Technical Note

Blood Attenuation With SSFP-CompatibleSaturation (BASS)Hung-Yu Lin, MS,1,2 Brian M. Dale, PhD,1,2 Chris A. Flask, PhD,1 andJeffrey L. Duerk, PhD1,2*

Purpose: To investigate a rapid flow-suppression method forimproving the contrast-to-noise ratio (CNR) between the vesselwall and the lumen for cardiovascular imaging applications.

Materials and Methods: In this study a new dark-bloodsteady-state free precession (SSFP) sequence utilizing two exci-tation pulses per TR was developed. The first pulse is appliedimmediately adjacent to the slice of interest, while the second isa conventional slice-selective pulse designed to excite an SSFPsignal for the static spins in the slice of interest. The slice-selec-tive pulse is followed by fully refocused gradients along all threeimaging axes over each TR. The signal amplitude (SA) from themoving spins excited by the “saturation” pulse is attenuatedsince they are not fully refocused at the TE.

Results: This work provides confirmation, by both simula-tion and experiments, that modest adaptations of the basicTrue-FISP structure can limit unwanted “bright blood” sig-nal within the vessels while simultaneously preserving thecontrast and speed advantages of this well-establishedrapid imaging method.

Conclusion: Animal imaging trials confirm that dark-bloodcontrast is achieved with the BASS sequence, which sub-stantially reverses the lumen-to-muscle CNR of a conven-tional True-FISP “bright blood” acquisition from 14.77(bright blood) to –13.96 (dark blood) with a modest increase(24.2% of regular TR of SSFP for this implementation) inacquisition time to accommodate the additional slab-selec-tive excitation pulse and gradient pulses.

Key Words: flow suppression; steady-state free precession;vessel wall image; dark blood; black bloodJ. Magn. Reson. Imaging 2006;24:701–707.© 2006 Wiley-Liss, Inc.

DARK-BLOOD CONTRAST MECHANISMS are essentialfor cardiovascular imaging applications, such as tissue

characterization (e.g., acute infarct), right ventriculardysplasia (e.g., fatty infiltration), and atheroscleroticplaque characterization, where the normal bright bloodsignal can make it difficult to distinguish features of thevessel wall from the vessel lumen. Several techniqueshave been developed to reduce the bright-blood signalfor cardiovascular MRI applications, such as doubleinversion recovery (DIR) magnetization preparation (1),tailored radiofrequency (RF) pulse design (2), and spa-tial saturation (3). The DIR technique is currently themost commonly used dark-blood acquisition strategy.The basic idea behind this sequence is to utilize thesignificant T1 relaxation difference between blood andsurrounding tissue in order to null the signal fromflowing blood but maintain normal tissue contrast forthe stationary tissues within the slice. However, theprohibitively long acquisition times may contribute tosubstantial motion artifacts that limit the use of DIRsequences in some routine cardiovascular MRI applica-tions. Therefore, rapid imaging applications, such astime-resolved imaging of the heart and real-time imag-ing, typically employ saturation pulses to eliminate in-flowing blood signal (4,5). Unfortunately, saturationmethods are not widely adaptable to steady-state freeprecession (SSFP) sequences because the saturationpulses perturb the normal steady-state equilibrium.

Recently the fully gradient refocused fast imagingwith steady-state precession (True-FISP) sequence (6,7)(see Fig. 1) has received increasing interest because ofits short acquisition time, high signal-to-noise ratio(SNR) per unit time, and relevant contrast for cardio-vascular applications. True-FISP produces a coherentsteady state by balancing the gradient lobes along allthree gradient axes, resulting in a characteristicallyhigh equilibrium signal. However, the SA from blood ina True-FISP image is greater than most static tissuesbecause of the magnetic properties of the blood, as wellas contributions from rapid through-plane flow. In thisstudy a pulse sequence is proposed that builds on thefundamental concepts of spatial presaturation to createtwo steady states: one for stationary tissue within theslice of interest, and one for tissue outside of the slice(and hence moving blood) (5). The blood attenuationwith SSFP-compatible saturation (BASS) sequence wasdeveloped to reduce the signal from moving spins priorto their entry into the imaging slice, and hence to im-prove both visualization of the vessel wall and robust-ness to flow artifacts. The major advantage of this tech-

