Technical Note

Quantitative Assessment of Cardiac Output and LeftVentricular Function by Noninvasive Phase-Contrastand Cine MRI: Validation Study With InvasivePressure-Volume Loop Analysis in a Swine Model

Hung-Yu Lin, PhD,1,2* Darren Freed, MD, PhD,3 Trevor W.R. Lee, MD,3,4

Rakesh C. Arora, MD, PhD,3 Ayyaz Ali, MD,5 Waiel Almoustadi, MD,3 Bo Xiang, MD,1

Fei Wang, MD,1 Stephen Large, MD,5 Scott B. King, PhD,1 Boguslaw Tomanek, PhD,1

and Ganghong Tian, MD, PhD1

Purpose: To validate noninvasive cardiac output meas-urements of phase-contrast magnetic resonance imaging(PC-MRI) and cine MRI using an invasive pressure-volume(PV) loop technique on a swine model.

Materials and Methods: We compared three methods forevaluating cardiac function at rest and under pharmaceu-tical low-dose inotropic infusion conditions: 1) phase-contrast MRI, 2) cine MRI, and 3) PV loop relationship.These measurements were made in 14 domestic pigsunder rest conditions. Identical MRI acquisitions and PVloop analysis were performed on six pigs from the samegroup that received an infusion of dobutamine 2.5 mg/kg/min. Cardiac outputs from all measurements were an-alyzed and compared using linear regression and Bland–Altman analysis.

Results: Noninvasive PC-MRI and cine MRI did not showany significant differences compared to an invasive PVloop technique for measurement of cardiac output underboth rest (PC-MRI, cine MRI, and PV loop, 3.17 6 0.45,3.18 6 0.61, 3.45 6 0.41 L/min, respectively) and phar-maceutical low-dose inotropic infusion conditions (PC-MRI, cine MRI, and PV loop, 4.78 6 0.53, 4.7 6 0.6, 4.966 0.48 L/min, respectively). Statistical analysis showed

good agreement of cardiac output measurements at rest(R2 ¼ 0.83) and under low-dose inotropic infusion condi-tions (R2 ¼ 0.74) using PC-MRI and PV loop techniques.Cardiac output measurement using cine MRI and PV looptechniques also showed good agreement at rest (R2 ¼0.85) and under low-dose inotropic infusion conditions(R2 ¼ 0.76). Furthermore, cardiac outputs determinedwith the three modalities showed good agreement over awide range of heart rates (90–180 bpm).

Conclusion: MRI can provide a reliable, noninvasive mea-surement of cardiac output that can be carried out with-out the complications that are inherent with current inva-sive procedures.

Key Words: phase-contrast MRI; cine MRI; pressure-vol-ume loop analysis; cardiac output; ventricular functionJ. Magn. Reson. Imaging 2011; 34:203–210.VC 2011 Wiley-Liss, Inc.

CONGESTIVE HEART FAILURE is a cardiac syndromethat occurs due to a severe decrease of left ventricularcontractility (1). The condition of the heart can bedetermined by its level of contractility to circulatingblood at an adequate rate within a body (2). Severalmethods have been developed for assessing ventricu-lar contractility to help early diagnosis and manage-ment of patients with heart disease. Existing techni-ques for assessment of left ventricular function areclassified into two major categories: invasive (ie,direct) measurements using a conductance catheter(3,4) and noninvasive (ie, indirect) estimations thatuse imaging techniques (5,6). Previous work hasshown that analysis of the time-varying ratio of in-stantaneous pressure (P) to volume (V) obtained froman indwelling conductance catheter yields accurateassessment of intrinsic ventricular diastolic and sys-tolic properties independently of loading conditions(7,8). Therefore, dynamic PV loop analysis has beenconsidered the gold standard in assessment of cardiac

1Institute for Biodiagnostics, National Research Council Canada,Winnipeg, MB Canada.2Department of Radiology, University of Manitoba, Winnipeg, MBCanada.3Cardiac Sciences Program, St. Boniface Hospital, Winnipeg, MBCanada.4Department of Anesthesia and Perioperative Medicine, St. BonifaceHospital, Winnipeg, MB Canada.5Department of Cardiothoracic Surgery, Papworth Hospital,Cambridge, United Kingdom.

