of phase-contrast MRangiography. AJR Am J Roentgenol. 1995;164:871-878.21. Lundin B, Cooper T, Meyer R, Potchen E.Measurement of total and unilateral renal bloodflow by oblique-angle velocity encoded 2D-CINEmagnetic resonance angiography. Magn ResonMed. 1992;11:51-59.22. Schoenberg S, Knopp M, Londy F, et al. Renalartery stenosis: Stenosis grading by combinedmorphologic and functional MR imaging: Results

of a multicenter multireader analysis. J Am SocNephrol. 2002;13:158-169.23. Weinman H, Brasch R, Press W, Wesbey G.Charcteristics of gadolinium-DTAP complex: Apotential NMR contrast agent. AJR Am JRoentgenol. 1984;142:619-624.24. Schima W, Mukerjee A, Saini S. Contrast-enhanced MR imaging. Clin Radiol. 1996;51:235-244.25. Goldstein H, Kashanian F, Blumetti R, et al.Safety assessment of gadopentetate dimegluminein U.S. clinical trials. Radiology. 1990;174:17-23.26. Niendorf H, Haustein J, Alhassan A, ClaussW. Safety of gadolinium-DTPA: Extended clini-cal experience. Presented as part of Brasch R,mod. Workshop on Contrast-Enhanced MagneticResonance. Society for Magnetic Resonance inMedicine Meeting, Napa, CA, 1991:70-79. 27. Haustein J, Niendorf H, Krestin G, et al. Renaltolerance of gadolinium-DTAP/dimeglumine inpatients with chronic renal failure. Invest Radiol.1992; 27:153-156.28. Prince M, Arnoldus C, Frisoli J. Nephrotoxic-ity of high dose gadolinium compared to iodinatedcontrast. J Magn Reson Imaging. 1996;6:162-166.29. Hohenschuh E, Watson A. Theory and mecha-nisms of contrast-enhancing agents. In: Higgins C,Hricak H, Helms C, eds. Magnetic ResonanceImaging of the Body. Philadelphia, Pa: Lippencott-Raven; 1997:1439-1464.30. Prince M, Yucel E, Kaufman J, et al. Dynamicgadolinium-enhanced three-dimensional abdomi-nal MR arterography. JMRI. 1993;3:877-881. 31. Runge V, Kirsch J, Lee C. Contrast-enhanceMR angiography. JMRI. 1993;3:233-239. 32. Ehman R, Revel D, Sievers R, Brasch R. Acutemyocardial ischemia. Magnetic resonance contrastenhancement with gadolinium-DTPA. Radiology.1984;153:157-163.33. Moseley M, Sawyer A. Imgaing Techniques:Pulse sequences. In: Higgins C, Hricak H, Helms C, eds. Magnetic Resonance Imaging of theBody. Philadelphia, Pa: Lippencott-Raven;1997:43-69.34. Wehrli F. Principles of magnetic resonance. In:Stark D, Bradley W, eds. Magnetic ResonanceImaging. St. Louis, MO: Mosby-Year Book;1992:3-20.35. Prince M, Narasimham D, Stanley J, et al.Breath-hold gadolinium-enhanced MR angiogra-phy of the abdominal aorta and its major branches.Radiology. 1995;197:785-792.36. Snidow J, Johnson S, Harris V, et al. Three-dimensional gadolinium-enhanced MR angiogra-phy for aortoiliac inflow assessment plus renalartery screening in a single breath hold. Radiology.1996;198:725-732.37. Leung D, McKinnon G, Davis C, et al. Breath-hold, contrast-enhanced, three-dimensional MRangiography. Radiology. 1996;201:569-571.38. Hendrick R, Roff U. Image contrast and noise.Chicago, Il: Mosby Yearbook; 1991.39. Weinman H, Laniado M, Mutzel W. Pharma-cokinetics of Gd-DTPA/dimeglumine after intra-venous injection into healthy volunteers. PhysiolChem Phys Med NMR. 1984;16:167-172.

