Interstudy reproducibility of biplane cine nuclear magnetic resonance measurements of left ventricular function

METHODS

Interstudy Reproducibility of Biplane Cine Nuclear Magnetic Resonance Measurements of Left Ventricular Function Halima Benjelloun, MD, Gregory B. Cranney, MB, BS, Katharine A. Kirk, PhD, Gerald G. Blackwell, MD, Chaim S. Lotan, MD, and Gerald M. Pohost, MD Cine nuclear magnetic resonance (NMR) imaging, as a noninvasive and high-resolution imaging modality, has been shown to be reliable for determining absolute left ventricular (LV) volumes and ejection fraction. A relatively new gradient echo tine NMR approach using 2 orthogonal long-axis planes (2- and 4-chamber) aligned with the true axes of the left ventricle has been previously developed and validated against radiographic biplane LV cineangiography. The aim of the present investigation was to determine the reproducibility of this more rapid tine NMR approach for the measurement of LV volumes and ejection fraction. Eighteen normal subjects underwent 2 tine NMR studies, on different days, using a 1.5~tesla clinical imaging system. Studies were analyzed on-line and blindly by 2 independent observers. lntraobserver error was also determined in a blinded manner. Mean values of measurements determined by this method in this group of normal subjects were end-diastolic volume (120 f 20 ml), endsystolic volume (39 f 9 ml) and ejection fraction (67 f 4%). Paired analysis of data revealed no significant bias between interstudy, interobserver or intraobserver measurements, except for interobserver end-diastolic volume, where the first observer measurements were slightly elevated (5.6 f 7.8 ml) compared with the second. This resulted in a small difference in ejection fraction (1.7 f 2.3%) between observers. The absolute variation between measurements (square root of variance components) was low From the Division of Cardiovascular Disease, Department of Medicine, University of Alabama at Birmingham, Birmingham, Alabama. Dr. Benjelloun (Department of Cardiology, University Mohammed V, CHU Avicenne, Rabat, Morocco) and Dr. Lotan (the Department of Cardiology, Hadassah Medical Center, Jerusalem, Israel) are recipients of the International Fulbright Scholarship, Washington, D.C. Manuscript received November 20, 1990; revised manuscript received and accepted February 21, 199 1. Address for reprints: Gerald M. Pohost, MD, Division of Cardiovascular Disease, University of Alabama at Birmingham, Birmingham, Alabama 35294.

foi all interstudy, interobserver and intraobserver comparisons: end-diastolic volume was <&6.7 ml, end-systolic volume C&3.5 ml and ejection fraction <&2.4%. These results demonstrate the potential of Ithe biplalre tine NMR intrinsic long-axis approach to obtaih reliable and sequential measurements of LV volumes and function in subjects without segmental wall motion abnormalities, using practical acquisition and analysis times. (Am J Cardiol 1991;67:1413-1420)

ssessmentof left ventricular (LV) volumes and ejection fraction is important for the clinical assessmentof cardiac disease and for determining prognosis.1-3Furthermore, often it is only by measuring serial changesin parameters of LV function that the effects of surgical and medical interventions can be evaluated.2,3Spin echo nuclear magnetic resonance (NMR) imaging4-7 and, more recently, gradient echo tine NMR, a relatively new noninvasive imaging method with excellent spatial and temporal resolution, have been widely described and are suggestedto be of clinical value for the assessmentof LV functions-l4 and mass.15-17Early tine NMR methods used mainly the transaxial approach, with imaging planes perpendicular to the external body axis.*m1°These studies were timeconsuming to acquire and analyze, were not aligned with the intrinsic axes of the heart and were unfamiliar to cardiologists trained in conventional methods. Recently, a more practical tine NMR approach, using a biplane long-axis acquisition strategy, for measurement of LV volumes has been developedand validated against radiographic ventricular cineangiography.14This approach has also been shown to be useful for assessingglobal and regional wall motion. l8 The imaging strategy uses planes aligned with the intrinsic axes of the heart and has the advantage of obtaining views comparable to radiographic ventricular cineangiography and echocardiography. Furthermore, acquisition and analysis times are shorter than the previous transaxial or short-axis approaches. The aim of the