Contract grant sponsor: Manitoba Medical Service Foundation(MMSF).

*Address reprint requests to: H.-Y.L., Institute for Biodiagnostics,National Research Council Canada, 435 Ellice Ave., Winnipeg, MBR3B 1Y6 Canada. E-mail: Hung-Yu.Lin@nrc-cnrc.gc.ca

Received September 15, 2010; Accepted March 4, 2011.

DOI 10.1002/jmri.22587View this article online at wileyonlinelibrary.com.

JOURNAL OF MAGNETIC RESONANCE IMAGING 34:203–210 (2011)

CME

VC 2011 Wiley-Liss, Inc. 203

conditions (9,10). Several load-independent indicesobtained from PV measurements are commonly usedto evaluate contractility of the left ventricle (LV),including slope of the end-systolic PV relationship,slope of the maximum first derivative of LV pressure(dP/dtmax), and preload-recruitable stroke work (11–13). These parameters have been extensively used tocharacterize the heart function under various patho-logical situations (14) and more recently for evaluationof diastolic LV stiffness (15). However, PV loop analy-sis requires an invasive procedure that is associatedwith a variety of complications and makes it difficultto perform routinely on patients.

Blood flow measurement using phase-contrast mag-netic resonance imaging (PC-MRI) has recently beendemonstrated as a noninvasive alternative for mea-surement of blood flow in a wide spectrum of cardio-vascular pathologies, including valvular disease andcongenital heart disease. The technique is particularlysensitive to valvular regurgitation (16,17), aortic valvestenosis (18,19), and shunt flow associated with atrialand ventricular septal defects (20,21). PC-MRI pro-vides unique advantages for noninvasive diagnosis asa result of the absence of ionizing radiation exposure,high reproducibility, and the ability to perform inte-grated assessments with cardiac MRI protocols thatprovide both anatomic and functional evaluations forpatients. Alternatively, fast steady-state gradient cinetechniques have been widely used in the clinic to eval-uate stroke volumes and cardiac outputs through ac-quisition of short-axial cine images of the LV (22,23).The assessment of LV volumes and cardiac functionusing PC-MRI and cine MRI techniques are promisingand widely used in the clinic. To date, the accuracy ofthe cardiac MRI technique has not been validated bycomparison with traditional invasive PV loop measure-ments. To our knowledge, the cardiac output obtainedfrom noninvasive PC-MRI, cine MRI with and withoutdobutamine low-dose inotropic infusion has not beenvalidated against the invasive PV loop technique. Theaim of our study was to determine the accuracy andreliability of PC-MRI and cine MRI for measurement ofcardiac output by comparing the results obtained fromthese methods to those yielded by the current goldstandard (ie, PV loop). Cardiac function was assessedusing PC-MRI, cine MRI, and PV loop techniques in pighearts under both resting and dobutamine low-doseinotropic infusion conditions.

MATERIALS AND METHODS

The project was designed to validate the correlation ofnoninvasive cardiac MRI examinations and invasive PVloop measurements through the following studies: PC-MRI measurements of blood flow at the descendingaorta; comparison of multiple segmented short-axis cineimaging of the LV; and comparison of the LV pressure-volume relation acquired using a conductance catheter.

Animal Preparation

Cardiac PC-MRI, cine MRI scans and invasive PV loopmeasurements were performed on 14 domestic pigs (n