40. Lee V, Flyer M, Weinreb J, et al. Image sub-traction in gadolinium-enhanced MR imaging.AJR Am J Roentgenol. 1996;167:1427-1432.41. Maki J, Chenevert T, Prince M. Optimizingthree-dimensional gadolinium-enhanced MRangiography. Invest Radiol. 1998;33:528-537.42. Verhoeven L. Digital subtraction angiography:The technique and an analysis of the physical fac-tors influencing the image quality [thesis]. In:Delft: Technische Hogeschule; 1985. 43. Barnes G, Lakshminarayann A. Conventionaland spiral Computed Tomography. In: Lee J,Sagel S, Stanley R, Heiken J, eds. Computed BodyTomgraphy with MRI Corellation. Philadelphia,Pa: Lippencott-Raven; 1998:1-20.44. Maki J, Prince M, Londy F, Chenevert T. Theeffects of time varying intravascular signal inten-sity on three-dimensional MR angiography imagequality. J Magn Reson Imaging. 1996;6:642-651.45. Riederer S, Tasciyan T, Farzaneh F. MRflouroscopy: Technical feasibility. Magn ResonMed. 1988;8:1-15.46. Fain S, Riederer S, Bernstein M, Huston J.Theoretical limits of spatial resolution in elliptical-centric contrast-enhanced 3d-MRA. Magn ResonMed. 1999;42:1106-1116.47. Maki J, Wilson G, Eubank W, Hoogeveen R.Utilizing SENSE to achieve lower station sub-mil-limeter isotropic resolution and minimal venousenhancement in peripheral MR angiography. JMagn Reson Imaging. 2002;15:484-491.48. Bampton A, Riederer S, Korin H. Centricphase-encoding order in three-dimensional MP-RAGE sequences: Application to abdominalimaging. J Magn Reson Imaging. 1992;2:327-334.49. Wilman A, Riederer S. Improved centric phaseencoding orders for three-dimensional magnetiza-tion-prepared MR angiography. Magn Reson Med.1996;36:384-392.50. Maki J, Chenevert T, Prince M. The effects ofincomplete breath holding on 3D MR image qual-ity. J Magn Reson Imaging. 1997;7:1132-1139.51. Wilman A, Riederer S, Breen J, Ehman R.Elliptical spiral phase-encoding order: An optimal,field-of-view-dependent ordering scheme forbreath-hold contrast-enhanced 3D MR angiogra-phy. Radiology. 1996;201:328-329.52. Hany T, Schmidt M, Hilfiker P, et al. Opti-mization of contrast dosage for gadolinium-enhanced 3D MRA of the pulmonary and renalarteries. Magn Reson Imaging. 1998;16:901-906.53. Shetty A, Bis K, Kirsch M, et al. Contrast-enhanced breath-hold three-dimensional magneticresonance angiography in the evaluation of renalarteries: Optimization of technique and pitfalls. JMagn Reson Imaging. 2000;12:912-923.54. Snidow J, Aisen A, Harris V, et al. Iliac arteryMR angiography: Comparison of three-dimen-sional gadolinium-enhanced and two-dimensionaltime-of-flight techniques. Radiology. 1995;196:371-378.55. Steiner P, Debatin J, Romanowski B, et al. Optimization of breath-hold 3D MR pulmonaryangiography [abstract]. Radiology. 1996;201(P):201.