1Department of Radiology, University Hospitals of Cleveland and CaseWestern Reserve University, Cleveland, OH, USA.2Department of Biomedical Engineering, Case Western Reserve Univer-sity, Cleveland, OH, USA.Hung-Yu Lin is now at the Department of Biomedical Engineering, OhioState University, Columbus, OH, USA.Brian M. Dale is now at Siemens Medical Solutions, Inc. Cary, NC, USA.*Address reprint requests to: J.L.D., Department of Radiology/MRI,B100 Bolwell Bldg., University Hospitals of Cleveland, 11100 EuclidAve., Cleveland, OH 44106.E-mail: duerk@uhrad.comReceived June 7, 2005; Accepted May 5, 2006.DOI 10.1002/jmri.20657Published online 4 August 2006 in Wiley InterScience (www.interscience.wiley.com).

JOURNAL OF MAGNETIC RESONANCE IMAGING 24:701–707 (2006)

© 2006 Wiley-Liss, Inc. 701

nique is that it can acquire dark-blood contrast SSFPimages without the long inversion time and single-slicelimitation required by the traditional DIR method. Spe-cifically, our work expands the application of dark-blood techniques to SSFP acquisitions by designing RFexcitation/gradient combinations that result in a fullyrefocused coherent steady state for static spins and adephased, incoherent steady state for flowing spins.

MATERIALS AND METHODS

Pulse Sequence Design of the BASS Method

A variant of a conventional True-FISP pulse sequence isdeveloped with two RF excitation pulses in series (�1 �90° for out-of-slice excitation, �2 � 60° for slice excita-tion, TR � 8.8 msec, TE � 3.6 msec, and image ma-trix � 192 � 256; Fig. 2). The first RF pulse (�1) excitesa large slab adjacent to the imaging slice. RF spoiling isapplied by varying the phase of �1 in subsequent exci-tation intervals. Refocusing gradients along the slice-select direction are not immediately applied prior to theimaging slice excitation pulse (�2). In theory, corruptionof the desired imaging slice signal by magnetizationfrom static spins in the out-of-slice slab is avoided byintentional lack of rephasing along the slice-select di-rection.

The excitation pulse �2 (i.e., for slice selection) isfollowed by a standard rephasing slice-select gradientwaveform, and fully refocused read and phase-encod-ing gradients. The slice-select gradients following �2 aredesigned to fully refocus the imaging slice magnetiza-tion prior to the echo time (TE) as well as immediatelyprior to the next slice-selective pulse to maintain thecoherent steady-state (i.e., equal but opposite polarityof gradient moment for shaded regions; Fig. 2). At thesame time, the slice-select gradients are intentionallydesigned to avoid refocusing the slab magnetization(i.e., moving spins), which experienced an additionalslab-saturation gradient at the TEs and prior to thenext slice-selective excitation. By the end of each repe-tition time (TR), a gradient is applied to achieve zerogradient moment within each TR and preserve the co-herent steady-state equilibrium (i.e., negative gradientmoment is balanced by refocused gradient prior to nextTR; Fig. 2).

Hence, we established two different steady-stateequilibria that correspond to fast low angle shot

(FLASH)-like magnetization for the stationary tissue inthe out-of-slice slab and the incoming moving spins via�1, and its RF spoiling and True-FISP contrast for sta-tionary spins in the imaging plane via �2 and the fullyrefocused gradients. The lack of refocusing following �1

further dephases the out-of-slice spins at the TE andencodes them at a different, nonzero, kz-space locationfrom those within the imaging slice. The reduction ofsignal from spins flowing into the imaging slice (i.e.,from fully relaxed to FLASH steady state) should lead tosignificant reductions in the overall lumen signal.Therefore, the BASS sequence acquires relatively highsignal amplitude (SA) for static spins with fully refo-cused gradients along all three imaging axes over eachTR via a coherent, refocused SSFP acquisition (e.g.,True-FISP, fast imaging employing steady-state acqui-sition (FIESTA), or SSFP). It also achieves signal atten-uation for moving spins by RF spoiling (e.g., withFLASH, spoiled gradient recalled acquisition in steadystate (GRASS), or incoherent SSFP) and dephasing theout-of-slice magnetization.