¼ 14) weighing 45–50 kg. A subgroup of six pigs (n ¼6) from the same animal group was given a continu-ous intravenous infusion of dobutamine (Sandoz Can-ada, Boucherville, QC, Canada) to evaluate the agree-ment between cardiac output measurements usingthe three methods under simulated exercise-stressconditions at a heart rate of �180 beats per minute.All pigs were anesthetized with 2% isoflurane beforeintravascular intervention and MRI examinations. At-ropine (0.05 mg/kg body wt) was given intramuscu-larly 30–50 minutes prior to the experiment. Anesthe-sia was induced with an intramuscular injection ofmidazolam (0.33 mg/kg body wt) and ketamine (0.22mg/kg body wt). Isoflurane was initially given by amask at a level of 4%–5% in 100% O2 and maintainedat 1.5�3%. The depth of anesthesia was controlled atthe level at which the animal did not show spontane-ous breath or muscular response to surgical proce-dures. After endotracheal intubation, the pigs wereventilated mechanically with 60% oxygen and 40%air. The ventilator rate and tidal volume were adjustedto maintain arterial CO2 tension between 35 and 45mmHg at normal temperature. Oxygenation, electro-cardiogram, and temperature were monitored andrecorded continuously. Heparin was given throughoutthe experiment. For the low-dose inotropic infusionMRI studies, dobutamine was infused intravenouslyat a rate of 2.5 mg/kg/min over 10 minutes to main-tain a targeted heart rate of �180 beats per minute.The pigs received care in compliance with the Guide tothe Care and Use of Experimental Animals publishedby the Canadian Council on Animal Care (Ottawa,ON, 1993).

Invasive PV Loop Assessments

The left carotid artery was isolated and instrumentedwith a 5F 7-electrode conductance catheter (MillarInstruments, Houston, TX) advanced into the LV toperform in vivo PV loop measurements. A 2-inch inci-sion was made on the abdomen for access to the infe-rior vena cava to control cardiac preload. Brief occlu-sion and release of the inferior vena cava (3–5seconds) were performed on pigs to reduce the pre-load for invasive PV loop measurements. All physio-logic parameters (LV pressure, LV volume, heart rate,and peak rate pressure development, dP/dtmax) weremonitored and recorded using a commercial PV loopsystem (MPVS Ultra system, Millar Instruments).

Noninvasive MRI Assessments

All imaging experiments were performed on a 3.0 Tscanner (Trio, Siemens Healthcare, Erlangen, Ger-many) with spine, cardiac phased-array coils, andprospective electrocardiography-signal gating.Through-plane blood flow measurements using a seg-mented gradient-echo PC-MRI sequence were used onan imaging plane perpendicular to subjects’ descend-ing aorta. The acquisition parameters of PC-MRI were:image matrix ¼ 256 � 144, flip angle ¼ 20�, spatialresolution ¼ 2.18 � 2.18 mm2, parallel accelerationrate ¼ 2, velocity encoding (VENC) ¼ 150 cm/sec, and

204 Lin et al.

TE/TR/temporal resolution ¼ 2.6/7.1/42.0 msec.Quantitative evaluation of cardiac output using MRIwas done by multiplying stroke volume (ie, the areaunder flow curve within one cardiac cycle) and thecorresponding heart rate of the animal. Alternate MRImeasurements of stroke volume and cardiac outputusing a segmented cine spoiled gradient-echosequence were performed in contiguous short-axislocations, parallel to the tricuspid valve annulus,spanning the LV from base to apex. Imaging parame-ters were as follows: image matrix ¼ 196 � 108, flipangle ¼ 12�, spatial resolution ¼ 2.84 � 2.9 mm2,

parallel acceleration rate ¼ 2, and TE/TR/temporalresolution ¼ 2.6/6.8/34 msec. Nine to twelve sliceswere acquired, with a slice thickness of 8 mm and2-mm gap between slices on short axis images. Thevalues of cardiac output were calculated using offlinecardiac analysis software (Argus, Siemens Healthcare)and standard cardiac analysis (24).

Statistical Analysis

Paired t-tests were used to compare stroke volumeand cardiac output between 1) PC-MRI and PV loop

Figure 1. Representative plot of dynamic LV (a) volume vs. time, (b) pressure vs. time curve acquired using a PV conduct-ance catheter under rest conditions. (c) ECG signal vs. time, and (d) beat-to-beat LV pressure vs. volume loop acquired usinga PV conductance catheter placed in the LV. The slopes of the EDPVR and the ESPVR reflect myocardial contractility.

