Water-Fat Separation Imaging of the Heart with StandardMagnetic Resonance bSSFP CINE Imaging

James W. Goldfarb1,2* and Sheeba Arnold-Anteraper1

Purpose: To study balanced steady-state free precession

CINE phase-sensitive water-fat separation imaging in four car-diac imaging planes to determine the necessary phase correc-

tion and image artifacts particular to this technique.Methods: Ten healthy volunteers and two subjects with knownheart pathologies were studied with standard balanced

steady-state free precession CINE imaging. Water-only andfat-only images were calculated using sign detection of thereal part of the complex image after phase correction with

constant and linear terms. Phase correction values were deter-mined using both manual and automated methods. Differen-

ces in phase correction values between imaging planes,cardiac phases, coil elements, automated image reconstruc-tion parameters as well as artifact scores between the auto-

mated and manual methods were studied with statistical tests.Results: Water-fat separation performed well in the heart after

constant and linear phase correction. Both constant (p ¼ 0.8)and linear x (p ¼ 1) and y (p ¼ 1) phase correction values didnot vary significantly across cardiac phases, but varied signifi-

cantly among the coils (p < 0.001) and imaging planes (p <

0.001). False water-fat separation artifacts were most frequent

in the chest/back and also were present at the mitral and aor-tic valves.Conclusion: Constant and linear phase correction is neces-

sary to provide consistent results in standard imaging planesusing a balanced steady-state free precession water-fat sepa-

ration postprocessing algorithm applied to standard cardiacCINE imaging. Magn Reson Med 71:2096–2104, 2014.VC 2013 Wiley Periodicals, Inc.

Key words: magnetic resonance imaging; fat water separa-

tion; cardiovascular magnetic resonance imaging; myocardialinfarction; cardiac tumor; lipid; image reconstruction

During a typical clinical magnetic resonance (MR) imag-ing study of the heart, multiple pulse sequences are usedfor a functional assessment and myocardial tissue char-acterization. CINE sequences are primarily used for theregional and global assessment of cardiac function andadditional single cardiac-phase sequences are typically

used for tissue characterization. MR imaging is nowwidely recognized as a standard for the quantitativeassessment of ventricular function (1) and myocardialmass (2). As in other parts of the body, MR imaging’sexcellent soft tissue contrast generating abilities lead tonumerous myocardial tissue characterization indications;for example tumor evaluation (3), myocardial ischemia(4), and viability (5). As a result of the breadth of cardio-vascular MR imaging applications, the examination timeis long: ranging from 30 min to over an hour due to theuse of multiple pulse sequences and imaging planes.

Balanced steady-state free precession (bSSFP) imaging(6,7) has become the recognized standard in the functionalassessment of the heart using CINE MR imaging due to itsefficiency and excellent myocardial to blood image con-trast (8). Conversely, both pericardial fluid and neighbor-ing fat have strong image intensities in CINE-bSSFPimages similar to blood and are not well identified basedon their image intensities. Therefore, characterization offatty masses, myocardial fat infiltration, and discrimina-tion of fluid and blood from fat are typically performedusing additional pulse sequences with T1 weighting orpreparation pulses such as chemical selective fat suppres-sion (9), short inversion time inversion recovery (10,11),or multiple gradient-echo imaging (12–14). Hargreaveset al. (15) described a phase-sensitive bSSFP techniquethat provides fat suppression without additional complex-ity or scan time in a standard bSSFP sequence with appli-cations in the extremities. It was shown that a typicalbSSFP acquisition with appropriate selection of the repe-tition time (TR) and center frequency results in water andlipid signals having opposite polarities. Due to the fre-quent use of bSSFP in cardiovascular applications, thiswater-fat separation technique would simultaneously pro-vide functional assessment and tissue characterization ina single pulse sequence, reducing imaging time withoutadditional complexities for the patient.