Simulations

The general formalism of signal behavior for static spinsis described by a matrix treatment by Jaynes (8). Amodification of an earlier steady-state flow phenome-non simulation was written in Mathematica (WolframResearch, Inc., Champaign, IL, USA). All simulationswere performed with blood-like (T1 � 1000 msec, T2 �150 msec, T2* � 75 msec) and muscle-like (T1 � 900msec, T2 � 50 msec) relaxation time constants. Analyticexpressions for the rotation matrices and magnetiza-tion distributions were generated to solve the Blochequations with plug and continuous flow assumptions.The SSFP signal behavior is determined by three matri-ces corresponding to 1) Rx(��, which describes an in-stantaneous RF pulse with a flip angle (FA) of � andalternated RF-phase for consecutive repetitions as ap-propriate for either RF spoiling or phase alternation; 2)FPz(�TR), which represents the phase evolution functionwithin one TR interval (e.g., due to gradients and off-resonance); and 3) two relaxation matrices representingT1 relaxation and T2 decay E2(TR,T1,T2) (9–11). Based onthe Bloch equations, it can be shown that the magne-tization directly after the nth RF pulse is given by thefollowing recursive expression:

Figure 1. Pulse sequence diagram for aconventional True-FISP sequence. Notethat all gradients are fully balanced onthree axes to maintain the coherentsteady state of longitudinal and trans-verse magnetization within a short ac-quisition time.

702 Lin et al.

M� n� � Rx� � ��FPz��TR�E2(TR,T1,T2)M� n � 1

� � E1(TR,T1)]

(1)

where

Rx(��) � � 1 0 00 Cos��� � Sin��)0 �Sin��� Cos���

�,FPz(�TR) �� Cos��TR� � Sin��TR) 0

Sin��TR� Cos��TR) 00 0 1

� (2)

E1(TR,T1) � � 00

Mo(1 � e�TR/T1)�,

and E2(TR,T1,T2) � � e�TR/T2 0 00 e�TR/T2 00 0 e�TR/T1

� (3)

Our simulation scheme for the BASS sequence com-bines incoming spins with unsaturated magnetization,establishment of an out-of-slice RF spoiled steady state

Figure 2. Pulse sequence diagram for theBASS flow suppression sequence. Note thatthe four shaded areas of the slice-section axisindicate regions of fully balanced gradient mo-ment within each TR. Moving spins continuedephasing after slice-selective excitation be-cause of an unbalanced gradient in the hol-lowed-shadowed region. During each TR aFLASH-like saturation pulse, �2, is applied(using RF spoiling) prior to exciting the imagingslice, �1 (using RF phase alternation). Note thatat the TE the slice-select gradient is refocusedfor spins excited by �2, but not for spins excitedby �1.

Figure 3. Note: To improve curve visualization, a different vertical scale is used in each figure. MR signal as a function ofdephasing/TR (resonance offset angle) for both flowing and stationary spins. Sequences and conditions simulated: (a) standardTrue-FISP with in-flow effect, (b) conventional True-FISP with in-flow effect and out-of-slice contribution, (c) BASS sequence within-flow effect, and (d) BASS sequence with in-flow effect and out-of-slice contribution. (TE/TR/FA-slab/FA-slice/T1blood/T2blood/T1Tissue/T2Tissue � 2 msec/4 msec/90°/60°/1000 msec/150 msec/900 msec/50 msec).

BASS 703

signal proximal to the imaging slice, a SSFP signal fromthe imaging slice, and out-of-slice contributions gener-ated by spins that have left the image slice but stillcontribute signal because of the zero gradient momentsof the SSFP acquisitions. In order to specifically de-scribe the moving spins that flow through an imageslice, blood velocity is encoded by the “inflow rate” pa-rameter, which represents the percentage of unsatur-ated spins that flow into the image slice within each TRinterval (this is similar to an earlier flow simulation byMarkl et al (12)). According to the current protocol(TR � 4 msec, slice thickness � 5 mm), an inflow rateequal to 20% of the slice thickness corresponds to ablood flow velocity of 25 cm/second.