Cardiac Output Validation 205

analysis and 2) cine MRI and PV loop analysis in thegroup of pigs. A value of P < 0.05 was consideredstatistically significant. All parameters are expressedas the mean 6 standard deviation (SD). Statisticalanalysis by the Bland–Altman method (25) was usedto evaluate the agreement between different measure-ment techniques. Agreement between the methodswas assessed by calculating the paired differencebetween two methods for each measurement and byestimating the bias and 95% limits of agreement rela-tive to the mean values of both methods.

RESULTS

MRI and PV Studies at Rest Conditions

Representative instantaneous volume and pressuremeasurements of LV from one sample of animal stud-ies under rest conditions are shown in Fig. 1a,b,respectively. Figure 1c shows a time-resolved electro-cardiogram (ECG) signal during the PV loop analysis.Representative beat-to-beat examinations of thepressure and volume relation for monitoring heartfunction are shown in Fig. 1d. End-systolic pressure-

Figure 2. (a) Time-resolvedaortic flow velocity at the de-scending aorta obtained fromPC-MRI under rest conditions.Cine MRI series of magnitudeimages acquired with ak-space segmented gradientecho sequence. (b,c) Acquiredusing cine MRI at end-systoleand end-diastole in the samesubject, respectively.

Figure 3. Bland–Altman anal-ysis of agreement in cardiacoutput under rest conditionsusing (a) PC-MRI and PV loop,(b) MRI cine and PV loop,respectively. The difference isplotted as a function of theaverage of the result from thetwo methods. The lines in thegraphs indicate the upper andlower limits of agreement,mean 62 SD.

206 Lin et al.

volume relations (ESPVR) and end-diastolic pressure-volume relation (EDPVR) indicate moderate myocar-dial contraction and complete diastolic relaxationunder rest conditions. Figure 2a shows the change inaortic flow with time over a single cardiac cycle usingthe PC-MRI technique. Frames taken from the cineMRI loop at end-systole (Fig. 2b) and end-diastole(Fig. 2c) are consistent with the PV loop relation (Fig.1d). The mean cardiac output was calculated to be3.17 6 0.45 L/min by PC-MRI, 3.18 6 0.61 L/min bycine MRI, and 3.45 6 0.41 L/min by PV measure-

ments. The cardiac output measurements obtainedfrom each subject using PC-MRI and PV loop agreed(y ¼ 0.84x þ 0.6, R2 ¼ 0.83) with a mean difference(bias) of 0.1 L/min of cardiac output or 3.2% (Fig. 3a).The upper and lower limits of agreement (bias 6 2SDs of the difference) were 0.27 and �0.44. Goodagreement was also observed between cine MRI resultand PV loop analysis (y ¼ 0.62x þ 1.3, R2 ¼ 0.85)with a mean difference of �0.05 L/min in cardiac out-put; the upper and lower limits were 0.51 and �0.61,respectively (Fig. 3b).

Figure 4. Representative plot of dynamic LV (a) volume vs. time, (b) pressure vs. time curve acquired using a PV conduct-ance catheter under stress conditions. (c) ECG signal vs. time, and (d) beat-to-beat LV pressure vs. volume loop acquiringusing a PV conductance catheter placed in the LV. The slopes of the EDPVR and the ESPVR reflect myocardial contractility.[Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Cardiac Output Validation 207

MRI and PV Studies Under Dobutamine Low-DoseInotropic Infusion Conditions

Representative measurements of cardiac functionobtained with the PV loop technique under dobut-amine low-dose inotropic infusion are displayed inFig. 4. Instantaneous volume (Fig. 4a), pressure (Fig.4b), and ECG signal (Fig. 4c) show an elevation inheart rate due to pharmaceutical stress, relative tothe control resting stage (Fig. 1c). ESPVR and EDPVRindicate increased myocardial contractility and lack ofcomplete relaxation during diastole under the dobut-amine low-dose inotropic infusion (Fig. 4d). Aorticflow, end-systolic, and end-diastolic short-axis imagesacquired in the segmented gradient echo sequence fol-