Phase-sensitive water-fat separation in standard bSSFPCINE imaging was studied in a series of healthy volun-teers and several patients with known fat pathologies infour cardiac imaging planes. Because both bSSFP imag-ing and water-fat separation are sensitive to main mag-netic field (B0) inhomogeneities, the main purpose ofthis study was to investigate and compare the necessaryphase correction through manual and automated techni-ques. Additionally, image artifacts particular to this tech-nique were determined.

METHODS

Patient Population

In this study, we enrolled 12 subjects: 10 healthy volun-teers (age ¼ 62.3 6 9.0 years, range 44.5–72.6; male (n ¼5)) and two male patients (ages 72.3 and 79.0) with

1Department of Research and Education, Saint Francis Hospital, Roslyn,New York, USA.2Program in Biomedical Engineering, SUNY Stony Brook, Stony Brook,New York, USA.

*Correspondence to: James W. Goldfarb, Ph.D., Department of Researchand Education: DeMatteis MRI, St. Francis Hospital, 100 Port WashingtonBoulevard, Roslyn, NY 11576. E-mail: james.goldfarb@chsli.org

Additional Supporting Information may be found in the online version ofthis article.

Received 17 April 2013; revised 18 June 2013; accepted 18 June 2013

DOI 10.1002/mrm.24879Published online 31 July 2013 in Wiley Online Library (wileyonlinelibrary.com).

Magnetic Resonance in Medicine 71:2096–2104 (2014)

VC 2013 Wiley Periodicals, Inc. 2096

known heart pathologies (cardiac tumor and myocardialinfarction with fat deposition). All subjects signed aninstitutional review board approved, Health InsurancePortability and Accountability Act compliant consentform before study initiation.

Imaging Protocol

All MR examinations were performed with a clinical 1.5T imager (Magnetom Sonata, Siemens Healthcare, Erlan-gen, Germany) with the subject in the supine positionand standard (“tune-up”) magnetic field shim. Imageswere acquired during suspended respiration at end-expiration with ECG gating. The standard circularlypolarized flexible four-element body and spine phasedarray coils were used for signal reception. bSSFP CINEimaging in mid-short axis, four-chamber (horizontal longaxis), two-chamber (vertical long axis), and three cham-ber (left ventricular outflow tract (LVOT)) imaging planeswas performed. bSSFP sequence parameters were 1 sliceper breath-hold (14 heartbeats), TR ¼ 3.2 ms, echo time¼ 1.6 ms, flip angle ¼ 60 degrees, bandwidth ¼ 930 Hz/pixel, in-plane spatial resolution ¼ 1.6 � 1.3 mm2, tem-poral resolution ¼ 50 ms, slice thickness ¼ 8 mm, rawdata matrix ¼ 210 � 256, rectangular field of view (FOV)¼ 81%, no parallel imaging or view sharing. Raw k-space data was automatically saved to the scanners harddrive and then transferred to a standalone computer forwater-fat separated image reconstruction.

Image Reconstruction

A custom computer application written in the JAVA pro-gramming language was developed for manual and auto-mated phase correction and image reconstruction. Theprogram reads the raw k-space data, [s(kx,ky) where x isthe readout and yis the phase endoding dimension], foreach coil element from disk and first reconstructs an ini-tial complex image for each coil element using a two-dimensional inverse Fourier transform:

Iðx; yÞ ¼ IFFT2D½sðkx ;kyÞ� [1]

The program then applies constant [wo] and linear [wx

and wy] phase corrections to the initial complex imageyielding phase-corrected images:

IPhaseCorrectedðx; yÞ ¼ Iðx; yÞeiðwoþwxxþwy yÞ [2]

where the linear phases correct for off-center echo posi-tion in two dimensions (readout and phase encoding)and the constant phase provides a global frequencyadjustment.