Simulated moving spins begin with full magnetiza-tion as they enter the saturation slab, where they expe-rience multiple RF pulses with RF spoiling. They thenprogress into the imaging slice with partially saturatedmagnetization, where they experience an RF pulse withslice-select gradient. They then move out of the slice,where only gradient activity is encountered. However,the gradient effects are not only confined to the imagingslice in the MR scanner. The gradient refocused echoesstill contribute SA due to the fully balanced nature ofthe True-FISP sequence; this signal contribution iscalled the “out-of-slice component” (12). This contribu-tion is calculated with signal attenuation from the mov-ing spins from T2* decay without further RF excitationbut with continued gradient activity. Solutions are alsoobtained for static spins within the imaging slice. Plotsof signal from moving spins and static spins within theimaging slice are obtained as a function of the reso-nance offset angle (13).

Animal Studies

We performed axial abdominal imaging on three Wa-tanabe heritable hyperlipidemic (WHHL) rabbits (meanage � 26 months) in a 1.5T Siemens Sonata MRI scan-ner (Siemens Medical Solutions, Erlangen, Germany) toobtain comparative images for both BASS (FA-slab �90°, FA-slice � 60°, image-slice thickness � 5 mm,saturation-slab thickness � 40 mm, gap between twoexcitations � 30 mm, bandwidth � 390Hz/pixel,FOV � 130 mm, spatial resolution � 0.5 � 0.5 mm,TE � 3.6 msec, TR � 11.0 msec, matrix � 192 � 256,NSA � 2, time � 4.1 sec) and conventional True-FISP(FA-slice � 60°/ image-slice thickness � 5 mm, band-width � 390 Hz/pixel, FOV � 130 mm, spatial resolu-tion � 0.5 � 0.5 mm, TE � 3.6 msec, TR � 8.8 msec,matrix � 192 � 256, NSA � 2, time � 3.3 sec) acquisi-tions. All of the animal imaging experiments were con-ducted under an approved institutional animal care

and use committee (IACUC) protocol. All of the acquisi-tions were performed without cardiac or respiratorygating.

RESULTS

Simulations

Figure 3 shows a simulation of a conventional True-FISP acquisition, the dependence of the signal with flowat 20% and 50% inflow rates (in comparison with statictissue), and the resonance-offset angle dependence.Signal increases above static tissue are associated withboth the in-flow effects (i.e., additional influx of unsat-urated magnetization; Fig. 3a) and the out-of-slice con-tributions (Fig. 3b), as previously reported by Markl andPelc (11). Figure 3c and d demonstrate significant flowsignal reduction (below that of stationary spins) whenthe BASS sequence is used. The SAs of moving spinsare predicted to be substantially attenuated from2.7638 and 8.0946 (arbitrary units) to 0.0158 and0.1007 in 180° resonance-offset angle for 20% and 50%inflow rate conditions, respectively. Note that as a func-tion of off-resonance, the overall signal variation of theflowing spins is significantly reduced in the BASS sim-ulation. Figure 3c also shows that stationary tissue hasa signal characteristic similar to that of the standardTrue-FISP pulse sequence (Fig. 3a). The inflow effectand out-of-slice contributions are reduced by at least98% of the magnitude in the BASS sequence, even inthe case of a 180° resonance offset angle (Table 1). Notethat even though it is difficult to measure the exactvelocity, relaxation properties, and pulsatile effects offlowing blood for the in vivo model, the simulation re-sults still predict the general flow signal reductiontrend—specifically, the reduction of the blood SA of aconventional True-FISP “bright-blood” acquisition from0.4365 (bright blood) to 0.0014 (dark blood) whilemaintaining the regular tissue SA of 0.0825. This rep-resents a predicted change in blood–tissue contrast((SAblood – SAtissue)/mean(SAtissue,SAblood)) from 1.4(bright blood) to –2.0 (dark blood).