lowing dobutamine injection are shown in Fig. 5a–c,respectively. Bland–Altman analysis (Fig. 6) indicatedthat PC-MRI and cine MRI both were in good agree-ment with PV loop measurements of cardiac outputs.Cardiac output obtained for each subject using PC-MRI and PV loop agreed (y ¼ 0.77x þ 1.27, R2 ¼ 0.74)with a mean difference (bias) of �0.18 L/min in car-diac output or 3.8% (Fig. 6a). The limits of agreement(62 SDs) between the two techniques were 0.36 and�0.73. Good agreement was also observed betweenresults obtained from cine MRI result and P-V loopanalysis (y ¼ 0.69x þ 1.7, R2 ¼ 0.76), with a meandifference of �0.26 L/min in cardiac output, andupper and lower limits of 0.33 and �0.85, respectively(Fig. 6b). No statistically significant differences were

Figure 5. (a) Time-resolvedaortic flow velocity at the de-scending aorta obtaining fromPC-MRI under stress condi-tions. Cine MRI series of mag-nitude images acquired with ak-space segmented gradientecho sequence. (b,c) Acquiredby cine MRI at end-systoleand end-diastole in the samesubject, respectively.

Figure 6. Bland–Altman anal-ysis of agreement in cardiacoutput (CO) under stress con-ditions using (a) PC-MRI andPV loop, (b) MRI cine and PVloop, respectively. The differ-ence is plotted as a functionof the average of the resultfrom two methods. The linesin the graphs indicate theupper and lower limits ofagreement, mean 62 SD.

208 Lin et al.

found in measurements in cardiac outputs betweenPC-MRI and PV loop measurements, or between cineMRI and PV loop methods.

DISCUSSION

We evaluated the agreement of cardiac output meas-urements derived from noninvasive MRI methodolo-gies relative to the gold standard invasive PV mea-surement technique using a porcine model (n ¼ 14under rest conditions, n ¼ 6 under low-dose inotropicinfusion conditions, respectively). The main findingsof this study are: 1) Cardiac outputs at rest assessedwith noninvasive PC-MRI and cine MRI are in goodagreement with those obtained from invasive PV loopmeasurements, as shown in Figs. 3 and 6. 2) Cardiacoutput under dobutamine-induced low-dose inotropicinfusion was slightly underestimated (�8%) using theMRI methodologies relative to PV loop analysis. Theresults of this study demonstrate that both PC-MRIand cine MRI are reliable techniques for assessmentof cardiac function under rest conditions. Under car-diac low-dose inotropic infusion conditions, the twoMRI techniques could slightly underestimate cardiacfunction when compared with the invasive PV looptechnique.

Assessment of cardiac output is essential in thetreatment of patients with suspected myocardial in-farction and heart failure. The PV conductance cathe-ter intervention provides the most direct and accuratemeasurements of pressure, volume, heart rate, andventricular contractility, and is considered the goldstandard for assessment of cardiac contractile func-tion. This methodology can, however, be associatedwith significant procedural complications, therebylimiting its clinical application in patients. Recentdevelopments in noninvasive medical imaging tech-nology have provided an attractive alternative toquantify cardiac function in clinical cardiovascularsettings. MRI is the best example of the noninvasiveimaging modalities that are able to provide compre-hensive evaluation of cardiac anatomy and function.Although noninvasive imaging technologies are widelyused in the clinic, their accuracy has never beendetermined relative to the gold standard invasive tech-nique. Our study was carried out to validate the non-invasive MRI techniques (PC-MRI and cine-MRI). Do-butamine was used as an inotropic agent in order toassess cardiac function under a wide range of heartrates.

Potential errors in calculating cardiac output fromPC-MRI arise mainly from improper correction ofbackground phase offsets. To address this issue, amulticenter, multivendor clinical trial of the accuracyof MRI flow measurements was recently conductedand reported (26). The results suggest that back-ground phase offsets contribute significant measure-ment errors to MRI flow analysis, and consequentlythe applicability of PC-MRI remains in question forsome cardiac diagnoses in clinic. In our study, an av-erage of 3%–8% measurement bias in cardiac outputwas observed using the phase-contrast technique,

which might explain the small offset in cardiac outputresult observed using PC-MRI compared to cine andPV loop techniques.