56. Hany T, Debatin J, Schmidt M, et al. Three-dimensional MR angiography of the renal arter-ies in a single breath-hold. Radiology. 1996;201:218.57. Finn J, Baskaran V, Carr J, et al. Thorax: Low-dose contrast-enhanced three-dimensional MRangiography with subsecond temporal resolutioninitial results. Radiology. 2002;224:896-904.58. Prince M, Grist T, Debatin J. 3D contrast MRangiography. Berlin: Springer-Verlag; 1997.59. Earls J, Rofsky N, DeCorato D, et al. Breath-hold single-dose gadolinium-enhanced three-dimensional MR aortography: Usefulness of atiming examination and MR power injector. Radi-ology. 1996:705-710.60. Krinsky G, Rofsky N, Flyer M, et al. Gadolin-ium-enhanced three-dimensional MR angiographyof acquired arch vessel disease. AJR Am JRoentgenol. 1996;167:981-987.61. Kreitner K, Kunz R, Kalden P, et al. Contrast-enhanced three-dimensional MR angiography ofthe thoracic aorta: Experiences after 118 examina-tions with a standard dose contrast administrationand different injection protocols. Eur Radiol.2001;11:1355-1363.62. Wilman A, Riederer S, King B, et al. Flouro-scopically-triggered contrast-enhanced threedimensional MR angiography with elliptical cen-tric view order: Application to the renal arteries.Radiology. 1997;205:137-146.63. Huston J, Fain S, Riederer S, Wilman A, Bern-stein M, Busse R. Carotid arteries: Maximizingarterial to venous contrast in fluoroscopically trig-gered contrast-enhanced MR angiography withelliptic centric view ordering. Radiology.1999;211:265-273.64. Fellner F, Fellner C, Wutke R, et al. Fluoro-scopically triggered contrast-enhanced 3D MRDSA and 3D time-of-flight turbo MRA of thecarotid arteries: First clinical experiences in corre-lation with ultrasound, x-ray angiography, andendarterectomy findings. Magn Reson Imaging.2000;18:575-585.65. Watts R, Wang Y, Redd B, et al. Recessedelliptical-centric view-ordering for contrast-enhanced 3D MR angiography of the carotid arter-ies. Magn Reson Med. 2002;48:419-424. 66. Levy R, Maki J. Three-dimensional contrast-enhanced MR angiography of the extracranialcarotid arteries: Two techniques. AJNR Am J Neu-roradiol. 1998;19:688-690. 67. Korosec F, Grist T, Frayne R, Mistretta C.Time-resolved contrast-enhanced 3D MR angiog-raphy. Magn Reson Med. 1996;36:345-351.68. Mistretta C, Grist T, Korosec F, Frayne R. 3Dtime-resolved contrast-enhanced MR DSA:Advantages and tradeoffs. Magn Reson Med.1998;40:571-581.69. Schoenberg S, Bock M, Floemer F, et al. High-resolution pulmonary arterio- and venographyusing multiple-bolus multiphase 3D-Gd-MRA. JMagn Reson Imaging. 1999;10:339-346.70. Schoenberg S, Essig M, Hallscheidt P, et al.Multiphase magnetic resonance angiography ofthe abdominal and pelvic arteries: Results of a

bicenter multireader analysis. Invest Radiol.2002;37:20-28.71. Stein B, DeMarco J, Zhou Y. 3D Time-resolved MRA with elliptic centric view ordering.Initial clinical experience to evaluate tibo-peronealarteries. Presented at the XIIIth Annual Inerna-tional Workshop on MRA. Madison, WI, 2001.72. Hennig J, Scheffler K, Laubenberger J,Strecker R. Time-resolved projection angiographyafter bolus injection of contrast agent. Magn ResonMed. 1997;37:341-345.73. Goyen M, Laub G, Ladd M, et al. Dynamic 3DMR angiography of the pulmonary arteries inunder four seconds. J Magn Reson Imaging.2001;13:372-377.74. Finn J, Carr J, Pereles S, et al. Time-resolved,contrast-enhanced MR angiography of the thorax.Presented at the XIIIth Annual InternationalWorkshop on MRA. Madison, WI, 2001. 75. Pruessmann K, Weiger M, Scheidegger M,Boesiger P. SENSE: Sensitivity encoding for fastMRI. Magn Reson Med. 1999;42:952-962.76. Finn J, Larson A, Moore J, Simonetti O. Sub-second 3D contrast-enhanced MRA of the thoraxwith radial k-space sampling. Presented at the XthScientific Meeting and Exhibition of the Interna-tional Society for Magnetic Resonance in Medi-cine. Honolulu, HI, 2002.77. Bock M, Schoenberg S, Floemer F, Schad L.Separation of arteries and veins in 3D MR angiog-raphy using correlation analysis. Magn ResonMed. 2000;43:481-487.78. Mistretta C, Grist T, Frayne R, Korosec F.Contrast and motion artifacts in 4D MR angiogra-phy. Radiology. 1996;201:238.79. Turski P, Korosec F, Carroll T, et al. Contrast-Enhanced magnetic resonance angiography of thecarotid bifurcation using the time-resolved imag-ing of contrast kinetics (TRICKS) technique. TopMagn Reson Imaging. 2001;12:175-181.80. Mazaheri Y, Carroll T, Du J, et al. Combinedtime-resolved and high-spatial-resolution 3DMRA using an extended adaptive acquisition. JMagn Reson Imaging. 2002;15:291-301.81. Peters D, Korosec F, Grist T, et al. Under-sampled projection reconstruction applied to MRangiography. Magn Reson Med. 2000;43:91-101.82. Barger A, Block W, Toropov Y, et al. Time-resolved contrast-enhanced imaging with isotropicresolution and broad coverage using an undersam-pled 3D projection trajectory. Magn Reson Med.2002;48:297-305.83. Willinek W, Gieseke J, Von Falkenhousen V,et al. CD-3D MRA of the supraaoritc arteries at512 and 1024 matrix: The use of randomly seg-mented central k-space ordering (CENTRA). Pre-sented at the Xth Scientific Meeting and Exhibitonof the International Society for Magnetic Reso-nance in Medicine. Honolulu, HI, 2002.84. Wood M, Runge V, Henkelman R. Overcom-ing motion in abdominal MR imaging. AJR Am JRoentgenol. 1988;150:513-522.85. Ehman R, Felmlee J. Adaptive technique forhigh-definition MR imaging of moving structure.Radiology. 1989;173:255-263.