Water-only [IWATER] and fat-only [IFAT images are cal-culated using sign detection of the real part of the phase-corrected complex image according to:

IWATERðx; yÞ ¼jIPhaseCorrectedðx; yÞj real

�Iðx; yÞ

�> 0

0 real�

Iðx; yÞ�� 0

8>><>>:

[3]

IFATðx; yÞ ¼jIPhaseCorrectedðx; yÞj real

�Iðx; yÞ

�< 0

0 real�

Iðx; yÞ�� 0

8>><>>:

[4]

Final image reconstruction is performed by summationof images over the four coils.

The developed computer program has sliders control-ling constant and linear x and y phases with immediateimage display feedback for interactive user reconstruc-tion. A single user with 2 years of cardiovascular MRexperience processed the raw data, adjusting the threesliders corresponding to the three phase terms of Eq. [2]to provide the best water-fat separation for each coil forthe first cardiac phase of each imaging plane.

Additionally, an automated reconstruction algorithmwas developed to find optimal phases [wo,wx, and wy]which minimized the imaginary part of the image similarto a method described by Hargreaves et al. (15). Anunconstrained nonlinear minimization of the imaginaryterm of each pixel multiplied by its magnitude,[jIðx; yÞj � imagðIPhaseCorrectedðx; yÞÞ�], was used to reduceeffects of noisy pixels. The sum of image pixels wasminimized using the Nelder Mead nonlinear optimiza-tion algorithm. Variables of the automated method werethe starting point and size of minimization subimage rel-ative to the imaging FOV. Minimization was not per-formed over the full FOV, as fold-over artifacts arecommon in CINE imaging. Automated reconstructionswere performed with full (1/1) FOV, 3=4 FOV, and 2/3FOV with the subimage center in the center of the FOV.Individual and coil mean phase determined by the man-ual reconstruction were used as starting points for theoptimization algorithm. The three phase correction val-ues were recorded for each coil for both manual andautomated techniques.

Image Analysis

Water-fat separated images from the 10 healthy volun-teers were evaluated for artifacts characterized by falselyseparated tissues at the chest/back, in atrial and ventric-ular blood, myocardium, and outside of the myocardium(extracardiac) by a single blinded observer with over 10years of cardiovascular MR experience. A three pointscale was used: 0 ¼ no artifact, 1 ¼ minor artifact, and 2¼ severe artifact. Presence of water-fat separation arti-facts at the mitral and aortic valves was recorded using atwo point scale: 0 ¼ no artifact and 1 ¼ artifact present.

Statistical Analysis

Statistical analysis was performed using Matlab version7.11, (Mathworks, Natick, MA). Continuous variables aresummarized as mean 6 standard deviation. Categoricalvariables are presented as frequency or percentage. Man-ual and automated phase correction values with differentstarting points and minimization subimage sizes werecompared using multiway (n-way) analysis of variance(ANOVA) with cardiac phase, coil element, technique,and imaging plane modeled as fixed effects. Subjectswere modeled as random effects. Using this technique,we tested for differences in phase correction values

bSSFP CINE Water-Fat Separation Imaging 2097

between imaging planes, cardiac phases, coil elements,starting points, and subimage minimization size. We alsotested for differences in water-fat separations artifactscores between the automated and manual methodsusing the Kruskal-Wallis one-way ANOVA test. P-valueswere adjusted for multiple comparisons. A P-value <0.05 was regarded as statistically significant.

RESULTS

Normal Volunteer Studies

The water-fat separation results from a representativenormal volunteer are displayed in Figure 1. Before phasecorrection water-fat separation was unreliable and failedin three of the four imaging planes. With constant and

linear phase corrections, water-fat separation performedwell in the heart. In Figure 2, an example is given fromanother normal volunteer showing water-fat separationwith a moderate pericardial effusion and fat having simi-lar signal intensities. Adjacent fluid and fat are easilyidentified in the separated images.