PC-MRI and cine MRI scans are typically performedwith limited spatial resolution in order to preserve suf-ficient temporal sampling rate and scan efficiency. Oneof the potential phase-velocity errors is caused by par-tial volume effects from mixing moving and static spinswithin a pixel. As a result, PC-MRI flow measurementhas a potential to slightly underestimate actual flowvolume due to averaged phase-velocity near vesselwall. However, PC-MRI only needs semiautomatic seg-mentation on single slice rather than the multipleshort-axial slices cine technique. Thus, accuracy andpostprocessing time of cardiac output evaluation canbe improved by reducing manual segmentation tasksin the daily clinical environment. Furthermore, thenumber of patient’s breathholds will significantly bereduced because of single-slice PC-MRI acquisition.The partial volume effect can also lead to miscalcula-tion of a patient’s end-systolic volume and end-dia-stolic volume in cine MRI, eventually causing inaccu-rate stroke volume and cardiac output. Advancedparallel imaging reconstruction could increase thedata matrix of MRI acquisitions by reconstructing finepixel resolution without loss of temporal resolution.The gain in spatial resolution of reconstructed imagespreserves in-plane spatial resolution, and minimizesthe measurement error of the partial volume artifact insmall vessels. Parallel imaging might be used to extendcurrent study to measure beat-to-beat (ie, real time)cardiac output with sufficient in-plane resolution andavoid possible overestimation of flow resulting frompartial volume effects (27).

In conclusion, our study showed good agreementbetween two MR techniques (ie, PC-MRI and cine MRI)and gold standard invasive PV loop analysis for assess-ment of LV function. The result in this study suggeststhat MRI offers a reliable, noninvasive method forassessment of cardiac function. Future work will aimat validating beat-to-beat monitoring of cardiac func-tions monitoring in response to nitric oxide inhalationusing real-time PC-MRI and cine imaging, which willbe compared with PV loop measurements.

REFERENCES

1. Alpert NR, Mulieri LA, Warshaw D. The failing human heart. Car-diovasc Res 2002;54:1–10.

2. Parmley WW. Pathophysiology of congestive heart failure. Am JCardiol 1985;56:7A–11A.

3. Nielsen JM, Kristiansen SB, Ringgaard S, et al. Left ventricularvolume measurement in mice by conductance catheter: evalua-tion and optimization of calibration. Am J Physiol Heart CircPhysiol 2007;293:H534–H540.

4. Pacher P, Mabley JG, Liaudet L, et al. Left ventricular pressure-volume relationship in a rat model of advanced aging-associatedheart failure. Am J Physiol Heart Circ Physiol 2004;287:H2132–H2137.

5. Bombardini T, Agrusta M, Natsvlishvili N, et al. Noninvasiveassessment of left ventricular contractility by pacemaker stressechocardiography. Eur J Heart Fail 2005;7:173–181.

6. Sandstede JJ, Lipke C, Kenn W, Beer M, Pabst T, Hahn D. CineMR imaging after myocardial infarction—assessment and follow-up of regional and global left ventricular function. Int J CardImaging 1999;15:435–440.

Cardiac Output Validation 209

7. Sagawa K. The end-systolic pressure-volume relation of the ven-tricle: definition, modifications and clinical use. Circulation1981;63:1223–1227.

8. Shoucri RM. The pressure-volume relation in the left ventricle andthe pump function of the heart. Ann Biomed Eng 1991;19:699–721.

9. Thompson RB, van den Bos EJ, Esposito DJ, Owen CH, GlowerDD. The effects of acute afterload change on systolic ventricularfunction in conscious dogs with normal vs. failing hearts. Eur JHeart Fail 2003;5:741–749.

10. Miyano H, Shishido T, Kawada T, et al. Acute effect of tumor ne-crosis factor-alpha is minimal on mechanics but significant onenergetics in blood-perfused canine left ventricles. Crit Care Med1999;27:168–176.