86. Vasbinder G, Maki J, Nijenhuis R, et al.Motion of the distal renal artery during 3D con-trast-enhanced breath-hold MRA. J MagneticReson Imaging. 2002;16:685-696.87. Holland G, Dougherty L, Carpenter J, et al.Breath-hold ultrafast three-dimensional gadolin-ium-enhanced MR angiography of the aorta andthe renal and other visceral abdominal arteries.AJR Am J Roentgenol. 1996;166:971-981.88. Carr J, Ma J, Desphande V, et al. High-resolu-tion breath-hold contrast-enhanced MR angiogra-phy of the entire carotid circulation. AJR Am JRoentgenol. 2002;178:543-549.89. Weiger M, Pruessmann K, Kassner A, et al.Contrast-enhanced 3D MRA using SENSE. JMagn Reson Imaging. 2000;12:671-677.90. Sodickson D, Manning W. Simultaneousacquisition of spatial harmonics (SMASH): Fastimaging with radiofrequency coil arrays. MagnReson Med. 1997;38:591-603.91. Wilson G, Maki J, Evitts M, et al. A directcomparison study to evaluate the usefulness ofSensitivity Encoding (SENSE) in evaluating renalartery stenosis using MR angiography. Presentedat the XIII Annual International Workshop onMRA. Madison, WI. 2001. 92. Maki J, Wilson G, Hoogeveen R. 3D Gdenhanced moving table MR angiography of theaorta and outflow vessels using SENSE to achievehigh resolution of the below knee vasculature. Pre-sented at the Seventh scientific meeting of theInternational Society for Magnetic Resonance inMedicine, Glasgow, Scotland, 2001. 93. Du Y, Parker D, Davis W, Gao T. Reduction ofpartial-volume artifacts with zero-filled interpola-tion in three-dimensional MR angiography. JMagn Reson. 1994;4:733-741.94. Ros P, Gauger J, Stoupis C, et al. Diagnosis ofrenal artery stenosis: Feasibility of combining MRangiography, MR renography, and gadopentetate-based measurements of glomerular filtration rate.AJR Am J Roentgenol. 1995;165:1447-1451.95. Schoenberg S, Bock M, Aumann S, et al.Quantitative recording of renal function with mag-netic resonance tomography. Radiologe. 2000;40:925-937.96. De Marco J, Nesbit G, Wesbey G, RichardsonD. Prospective evaluation of extracranial carotidstenosis: MR angiography with maximum-inten-sity projections and multiplanar reformation com-pared with conventional angiography. AJR Am JRoentgenol. 1994;163:1205-1212.97. Laub G. Displays for MR angiography. MagnReson Med. 1990;14:222-229.98. Rossnick S, Laub G, Braeckle R. Three dimen-sional display of blood vessels in MRI. In: Pro-ceedings of the IEEE: Computers in Cardiology.New York: IEEE; 1986:193-195. 99. Anderson C, Saloner D, Tsuruda J, et al. Arti-facts in maximum intensity projection display ofMR angiograms. AJR Am J Roentgenol. 1990;154:623-629.100. Baskaran V, Pereles F, Nemcek AJ, et al.Gadolinium-enhanced 3D MR angiography ofrenal artery stenosis: A pilot comparison of maxi-