In Vivo Imaging Studies

In vivo images obtained with a conventional True-FISPsequence and the BASS sequence are displayed in Fig.4. Figure 4a shows an axial image of the standard True-FISP acquisition in the abdomen of a WHHL rabbit.Note that the blood flow gives rise to the familiar bloodflow artifact (arrows in Fig. 4a). Figure 4b–d were ac-quired by placing the saturation slab of the BASS ac-quisition superior, inferior, and both superior and in-

Table 1Signal Amplitude Comparison of Static and Moving Spins With 20% and 50% Inflow Rate in the Simulation Model

SequencesMoving spins (20% inflow rate) Moving spins (50% inflow rate) Static spins

On resonance Off resonance On resonance Off resonance On resonance Off resonance

Standard True-FISP 0.4365 2.7637 0.4317 8.0946 0.0825 0.0038BASS 0.0014 0.0158 0.0008 0.1007 0.0825 0.0038SA reduction (%) 99.68 99.43 99.81 98.76 0.00 0.00

SA � signal amplitude, BASS � blood attenuation with SSFP-compatible saturation, SA reduction � (SATrue-FISP – SABASS)/SATrue-FISP.

704 Lin et al.

ferior (two saturation slabs) to the image slice,respectively. The initial results in Fig. 4b–d demon-strate the excellent dark-blood contrast that can beachieved using this technique while preserving the con-trast properties of other tissues. CNR measurementsare based on the difference in mean pixel value betweentwo regions of interest (ROIs) and random noise. Notethat the images with arterial suppression do not showghost artifacts even though cardiac gating is not used.For the purpose of consistent measurement, we chose

the same ROIs to measure the mean and standarddeviation (SD) of the SA in the conventional and BASSacquisitions. The lumen-to-muscle CNR value is calcu-lated by the following equation:

CNRLumen�Muscle � �SALumen � SAMuscle�/SANoise (4)

For the BASS method, the lumen-to-muscle contrastwas opposite that of the True-FISP acquisition. Thelumen-to-muscle CNR was dramatically decreasedfrom 14.77 and 5.82 (bright blood, in the vein andartery, respectively, for the conventional True-FISP im-age, and to –13.96 and –13.13 (dark blood) with theBASS technique (Table 2). Note the excellent suppres-sion of inflowing spins in the vein (86%) and artery(78%) blood signal reduction, respectively, yet consis-tent True-FISP contrast for the stationary tissues.These images demonstrate the typical SSFP high SAand contrast achieved between flowing blood and nor-mal stationary tissue in conventional coherent SSFPacquisitions, and the blood flow signal attenuationachieved with the new BASS sequence.

DISCUSSION

In this study we show that the BASS sequence canachieve desirable dark-blood contrast in an SSFP se-quence with short TR. Computer simulations and invivo imaging results confirm that the SA of movingspins is significantly reduced while the signal fromstatic spins is maintained (Figs. 3 and 4). The computersimulations also indicate that, compared to the stan-dard True-FISP sequence, the BASS sequence is muchless sensitive to off-resonance effects for flowing spins,while it maintains the signal properties for stationaryspins (Fig. 3a and b vs. c and d). Reasonable agreementwas achieved between simulation results, and CNR/SNR measurements in in vivo experiments (Tables 1and 2) confirm the reduction of the inflow signal (viaout-of-slice spoiled steady-state signal generation) evenin the presence of uncertainties in blood relaxationtimes, off-resonance, blood velocity, and pulsatile ef-fects found in vivo. It appears that the simulation modelcan practically predict the signal trend of stationaryand moving spins.

Figure 4. WHHL rabbit abdominal aorta wall image using (a)conventional True-FISP (FA-slice � 60°, TE � 3.6 msec, TR �8.8 msec, matrix � 192 � 256, NSA � 2, time � 3.3 sec); (b)BASS sequence with arterial saturation (FA-slice � 60°, TE �3.6 msec, TR � 8.8 msec, matrix � 192 � 256, NSA � 2, time �3.3 sec); (c) BASS sequence with venous saturation (FA-slab �90°, FA-slice � 60°, TE � 3.6 msec, TR � 8.8 msec, matrix �192 � 256, NSA � 2, time � 3.3 sec); and (d) BASS sequencewith arterial and venous saturation (FA-slab � 90°, FA-slice �60°, TE � 3.6 msec, TR � 11.0 msec, matrix � 192 � 256,NSA � 2, time � 4.1 sec). Note that the arrows in image aindicate the flow artifact, which is spread out in the phase-encoding direction.