Several image artifacts particular to water-fat separa-tion were identified (Figs. 3 and 4). Banding artifacts(16) common in bSSFP imaging due to off-resonanceoften caused false water-fat separation (Fig. 3). An apicalfalse separation artifact (Fig. 3: middle row) due tounknown origins was also noted (see also Figure 8).False separation artifacts with image fold-over due to afailure of the linear phase correction occurred whenimage fold-over was present (Fig. 3: bottom row). False

FIG. 1. Water-fat separation examples from a normal volunteer without phase correction (noPC) and with manual (ManPC) phase correc-tion. Displayed are the conventional (ORIG), water-only (WATER), and fat-only (FAT) images for the standard four-chamber (horizontallong axis), three-chamber (LVOT), two-chamber (vertical long axis), and mid short-axis (SA) imaging planes. Water-fat separation is unre-

liable without phase correction and performs well around the heart in all imaging planes with appropriate phase correction.

FIG. 2. Four-chamber images from anormal volunteer showing a mild peri-cardial effusion. Fluid around the heart

(small arrowheads) in the water-onlyimage is easily differentiated from epi-

cardial fat in the fat-only image. Largearrowhead shows cerebrospinal fluid inthe water-only image.

2098 Goldfarb and Arnold-Anteraper

FIG. 3. Example of false water-fat sepa-

ration artifacts. Top row: false separa-tion of chest wall fat (arrowheads). Notebanding artifacts due to off-resonance

in the original image. Middle row: falseseparation of chest wall fat and apical

epicardial fat (arrowheads). Bottom row:false separation of posterior fat due tofoldover aliasing artifacts. Phase cor-

rection fails due to the fat being on theincorrect (anterior) side of the imaging

field of view (arrowheads). There is alsoa false separation of a small region ofchest wall fat (large arrowhead). Note

the off-resonance banding artifact inthe original image.

FIG. 4. Example of blood water-fat sep-

aration artifacts. Three-chamber imagesfrom a healthy volunteer with mitralvalve prolapse at systole (top row) and

diastole (bottom row). Fast movingblood creates a false separation artifact

distal to the aortic and mitral valves(arrowheads).

bSSFP CINE Water-Fat Separation Imaging 2099

separation due to high flow at the aortic and mitral valeswas also noted. There were five (50%) false separationsat the aortic valve in the LVOT imaging plane and seven(23%) false separations at the mitral valve (1 in horizon-tal long axis (10%), 2 in LVOT (20%), and 4 in verticallong axis (40%)). Example images in the LVOT imagingplane are given in Figure 4. These artifacts were presentfor a few cardiac phases only when flow across the valvewas high and easily identified.

Manual and Automated Phase Correction Values

Both constant (p ¼ 0.8) and linear x (p ¼ 1) and y (p ¼1) phase correction values did not vary significantlyacross cardiac phases. All phase correction values variedsignificantly among the coils (p < 0.001) and imagingplanes (p < 0.001) (Fig. 5). One can clearly see the largedifference between anterior (red and green) and posterior(blue and black) coil elements. There was no significantdifference between horizontal long axis and LVOT imag-ing planes as they are similarly positioned, but larger dif-ferences were seen with short axis and vertical long axis

imaging planes. The starting points investigated in thisstudy and displayed in Figure 5 did not produce signifi-cantly different phase correction values. Changes in min-imization subimage size did not yield significantlydifferent constant phase correction values (p ¼ 0.2), butdid yield significantly different linear x (p < 0.001) andy (p < 0.001) phase values. Artifacts were most frequentin the chest/back. There was a significant difference inartifact score with phase correction, but not betweentested methods (Fig. 6).

Patient Studies

The patient with a prior myocardial infarction and fatdeposition showed a fatty lesion in water-fat separatedimages (Fig. 7). Fat in the inferior wall was easily identi-fied in separated images. CINE images (Supporting Infor-mation online video 1) show the motion of the fatthroughout the cardiac cycle. A right atrial mass isdepicted in Figure 8. The mass has the same signalintensity as atrial blood, but is well characterized bywater-fat separation. Its mobility is well visualized in

FIG. 5. Constant and linear phase corrections from manual (solid) automated with user manual starting point (dashed) and automated

with median coil starting point (dotted) for the each coil element and imaging plane. Coil element location: red ¼ anterior inferior, green¼ anterior superior, blue ¼ posterior inferior, and black ¼ posterior superior. Constant and linear phases were significantly differentbetween the anterior and posterior coils. Note the similarities and variances between imaging planes.