mum intensity projection, multiplanar reformat-ting, and 3D volume-rendering postprocessingalgorithms. Acad Radiol. 2002;9:50-59.101. Douek P, Revel D, Chazel S, et al. Fast MRangiography of the aortoiliac arteries and arteriesof the lower extremity: Value of bolus-enhanced,whole-volume subraction technique. AJR Am JRoentgenol. 1995;165:431-437.102. Wang Y, Johnston DL, Breen JF, et al.Dynamic MR digital subtraction angiographyusing contrast enhancement, fast data acquisition,and complex subtraction. Magn Reson Med.1996;36:551-556. 103. Wang Y, Winchester P, Khilnani N, et al.Contrast-enhanced peripheral MR angiographyfrom the abdominal aorta to the pedal arteries:Combined dynamic two-dimensional and bolus-chase three-dimensional acquisitions. InvestRadiol. 2001;36:170-177.104. Hoogeveen R, Bakker C, Viergever M. Lim-its to the accuracy of vessel diameter measurementin MR angiography. J Magn Reson Imaging.1998;8:1228-1235.105. Wilson G, Haynor D, Maki J. Resolutionrequirements for grading stenoses in 3D CE-MRA. Presented at the Sixth scientific meeting ofthe International Society for Magnetic Resonancein Medicine. Denver, CO, 2000. 106. Stafford-Johnson D, Hamilton B, Dong Q, etal. Vascular Complications of Liver Transplanta-tion: Evlauation with gadolinium-enhanced MRangiography. Radiology. 1998;207:153-160.107. Wutke R, Lang W, Fellner C, et al. High-res-olution, contrast-enhanced magnetic resonanceangiography with elliptical centric k-space ordering of supra-aortic arteries compared withselective X-ray angiography. Stroke. 2002;33:1522-1529.108. Murray J, Manisali M, Flamm S, et al. Intra-mural hematoma of the thoracic aorta: MR imagefindings and their prognostic implications. Radiol-ogy. 1997;204:349-355.109. Prince M, Schoenberg S, Ward J, et al.Hemodynamically significant atherosclerotic renalartery stenosis: MR angiographic features. Radiol-ogy. 1997;205:128-136.110. Hood M, Ho V, Corse W. Three-dimensionalphase-contrast magnetic resonance angiography:A useful clinical adjunct to gadolinium-enhancedthree-dimensional renal magnetic resonanceangiography? Mil Med. 2002; 167:343-349.111. Johnson D, Lerner C, Prince M, et al.Gadolinium-enhanced magnetic resonanceangiography of renal transplants. Magn ResonImaging. 1997;15:13-20.112. Ferreiros J, Mendez R, Jorquera M, et al.Using gadolinium-enhanced three-dimensionalMR angiography to assess arterial inflow stenosisafter kidney transplantation. AJR Am JRoentgenol. 1999;172:751-757.113. Laissy J, Trillaud H, Douek P. MR angiogra-phy: noninvasive vascular imaging of theabdomen. Abdom Imaging. 2002;27:488-506.114. Baden J, Racy D, Grist T. Contrast-enhancedthree-dimensional magnetic resonance angiogra-