Table 2Signal-to-Noise and Contrast-to-Noise Measures in the Abdomen of a WHHL Rabbit

Sequences SA in artery SA in vein SA in muscle SA of Noise

Standard True-FISP 381.4 � 55.1 518.3 � 41.2 292.3 � 15.4 48.5 � 15.3Arterial saturation in BASS 72.5 � 23.6 469.7 � 45.8 367.8 � 24.3 49.5 � 15.3Venous saturation in BASS 192.5 � 71.3 116.0 � 44.5 358.5 � 29.8 48.8 � 13.7Arterial and venous saturation in BASS 82.3 � 26.3 69.7 � 20.6 280.5 � 25.4 49.4 � 15.1

SequencesFlow SA

reduction in arteryFlow SA

reduction in veinCNR in artery CNR in vein

Standard True-FISP N/A N/A 1.84 4.66Arterial saturation in BASS 80.99% 9.38% �5.97 2.06Venous saturation in BASS 69.59% 77.62% �3.40 �4.97Arterial and venous saturation in BASS 78.42% 86.55% �4.01 �4.27

SA � signal amplitude, BASS � blood attenuation with SSFP-compatible saturation, CNR � (SALumen � SAMuscle)/SANoise.

BASS 705

The major advantage of the BASS sequence is thatone can readily achieve dark-blood contrast in an SSFPsequence by introducing an additional RF pulse andtailored gradient moment design for suppression ofmoving spins. One minor advantage is that with arterialflow suppression there are no ghost artifacts, evenwithout the use of cardiac gating, as shown in Fig. 4band d. This decreases acquisition time for the vascularapplications considered here.

Moreover, this technique differs from the conven-tional presaturation method (4,5) in a number of ways.First, by applying the saturation pulse every TR, theBASS method establishes a true steady state for boththe imaging slice and the saturation slab. In contrast,the presaturation method interrupts the steady state ofthe base sequence by applying the saturation only onceper image. This leads to a suppression effect that de-cays in time, often requires centric reordering, and of-ten is effective for only a single slice. Second, the sup-pression slab signal is not assumed to be zero in theBASS method, although it is intentionally not refocusedat the echo. Finally, the saturation method has beenused for all sequences, whereas to date the BASSmethod has been applied only for True-FISP imagingsequences.

It is important to utilize a large FA for the FLASH-likeslab saturation to ensure that rapidly moving spins arein incoherent steady-state equilibrium even thoughthey may only experience a single or limited number ofexcitations. Further, the large FA results in a low signalfrom the FLASH-like steady state in the slab, which isimportant for avoiding unwanted signal contamination.Unlike earlier spatial presaturation methods that wereintended primarily for spin-echo (SE) methods, we con-sider the RF amplitude, RF phase, post-RF pulse gra-dient structure, and resultant flowing signal character-istics in this steady-state (vs., for example, SE) regime.The inflow effect could potentially be further reduced bydesigning improved slice excitation profiles to place thesaturation slab as close as possible to the imaging slice.In this preliminary trial, simple truncated sinc pulseswere used. Improved excitation would limit both cross-talk and signal regrowth as slowly moving spins passthrough the gap between the saturation slab and theimaging slice. It is also worth pointing out that theout-of-slice and slice-selective RF pulses can be imple-mented either separately or as a family of combinedexcitations with varying phase in the slab region andalternating phases in the slice region. Moreover, maxi-mum-phase pulses designed with the Shinnar-LeRouxtechnique (14) may be desirable for the saturation slabbecause of their non-refocusing properties. The tailoredRF pulse design, which could combine a dual bandsaturation slab and regular slice excitation within onecomposite RF pulse, would also allow further reductionin the duration of the RF excitation pulse train andsubsequent reduction in the overall acquisition time.Such optimization may be particularly important forclinical use, since it should mitigate some of the addi-tional banding artifact that could arise from the current24% increase in TR. Finally, optimization of the RFpulses may mitigate some or all of the increase in spe-cific absorption rate (SAR) experienced with this se-

quence. Even though the total energy deposition morethan doubles with the current implementation of BASS(�1 � 90° � �2), the SAR increase is not as dramaticbecause of the concurrent increase in TR. Many True-FISP applications are not SAR-limited, but without theaforementioned optimization, BASS may not be appro-priate for applications that are SAR-limited.