2100 Goldfarb and Arnold-Anteraper

Supporting Information online video 2. Artifacts arepresent in this study in the anterior chest wall and heart,but do not interfere with myocardial and blood water-fatseparation.

DISCUSSION

As previously shown in other part of the body (15), wehave shown that the resonant frequency dependence ofthe signal phase in standard bSSFP CINE images of theheart can be used to produce water and fat separatedimages. A zeroth- and first-order phase correction of con-ventional complex image data for each coil element wasnecessary to provide consistent water-fat separation.Simple image calculations such as phase multiplicationand sign detection were used in water-fat separationprocessing. Optimization of the phase correction valueswas performed with manual and more complex mathe-matical calculations.

Water-fat separation via resonant frequency depend-ence has a long history (17–20). The method describedin this paper differs from most methods and could prob-ably be termed water-fat identification as it does not pro-vide a water-fat fraction, but identifies image pixels with

a predominance of water or fat. Additionally, theapproach used in this paper uses a standard single-echoCINE acquisition rather than a multiecho approach(12,20). Recent research has concentrated on improvedpostprocessing methods of dedicated multiecho acquisi-tions (18–20). Alternatively, Hargreaves et al. (15)showed that water-fat separation information is alreadyincluded in conventional bSSFP acquisitions and mustonly be extracted, requiring no pulse sequence modifica-tions or additional complexities if certain requirementsare met. The main criterion is that the echo time is halfway between the RF excitation pulses (TR), which is acommon attribute of bSSFP sequences. TR should be lessthan the reciprocal of off resonance frequency differenceor TR < 4.5 ms at 1.5 T. We used a TR of 3.2 ms, com-monly used for cardiac CINE imaging to reduce bandingartifacts. Increases in TR may be tolerated, but at theexpense of additional banding artifacts.

In this pilot study, a basic CINE technique was used.This resulted in a somewhat low temporal resolution (50ms). Temporal resolution could be improved using k-space view sharing (21) and a parallel image acquisitionand reconstruction (22); two techniques commonly usedin clinical situations. The presented water-fat separation

FIG. 6. Qualitative image evaluation results. There was a significant difference between no phase correction and phase correctedimages for all areas evaluated (p < 0.001) and not a significant difference between evaluated phase correction methods. NoPC ¼ no

phase correction, ManPC ¼ user interactive, AutUser ¼ automated with user manual starting point, and AutMed ¼ automated withmedian coil starting point.

bSSFP CINE Water-Fat Separation Imaging 2101

technique should be compatible with these temporalresolution improvement methods and might benefit fromphase corrections from autocalibration data. A bSSFPacquisition was used, but a spoiled gradient-echo acqui-sition could also be used if an echo time was chosensuch that water and fat had opposed phases.

Although confidence in water-fat separation isimproved by robust performance over an extended FOV,immediate cardiac applications would be a quick or ret-rospective identification of fatty masses. Also, the tech-nique may find use in the assessment of arrhythmogenicright ventricular dysplasia/cardiomyopathy for thesimultaneous functional assessment and fat infiltrationassessment, which now is performed with multiple pulsesequences (23). Lastly, the technique could also be usedto improve confidence through increased contrastbetween fluid, fat, pericardium, and myocardium. Retro-spective water-fat separation is also possible providingthat the complex image or raw k-space data are available,allowing a physician to evaluate an unsuspected cardiac

mass with bSSFP CINE images without routine fat sensi-tive imaging (an real-world example is given in Figure 7).