phy of the mesenteric vasculature. J Magn ResonImaging. 1999;10:369-375.115. Stafford-Johnson D, Chenevert T, Cho K,Prince M. Portal venous magnetic resonanceangiography. A review. Acad Radiol. 1998;5:289-305.116. Rofsky N, Lee V, Laub G, et al. AbdominalMR imaging with a volumetric interpolatedbreath-hold examination. Radiology. 1999;212:876-884.117. Oudkerk M, van Beek E, Wielopolski P, et al.Comparison of contrast-enhanced magnetic reso-nance angiography and conventional pulmonaryangiography for the diagnosis of pulmonaryembolism: a prospective study. Lancet. 2002;359:1643-1647.118. Meaney J, Johansson L, Ahlstrom H, PrinceM. Pulmonary magnetic resonance angiography. JMagn Reson Imaging. 1999;10:326-338.119. Wikstrom J, Johansson L, Rossitti S, et al.High resolution carotid artery MRA. Comparisonwith fast dynamic acquistion and duplex ultra-sound scanning. Acta Radiol. 2002;43:256-261.120. North American Symptomatic CarotidEndarterectomy Trial Collaborators. Beneficialeffect of carotid endarterectomy in symptomaticpatients with high-grade carotid stenosis. N Engl JMed. 1991;325:445-453.121. Sueyoshi E, Sakamoto I, Matsuoka Y, et al.Aortoiliac and lower extremity arteries: Compari-son of three-dimensional dynamic contrast-enhanced subtraction MR angiography andconventional angiography. Radiology. 1999;210:683-688.122. Ho K, Leiner T, DeHaan M, et al. Peripheralvascular tree stenoses: evaluation with moving-bed infusion-tracking MR angiography. Radiol-ogy. 1998;206:683-692.123. Meaney J, Ridgway J, Chakraverty S, et al.Stepping-table gadolinium-enhanced digital sub-traction MR angiography of the aorta and lowerextremity arteries: Preliminary experience. Radi-ology. 1999;211:59-67.124. Ho V, Choyke P, Foo T, et al. Automatedbolus chase peripheral MR angiography: Initialpractical experiences and future directions of thiswork-in-progress. J Magn Reson Imaging.1999;10:376-388.125. Goyen M, Quick H, Debatin J, et al. Whole-body three-dimensional MR angiography with arolling table platform: Initial clinical experience.Radiology. 2002;224:270-277.126. Ho K, de Haan M, Kessels A, et al. Peripheralvascular tree stenoses: Detection with subtractedand nonsubtracted MR angiography. Radiology.1998;206:673-681.127. Ho K. Thesis: MR Angiography of the LowerExtremities. Maastricht, The Netherlands: Univer-sity of Maastricht; 1999:93-108. 128. Ho K. First experiences with BPA for periph-eral MRA. Presented at the XI International Work-shop on MRA. Lund, Sweden, 1999.129. Maki J, Wilson G, Eubank W, et al. 3D Gd-enhanced moving table peripheral MR angiogra-phy using multi-station SENSE to include the

pedal vasculature. Presented at the Tenth ScientificMeeting and Exhibition of the International Soci-ety for Magnetic Resonance in Medicine. Hon-olulu, HI, 2002.130. Leiner T, Kessels A, van Engelshoven J.Total runoff peripheral MRA in patients with criti-cal ischemia and tissue loss can detect more patentarteries than IA-DSA. In: Tenth Scientific Meetingand Exhibition: International Society for MagneticResonance in Medicine. 2002;210. 131. Wang Y, Lee H, Avakian R, et al. Timingalogrithm for bolus chase MR digital subtractionangiography. Magn Reson Med. 1998;39:691-696.132. Ruehm S, Zimny K, Debatin J. Direct con-trast-enhanced 3D MR venography. Eur Radiol.2001;11:102-112.133. Li W, Kaplan D, Edelman R. Three-dien-sional low dose gadolinium-enhanced peripheralMR venography. J Magn Reson Imaging. 1998;8:630-633.134. Thornton M, Ryan R, Varghese J, et al. Athree-dimensional gadolinium-enhanced MRvenography technique for imaging central veins.AJR Am J Roentgenol. 1999;173:999-1003.135. de Haën C, Cabrini M, Akhnana L, et al.Gadobenate dimeglumine 0.5 M solution for injec-tion (MultiHance) pharmaceutical formulation andphysicochemical properties of a new magnetic res-onance imaging contrast medium. J Comput AssistTomogr. 1999;23 Suppl 1:S161-S168.136. Knopp M, von Tengg-Kobligk H, Floemer F,Schoenberg S. Contrast agents for MRA: Futuredirections. J Magn Reson Imaging. 1999;10:314-316.137. Tombach B, Reimer P, Prumer B, et al. Doesa higher concentration of gadolinium chelatesimprove first-pass cardiac signal changes? J MagnReson Imaging. 1999;10:806-812.138. Goyen M, Lauenstein T, Herborn C, et al. 0.5M Gd chelate (Magnevist) versus 1.0 M Gdchelate (Gadovist): Dose-independent effect onimage quality of pelvic three-dimensional MR-angiography. J Magn Reson Imaging. 2001;14:602-607.139. Herborn C, Lauenstein T, Ruehm S, et al.Intraindividual comparison of gadopentetatedimeglumine, gadobenate dimeglumine andgadobutro for pelvic 3D magnetic resonanceangiography. Invest Radiol. 2003;38:27-33.140. Knopp M, Schoenberg S, Rehm C, et al.Assessment of gadobenate dimeglumine (Gd-BOPTA) for MR angiography: Phase I studies.Invest Radiol. 2002;37:706-715.141. Völk M, Strotzer M, Lenhart M, et al. Renaltime-resolved MR angiography: Quantitativecomparison of gadobenate dimeglumine andgadopentetate dimeglumine with different doses.Radiology. 2001;220:484-488.142. Cavagna F, Maggioni F, Castelli P, et al.Gadolinium chelates with weak binding to serumproteins. A new class of high-efficiency, generalpurpose contrast agents for magnetic resonanceimaging. Invest Radiol. 1997;32:780-796.143. Kroencke T, Wasser M, Pattynama P, et al.Gadobenate dimeglumine-enhanced magnetic res-