A

REPRODUCIBILITY OF CINE NMR MEASUREMENTS

14113

cordingly, 2 long-axis tine NMR planes (Figure 1) were acquired, as previously described14:(1) 2-chamber plane-a single angulated plane parallel to the interventricular septum intersecting the apex and the midmitral valve, and (2) 4-chamber plane-a double-angulated plane perpendicular to the interventricular septum and orthogonal to the 2-chamber plane. METHODS Cine NMR acquisition parameters were: pulse flip Patients: The study group consisted of 20 normal volunteers who were asymptomatic and had normal angle 45 to 50°, echo time 12 to 14 ms, 4 measurephysical examinations and 12-lead electrocardiograms ments, 16 phases,and slice thickness of 8 mm. Longat rest. No history of cardiac diseasewas recorded. In- axis images were acquired with 128 X 128 matrix resoformed consent was obtained from all subjects before lution interpolated to 256 X 256 during reconstruction. NMR imaging. Subsequently,2 subjectswere excluded Repetition times of 30 to 40 ms and acquisition times after their first scan due to severeimage artifacts pre- (6 to 10 min/scan) varied depending on the heart rate cluding analysis. These subjects were anxious and had (range 48 to 88 beats/min). Total study time, includpoorer quality images due to excessiverespiratory mo- ing the initial scout views, was approximately 25 to 30 tion. Therefore, the final study group comprised 18 nor- minutes. mal volunteers (8 men) aged between 19 and 69 years Cine nuclear inagnetic resohance imlige analysis: (mean f standard deviation 41 f 15). All subjects All image analysis was performed on-line, using softwere studied on 2 separate occasionsby NMR with a ware available on the imaging system. NMR images l- to 30-day interval between studies. were initially displayed in a tine ioop for viewing. The Nuclear magnetic resonance imaging: Studies were end-diastolic frame was the first image of the seriesobperformed on a 1.5~teslaimaging system with cardiac tained immediately (12 ms) after the R wave and beanalysis software (Philips Gyroscan, Shelton, Connecti- fore the beginning of mechanical contraction. The endcut). Spin echo scout images were used to determine systolic frame was chosen as the frame just before mithe orientation of gradient echo tine NMR planes. Ac- tral valve opening, when the LV cavity is smallest. The present study was to determine the reproducibility of this new approach for measuring LV volumes and ejection fraction in subjects with normal wall motion, where day-to-day variation due to biologic factors is minimized.

BIPLANE NMR

FIG& 1. Two-chamber and orthogonal 4-chainBer long-axis nuclear magnetic respmce (NhnR) views in both diastoie and syktote. The k$t ventrick is planbnetored by the operator, and the long ?I+ is dettnsd b.y a line joirdng the apex and midmitral valve. (From Cranney et al.13 Reproduced by permission of the American Heart Association, Im., and the authors.)

Diastole 1414

THE AMERICAN JOURNAL OF CARDIOLOGY VOLUME 67

JUNE 15, 1991

LV cavity area and length were determined by using a tracker ball (Figure 1). The mitral valve plane was defined by a straight line joining the hinge points of the mitral leaflets, which had previously been identified on the tine display. The length of the ventricle is determined from a line joining the apex to the midpoint of the mitral valve plane. Long-axis NMR volumes were calculated by using the Sandler-Dodge biplane formula, which is used for LV volume determination by contrast ventriculography and which corrects for foreshortening of 1 of the views when present.19In NMR imaging, dimensional correction factors are not needed,as the calibration is regularly checked with standard phantoms and rarely requires adjustment. Therefore, the equation simplifies to: volume = (8 X A2 X Ad )/(3 X 7r X L,i,), where AZ is the 2-chamber area, A4 is the 4-chamber area and Lk,, is the length of the shorter long axis. Two observers performed all measurements. Each observer was blinded to the results of the other. For assessmentof intraobserver variability, one observerreanalyzed 18 studies randomly selectedfrom the total of 36 studies available. This secondmeasurementwas obtained, after an interval of 23 months, again without knowledge of the original measurements. The mean analysis time for each study was approximately 10 minutes. &atistical methods: Interstudy, interobserver and intraobserver variability were assessedby using paired t tests to addresspotential systematic bias from study to study or from observer to observer. Estimation of variance componentswas used to estimate the amount of variance in a random measurement that is attributable to study or to observer differences. The results of the latter method are reported in 2 different ways: (1) as the intraclass correlation coefficient measuring the strength of agreement between measures, and (2) as the raw amount of uncertainty (reported in the measurement units) attributable to each source of variation. This method avoids the commonplace pitfall of using simple linear regression (and the Pearson correlation coefficient), which does not measure reliability or agreement, but which, assuming linearity, measuresa general trend. Also, absolute differencesbetween measurementswere calculated to assess variation and to allow comparison with other imaging modality studies that have used this method. RESULTS