The technique is mainly limited by B0 homogeneity.Also partial voluming of fat and water limits the sizeof fat that can be identified. This is not a limitation ofmultiecho techniques as they separated water and fatcontributions to individual pixels rather than identify-ing water and fat pixels. The main artifact of the tech-nique was false water-fat separation and furtheroptimization could be accomplished in the future. Thiscould possibly be resolved with dedicated patient andimage plane magnetic field shimming, which was notused in this study. Also low frequency phase correc-tions from the bSSFP data or additional field mapscould be used for phase correction. It seems natural toconsider the of use higher order phase correctionsterms (x2, y2, and xy), these do not have a clear theo-retical effect as do constant and linear term. Additionof higher order terms was explored, but did not have apositive effect. The artifact from fast moving blood

FIG. 7. a: Two chamber (top row)and short axis (bottom row)

images from a patient with aninferior wall chronic myocardial

infarction and mid-wall fat depo-sition (arrowheads). SupportingInformation Online Movie 1

shows good separation acrossthe entire cardiac cycle. Artifactsare seen in the liver due to off

resonance and in the chest walldue to sternal wires from a prior

coronary artery bypass graft(CABG) operation. b: Spoiled-gradient echo images without

(left) and with (right) chemical fatsuppression shows fat deposition

(arrowheads).

2102 Goldfarb and Arnold-Anteraper

warrants further study and has not yet been reportedin water-fat separation as many reports use dark bloodpreparation or gradient moment nulling for flow com-pensation. It may provide an additional tool for sensi-tivity to regurgitant flow jets or possibly be eliminatedwith additional flow compensation (24). Additionally,the on-resonance water center frequency in this workwas set to the center of the main spectral lobe result-ing in an asymmetry of water and fat resonancesaround the first null band. Improvements may be pos-sible if the null band is optimally shifted betweenwater and fat resonances by adjustment of the centerfrequency.

In conclusion, we have implemented a previouslyreported bSSFP water-fat separation postprocessing algo-rithm for cardiac CINE imaging. Zeroth- and first-orderphase correction was necessary to provide consistentresults in standard cardiac imaging planes. Artifacts ofthe technique include false water-fat separation due tomagnetic field inhomogeneities and fast moving blood atthe heart valves. The technique works with existing car-diac protocols and should provide an additional tool fortissue characterization in cardiac studies.

REFERENCES

1. Natori S, Lai S, Finn JP, Gomes AS, Hundley WG, Jerosch-Herold M,

Pearson G, Sinha S, Arai A, Lima JA, Bluemke DA. Cardiovascular

function in multi-ethnic study of atherosclerosis: normal values by

age, sex, and ethnicity. AJR Am J Roentgenol 2006;186:S357–S365.

2. Keenan NG, Pennell DJ. CMR of ventricular function. Echocardiogra-

phy 2007;24:185–193.

3. Fieno DS, Saouaf R, Thomson LE, Abidov A, Friedman JD, Berman

DS. Cardiovascular magnetic resonance of primary tumors of the

heart: a review. J Cardiovasc Magn Reson 2006;8:839–853.

4. Greenwood JP, Maredia N, Younger JF, et al. Cardiovascular magnetic

resonance and single-photon emission computed tomography for

diagnosis of coronary heart disease (CE-MARC): a prospective trial.

Lancet 2012;379:453–460.

5. Kim RJ, Wu E, Rafael A, Chen EL, Parker MA, Simonetti O, Klocke

FJ, Bonow RO, Judd RM. The use of contrast-enhanced magnetic reso-

nance imaging to identify reversible myocardial dysfunction. N Engl J

Med 2000;343:1445–1453.

6. Carr H. Steady-state free precession in nuclear magnetic resonance.

Phys Rev 1958;112:1693–1701.

7. Oppelt A. FISP-a new fast MRI sequence. Electromedica 1986;54:15–

18.

8. Carr JC, Simonetti O, Bundy J, Li D, Pereles S, Finn JP. Cine MR angi-

ography of the heart with segmented true fast imaging with steady-

state precession. Radiology 2001;219:828–834.