onance angiography of the abdominal aorta andrenal arteries. AJR Am J Roentgenol. 2002;179:1573-1582.144. Ruehm S, Goyen M, Barkhausen J, et al.Rapid magnetic resonance angiography for detec-tion of atherosclerosis. Lancet. 2001;357:1086-1091.145. Goyen M, Herborn C, Lauenstein T, et al.Optimization of contrast dosage for gadobenatedimeglumine-enhanced high-resolution wholebody 3D MR angiography. Invest Radiol. 2002;37:263-268.146. Kirchin M, Pirovano G, Venetianer C,Spinazzi A. Safety assessment of gadobenatedimeglumine (Multihance): Extended clinicalexperience from phase I studies to post-marketingsurveillance. J Magn Reson Imaging. 2001;14:281-294.147. Grist T, Korosec F, Peters D, et al. Steady-state and dynamic MR angiography with MS-325:Initial experience in humans. Radiology. 1998;207:539-544.148. Lauffer R, Parmelee D, Dunham S, et al. MS-325: Albumin-targeted contrast agent for MRangiography. Radiology. 1998;207:529-538.149. Bluemke D, Stillman A, Bis K, et al. CarotidMR angiography: Phase II study of safety and efficacy of MS-325. Radiology. 2001;219:114-122.150. La Noce A, Stoelben S, Scheffler K, et al.B22956/1, a new intravascular contrast agent forMRI: First administration to humans-preliminaryresults. Acad Radiol. 2002;9 Suppl:S404-406.151. Zheng J, Li D, Cavagna F, et al. Contrast-enhanced coronary MR angiography: Relationshipbetween coronary artery delineation and blood T1.J Magn Reson Imaging. 2001;14:348-354.

152. Stuber M, Botnar R, Danias P, et al. Contrastagent-enhanced, free-breathing, three-dimensionalcoronary magnetic resonance angiography. JMagn Reson Imaging. 1999;10:790-799.153. Huber M, Paetsch I, Schnackenburg B, et al.Performance of a new gadolinium-based intravas-cular contrast agent in free-breathing inversion-recovery 3D coronary MRA. Magn Reson Med.2003;49:115-121.154. Dong Q, Hurst D, Weinmann H, et al. Mag-netic resonance angiography with gadomer-17. Ananimal study original investigation. Invest Radiol.1998;33:699-708.155. Gaillard S, Kubiak C, Stolz C, et al. Safetyand pharmacokinetics of p792, a new blood-poolagent: results of clinical testing in nonpatient vol-unteers. Invest Radiol. 2002;37:161-166.156. Taupitz M, Schnorr J, Wagner S, et al. Coro-nary magnetic resonance angiography: experimen-tal evaluation of the new rapid clearance bloodpool contrast medium P792. Magn Reson Med.2001;46:932-938.157. Li D, Zheng J, Weinmann H. Contrast-enhanced MR imaging of coronary arteries: Com-parison of intra- and extravascular contrast agentsin swine. Radiology. 2001;218:670-678.158. Reimer P, Allkemper T, Matuszewski L,Balzer T. Contrast-enhanced 3D-MRA of theupper abdomen with a bolus-injectable SPIO(SH U 555 A). J Magn Reson Imaging. 1999;10:65-71.159. Mayo-Smith W, Saini S, Slater G, et al. MRcontrast material for vascular enhancement: Valueof superparamagnetic iron oxide. AJR Am JRoentgenol. 1996;166:73-77.160. Weishaupt D, Ruhm S, Binkert C, et al. Equi-librium-phase MR angiography of the aortoiliac