In this group of normal subjects, end-diastolic volume ranged between 89 and 164 ml (mean 120 f 20), end-systolic volume ranged between 26 and 64 ml (mean 39 f 9) and ejection fraction ranged from 61 to 75% (mean 67 f 4) (first study and first observermea-

TABLE I Comparison Between First and Second Study HR (beats/min)

EDV (ml)

ESV (ml) ~

EF (%I ~

Pt. ID

1st

2nd

1st

2nd

1st

2nd

1st

2nd

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

72 70 55 60 75 58 85 58 88 54 57 79 60 75 75 76 60 63

70 66 57 65 80 55 87 55 86 48 54 75 62 71 72 74 65 60

147 164 119 131 104 119 127 89 150 120 110 115 133 113 104 109 90 121

138 156 109 133 93 112 121 88 162 122 122 107 131 115 92 102 91 115

47 64 44 46 41 44 32 30 48 42 35 30 47 30 34 34 26 36

44 62 42 44 39 38 33 31 50 39 38 31 44 31 29 33 32 35

68 61 63 65 61 63 75 67 68 65 68 74 65 74 66 69 71 70

6860 61 67 58 66 72 65 69 68 69 71 66 73 65 68 65 71

Mean sp

68 11

67 11

i20 20

117 21

39 9

39 8

67 4

67 4

Measurements were made by the same observer. EDV = end-diastolic volume,; EF = ejection fraction; ESV = end-systolic vo~urne; HR = heart rate; ID = subject IdentifCation number; SD = kmdard deviation; 1st = first study; 2nd = second study.

surements). When the volumes were indexed to the body surface area (mean 1.73 f 0.22 m2), end-diastolic volume ranged between 55 and 86 ml/m2 (mean 70 f 9) and end-systolic volume ranged between 16 and 30 ml/m2 (mean 23 f 5). lnterstudy variability: Two studies, acquired on different occasions, were obtained for all 18 subjects. There was no significant difference between the subject’s heart rate at rest during the first (68 f 11 beats/ min) and secondstudy (67 f 11 beats/min). The maximal difference in heart rate between studies for any subject was 9 beats/min (Table I). The study-to-study comparisonsof LV volumes and ejection fraction were based on data obtained by the same observer and are listed in Table i and displayed about the line of identity in Figure 2. There was no significant bias between the studies for the measurements obtained (Table II), except for end-diastolic volume, where the averageinterstudy difference (mean f standard deviation, 3.1 f 7.2 ml) suggesteda tendency for the secondstudy readings to be slightly lower than those from the first study. The mean absolute difference between studies (Table III) was low for all measurements: end-diastolic volume 6.6 f 4.0 ml, endsystolic volume 2.5 f 1.6 ml and ejection fraction 1.9 f 1.4%. Estimation of the variance component was calculated by using the square root component (s). A low component of the total variance was associatedwith study differences (end-diastolic volume: s = f5.4 ml; endREPRODUCIBILITY OF CIr’iE NMR MEASUREMENTS

1415

I

TABLE II Systematic

Differences

Interstudy End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (%) Interobserver End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (%) lntraobserver End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (%)

Between

TABLE Ill Absolute

Measurements

Mean f SD

p Value*

3.1 k 7.2

0.08 0.24 0.29

0.8 t 2.9 0.6 f 2.4 5.6 + 7.8 -0.2 f 2.7 1.7 f 2.2

10.01 0.80

-0.3 f 8.4 -1.6 f 4.7 1.2 2 3.1

0.87 0.16 0.12

Differences

Between

Measurements Mean f SD

Interstudy End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (%I Interobserver End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (%) Intraobserver End-diastolic volume (ml) End-systolic volume (ml) Ejection fraction (“/.I


t

*Paired test. Abbreviation as in Table I.