The simulation results and in vivo CNR/SNR mea-surements (Figs. 3 and 4) show that perfect RF excita-tion pulses are not necessary to suppress the blood flowsignal in the BASS sequence. Therefore, the CNR per TR(minimum) is anticipated to be more important thanperfect RF excitation when the method is attempted inthe clinical setting.

The BASS dark-blood sequence is designed forthrough-plane flow suppression. It is less efficient atsuppressing in-plane flow because the moving spinshave zero velocity along the slice-selection axis. There-fore, the saturation pulse will be ineffective at alteringthe incoming magnetization. However, the reduction inacquisition time coupled with the improvement in se-quential multislice capabilities makes BASS suitablefor many clinical applications. For example, atheroscle-rotic plaque characterization typically utilizes an imag-ing plane that is perpendicular to the axis of vessel.

The simulation tool created here allows adjustment ofall aspects of the pulse sequence, including the RF tipangles, RF phases, dephasing intervals, inflow rate, TR,and TE. Prior to phantom and animal experiments,future research can explore modifications of the basicsequence in order to achieve further improvements inflow suppression.

In conclusion, this work confirmed by both simula-tions and experiments that modest adaptations of thebasic True-FISP structure can prevent unwantedbright-blood signal within the vessels while simulta-neously preserving the static-tissue contrast and speedadvantages of this well-established rapid imagingmethod. The results are achieved via the establishmentof two separate steady states within one sequence. Thisprovides successful flow suppression and high CNRbetween stationary tissues and flowing blood. An ex-plicit comparison of the experimental results and the-oretical simulations is hindered by uncertainty abouttissue and blood relaxation properties, rabbit aorticflow rates, etc. However, the general nature of signifi-cant flow reductions and almost complete flow suppres-sion were observed. This technique should permit rapidimaging of vessel walls, and may have a significantimpact on the utility of MRI for routine screening ofatherosclerotic plaques.

REFERENCES

1. Edelman RR, Chien D, Kim D. Fast selective black blood MR imag-ing. Radiology 1991;181:655–660.

2. Ro YM, Cho ZH. A novel flow-suppression technique using tailoredRF pulses. Magn Reson Med 1993;29:660–666.

3. Edelman RR, Mattle HP, Wallner B, et al. Extracranial carotidarteries: evaluation with “black blood” MR angiography. Radiology1990;177:45–50.

4. Nayak KS, Rivas PA, Pauly JM, et al. Real-time black-blood MRIusing spatial presaturation. J Magn Reson Imaging 2001;13:807–812.

706 Lin et al.

5. Felmlee JP, Ehman RL. Spatial presaturation: a method for sup-pressing flow artifacts and improving depiction of vascular anatomyin MR imaging. Radiology 1987;164:559–564.

6. Oppelt A, Graumann R, Barfuss H, Fischer H, Hartl W, Schajor W.FISP: a new fast MRI sequence. Electromedica (Engl Ed) 1986;54:15–18.

7. Hawkes RC, Patz S. Rapid Fourier imaging using steady-state freeprecession. Magn Reson Med 1987;4:9–23.

8. Jaynes ET. Matrix treatment of nuclear induction. Phys Rev 1955;98:1099–1105.

9. Scheffler K. A pictorial description of steady-states in rapid mag-netic resonance imaging. Concepts Magn Reson 1999;11:291–304.

10. Scheffler K, Lehnhardt S. Principles and applications of balancedSSFP techniques. Eur Radiol 2003;13:2409–2418.

11. Markl M, Pelc NJ. On flow effects in balanced steady-state freeprecession imaging: pictorial description, parameter dependence,and clinical implications. J Magn Reson Imaging 2004;20:697–705.

12. Markl M, Alley MT, Elkins CJ, Pelc NJ. Flow effects in balancedsteady state free precession imaging. Magn Reson Med 2003;50:892–903.

13. Haacke EM, Wielopolski PA, Tkach JA, Modic MT. Steady-state freeprecession imaging in the presence of motion: application for im-proved visualization of the cerebrospinal fluid. Radiology 1990;175:545–552.

14. Pauly J, Le Roux P, Nishimura D, Macovski A. Parameter relationsfor the Shinnar-Le Roux selective excitation pulse design algorithm[NMR imaging]. IEEE Trans Med Imaging 1991;10:53–65.

BASS 707