11. Aghajani E, Muller S, Kjorstad KE, et al. The pressure-volumeloop revisited: Is the search for a cardiac contractility index a fu-tile cycle? Shock 2006;25:370–376.

12. Faber MJ, Dalinghaus M, Lankhuizen IM, et al. Right and leftventricular function after chronic pulmonary artery banding inrats assessed with biventricular pressure-volume loops. Am JPhysiol Heart Circ Physiol 2006;291:H1580–H1586.

13. Joho S, Ishizaka S, Sievers R, Foster E, Simpson PC, GrossmanW. Left ventricular pressure-volume relationship in consciousmice. Am J Physiol Heart Circ Physiol 2007;292:H369–H377.

14. Penicka M, Bartunek J, Trakalova H, et al. Heart failure with pre-served ejection fraction in outpatients with unexplained dyspnea:a pressure-volume loop analysis. J Am Coll Cardiol 2010;55:1701–1710.

15. Westermann D, Kasner M, Steendijk P, et al. Role of left ventricu-lar stiffness in heart failure with normal ejection fraction. Circu-lation 2008;117:2051–2060.

16. Dulce MC, Mostbeck GH, Osullivan M, Cheitlin M, Caputo GR,Higgins CB. Severity of aortic regurgitation — interstudy reprodu-cibility of measurements with velocity-encoded cine MR-imaging.Radiology 1992;185:235–240.

17. Fujita N, Chazouilleres AF, Hartiala JJ, et al. Quantification ofmitral regurgitation by velocity-encoded cine nuclear magneticresonance imaging. J Am Coll Cardiol 1994;23:951–958.

18. Caruthers SD, Lin SJ, Brown P, et al. Practical value of cardiacmagnetic resonance imaging for clinical quantification of aorticvalve stenosis comparison with echocardiography. Circulation2003;108:2236–2243.

19. Sondergaard L, Stahlberg F, Thomsen C, Stensgaard A, LindvigK, Henriksen O. Accuracy and precision of MR velocity mappingin measurement of stenotic cross-sectional area, flow-rate, andpressure-gradient. J Magn Reson Imaging 1993;3:433–437.

20. Beerbaum P, Korperich H, Gieseke J, Barth P, Peuster M, MeyerH. Rapid left-to-right shunt quantification in children by phase-contrast magnetic resonance imaging combined with sensitivityencoding (SENSE). Circulation 2003;108:1355–1361.

21. Hundley WG, Li HF, Willard JE, et al. Magnetic resonance imag-ing assessment of the severity of mitral regurgitation. Compari-son with invasive techniques. Circulation 1995;92:1151–1158.

22. Moon JC, Lorenz CH, Francis JM, Smith GC, Pennell DJ. Breath-hold FLASH and FISP cardiovascular MR imaging: left ventricularvolume differences and reproducibility. Radiology 2002;223:789–797.

23. Myerson SG, Bellenger NG, Pennell DJ. Assessment of left ven-tricular mass by cardiovascular magnetic resonance. Hyperten-sion 2002;39:750–755.

24. Hudsmith LE, Petersen SE, Francis JM, Robson MD, NeubauerS. Normal human left and right ventricular and left atrial dimen-sions using steady state free precession magnetic resonanceimaging. J Cardiovasc Magn Reson 2005;7:775–782.

25. Bland JM, Altman DG. Statistical methods for assessing agree-ment between two methods of clinical measurement. The Lancet1986;327:307–310.

26. Gatehouse PD, Rolf MP, Graves MJ, et al. Flow measurement bycardiovascular magnetic resonance: a multi-centre multi-vendorstudy of background phase offset errors that can compromise theaccuracy of derived regurgitant or shunt flow measurements. JCardiovasc Magn Reson 2010;12:5.

27. Tang C, Blatter DD, Parker DL. Accuracy of phase-contrast flowmeasurements in the presence of partial-volume effects. J MagnReson Imaging 1993;3:377–385.

210 Lin et al.