and renal arteries using a blood pool contrastagent. AJR Am J Roentgenol. 2000;175:189-195.161. Taylor A, Panting J, Keegan J, et al. Safetyand preliminary findings with the intravascularcontrast agent NC100150 injection for MR coro-nary angiography. J Magn Reson Imaging. 1999;9:220-227.162. Bachmann R, Conrad R, Kreft B, et al. Evalu-ation of a new ultrasmall superparamagnetic ironoxide contrast agent Clariscan (NC100150) forMRI of renal perfusion: Experimental study in ananimal model. J Magn Reson Imaging. 2002;16:190-195.163. Campeau NG, Huston J 3rd, Bernstein MA,er al. Magnetic resonance angiography at 3.0Tesla: Initial clinical experience. Top Magn ResonImaging. 2001;12:183-204. 164. Al-Kwifi O, Emery D, Wilman A. Vesselcontrast at three Tesla in time-of-flight magneticresonance angiography of the intracranial andcarotid arteries. Magn Reson Imaging. 2002;20:181-187.165. Reichenbach J, Barth M, Haacke E, et al.High-resolution MR venography at 3.0 Tesla. JComput Assist Tomogr. 2000;24:949-957.166. Hugg J, Rofsky N, Stokar S, et al. Clinicalwhole body MRI at 3.0 T - initial experience. Pre-sented at the Xth Scientific Meeting and Exhibi-tion of the ISMRM. Honolulu, HI, 2002. 167. Leiner T, Vasbinder B, De Vries M, et al.Contrast-Enhanced Peripheral MRA at 3.0T: Ini-tial results. Presented at the 10th Scientific Meet-ing and Exhibition of the ISMRM. Honolulu, HI,2002. 168. Stuber M, Botnar R, Fischer S, et al. Prelimi-nary report on in vivo coronary MRA at 3 Tesla inhumans. Magn Reson Med. 2002;48:425-429.

This activity has been planned and implemented in accordance with the

Essential Areas and Policies of the Accreditation Council for Continuing

Medical Education (ACCME) through the joint sponsorship of the Institute

for Advanced Medical Education and Anderson Publishing, Ltd. The Insti-

tute for Advanced Medical Education is accredited by the Accreditation

Council for Continuing Medical Education (ACCME) to sponsor continu-

ing medical education for physicians.

The Institute for Advanced Medical Education designates this continuing

medical education activity for a maximum of 3 Category 1 credits toward the

AMA Physician’s Recognition Award. Each physician should claim only

those credits that he/she actually spent in the activity.

See page 182 for instructions on how to participate in the program.

As of January 1, 2003, courses approved for AMA Category 1 CME credit

that are relevant to the radiologic sciences are accepted for Category B

CE credit on a one-to-one basis by the American Registry of Radiologic

Technologists (ARRT).

In compliance with the Essentials and Standards of the ACCME, the

authors of this CME tutorial are required to disclose any significant finan-

cial or other relationships they may have with the manufacturer(s) of any

commercial product(s) or provider(s) of any commercial service(s) dis-

cussed in this program. Dr. Maki reports a relationship with EPIX Medical

as a consultant and through research funding. Dr. Knopp reports relation-

ships with Amersham Health, Berlex Imaging/Schering, and Bracco Diag-

nostics as a consultant. Dr. Prince reports relationships with TopSpins,

Inc. as a major shareholder; as well as patent agreements with GE Medical

Systems, Siemens Medical Systems, Mallinckrodt, Bracco Diagnostics,

and Medrad, Inc.

CME Information

Documents

Contrast-Enhanced MR Angiography CATEGORY 1Contrast-Enhanced MR Angiography Jeffrey H. Maki, MD, PhD; Michael V. Knopp, MD, PhD; Martin Prince, MD, PhD Dr. Maki is an Associate Professor