9. Haase A, Frahm J, Hanicke W, Matthaei D. 1H NMR chemical shift

selective (CHESS) imaging. Phys Med Biol 1985;30:341–344.

FIG. 8. a: Right ventricular inflow

tract (RVIT) and four-chamberimages of a patient with a rightatrial 3 cm tumor (large arrow-

head). The tumor is well depictedin the fat-only image and arisesfrom an area that shows fat

around the inferior vena cava.The mass is attached to the right

atrial wall and is mobile withinthe right atrial cavity (SupportingInformation Online Movie 2).

False separation artifacts arepresent in the chest wall and api-

cal epicardial fat (small arrow-heads). b: Dark blood turbo-spin-echo images without (left)

and with (right) chemical fat sup-pression shows a change in sig-

nal of the right atrial tumor(arrowheads).

bSSFP CINE Water-Fat Separation Imaging 2103

10. Bydder GM, Pennock JM, Steiner RE, Khenia S, Payne JA, Young IR.

The short TI inversion recovery sequence—an approach to MR imag-

ing of the abdomen. Magn Reson Imaging 1985;3:251–254.

11. Goldfarb JW. Targeted ROTational magnetic resonance angiography

(TROTA). Magn Reson Med 2007;58:448–453.

12. Reeder SB, Markl M, Yu H, Hellinger JC, Herfkens RJ, Pelc NJ. Car-

diac CINE imaging with IDEAL water-fat separation and steady-state

free precession. J Magn Reson Imaging 2005;22:44–52.

13. Goldfarb JW. Fat-water separated delayed hyperenhanced myocardial

infarct imaging. Magn Reson Med 2008;60:503–509.

14. Goldfarb JW, Roth M, Han J. Myocardial fat deposition after left ven-

tricular myocardial infarction: assessment by using MR water-fat sep-

aration imaging. Radiology 2009;253:65–73.

15. Hargreaves BA, Vasanawala SS, Nayak KS, Hu BS, Nishimura DG.

Fat-suppressed steady-state free precession imaging using phase

detection. Magn Reson Med 2003;50:210–213.

16. Scheffler K, Heid O, Hennig J. Magnetization preparation during the

steady state: fat-saturated 3D TrueFISP. Magn Reson Med 2001;45:

1075–1080.

17. Dixon WT. Simple proton spectroscopic imaging. Radiology 1984;

153:189–194.

18. Glover GH, Schneider E. Three-point Dixon technique for true water/

fat decomposition with B0 inhomogeneity correction. Magn Reson

Med 1991;18:371–383.

19. Ma J. Breath-hold water and fat imaging using a dual-echo two-point

dixon technique with an efficient and robust phase-correction algo-

rithm. Magn Reson Med 2004;52:415–419.

20. Reeder SB, McKenzie CA, Pineda AR, Yu H, Shimakawa A, Brau AC,

Hargreaves BA, Gold GE, Brittain JH. Water-fat separation with

IDEAL gradient-echo imaging. J Magn Reson Imaging 2007;25:644–

652.

21. Foo T, Bernstein MA, Aisen AM, Hernandez RJ, Collick BD,

Bernstein T. Improved ejection fraction and flow velocity estimates

with use of view sharing and uniform repetition time excitation with

fast cardiac techniques. Radiology 1995;195:471–478.

22. Griswold MA, Jakob PM, Heidemann RM, Nittka M, Jellus V, Wang J,

Kiefer B, Haase A. Generalized autocalibrating partially parallel

acquisitions (GRAPPA). Magn Reson Med 2002;47:1202–1210.

23. Tandri H, Macedo R, Calkins H, et al. Role of magnetic resonance

imaging in arrhythmogenic right ventricular dysplasia: insights from

the North American arrhythmogenic right ventricular dysplasia

(ARVD/C) study. Am Heart J 2008;155:147–153.

24. Bieri O, Scheffler K. Flow compensation in balanced SSFP sequen-

ces. Magn Reson Med 2005;54:901–907.

2104 Goldfarb and Arnold-Anteraper