6.6 AZ4.0 2.5 zc 1.6 1.9 r 1.4

8.1 k 4.9 2.4 f 1.1 2.1 2 1.7 6.5 f 5.1 4.0 f 2.7 2.6 t- 1.8

Mean = mean absolute difference between measurements. Abbreviation as in Table I.

systolic volume: s = f2.1 ml; ejection fraction: s = f 1.69%). The intraclass correlation coefficient was 0.93, 0.94 and 0.83 for end-diastolic volume, endsystolic volume and ejection fraction, respectively (p
sure higher volumes than the second.The average difference was 5.6 f 7.8 ml (p
EDV (ml)

1st Study

FIGURE 2. Interstudy variability. Plots compare data obtained from the first study with data obtained from the second study on the same subjsct. The line of idenfity is shown. EDV = end-diastolic volume; EF = ejection fraction; ESV = end-systolic v&me.

1416

THE AMERICAN

JOURNAL

OF CARDIOLOGY

VOLUME

67

JUNE 15, 1991

amount of variance in a given measurement assignable to intraobserver difference. Possible intraobserver bias and variations in a random reading due to intraobserver variation were considered. Data are displayed in Figure 4. There was no significant bias between the first and second measurement for any of the parameters (Table II). Mean differences were as follows: end-diastolic volume -0.3 f 8.4 ml, end-systolic volume -1.6 f 4.7 ml and ejection fraction 1.2 f 3.1%. The mean absolute difference between studies (Table III) was also low: end-diastolic volume 6.5 f 5.1 ml, end-systolic volume 4.0 f 2.7 ml and ejection fraction 2.6 f 1.8%. The contribution of the intraobserver variance component to total variance was as follows: end-diastolic volume: s = 5.8 ml; end-systolic volume: s = f3.4 ml; ejection fraction: s = f2.3%). The intraclass correlation coefficient was high for end-diastolic and end-sys-

EDV (ml)

tolic volumes (0.95 and 0.91) but lower for ejection fraction (0.68). DISCUSSION The present study was conducted to evaluate the variability of volume and ejection fraction measurements in subjects with normal wall motion by using a recently reported gradient echo tine NMR approach. This new approach uses 2 imaging planes: ( 1) a longaxis, 2-chamber plane; and (2) a long-axis, 4-chamber plane. These planes are mutually perpendicular and aligned with respect to the true axes of the heart rather than the axes of the body. The determination of reproducibility and interobserver variability provides a frame of reference for applying this method. Accordingly, to focus on variation due to the tine NMR technique rather than variation due to a change in clinical status, normal volunteers

ESV (ml)

EF (%I

w

FIGURE 3. Ier vadabiiii. Plots compare Iffy is shown. Abbreviations as in Figure 2.

data

obtained

EDV (ml)

1st Measure 4. lntraobserver Remea-

VsriabiHy. were made

70

M

by dii

observers

from

the same study.

The /he

of i&m

ESV (ml)

eoM)Mo 110140Ia0 a0 200 FIGURE server.

w

1st Observer

1st Observer

D

10

20

SD

40

60

eo

70

80

1st Measure Plots after

compare data obtained from the same study on 2 separate occasions 13 months. The he of identity is shown. Abbrevhtions as in Figure

REPRODUCIBILITY

OF CINE NMR

by the same 2.

MEASUREMENTS

ob-

1417

were used so that the possibility of a true change in LV volumes was minimized. Two studies were excluded due to poor quality images, whereas 18 were included for analysis. Reproducibility of LV volumes and ejection fraction has been reported by different modalities.20-28 Cohn et a120demonstrated that the variation in ejection fraction between sequential radiographic LV cineangiography during comparable hemodynamic states was approximately 10 to 15%. This difference was explained mainly by the limited resolution of cineangiography and by interobserver variability. McAnulty et a121 showed, in another radiographic LV cineangiography study, substantial day-to-day high fluctuation as evidenced by large standard deviations of the mean LV volumes. Gordon et a124reported similar results for the reproducibility of LV volumes by 2-dimensional echocardiography. Little variability was found in mean volume measurements, but the wide standard deviations for paired values indicated substantial differences between individual subjects in the measurement of volumes and ejection fraction. With use of radionuclide angiographic methods, the reproducibility of measurements of ejection fraction in normal subjects25-27 is reported to be around 5% with wide variability, as shown by the standard deviation for paired differences. Variability in enddiastolic volumes between studies reported by Upton et a127was 9.9 f 5.1 ml in normal subjects at rest, and 9.8 f 6.2 ml during exercise; they concluded that a difference in end-diastolic volume of 220 ml between studies would be required for the change to be considered meaningful. Recently, Semelka et a128 examined the reproducibility of a stacked-slice short-axis tine NMR technique for determining LV volumes in 11 normal subjects. They also found good reproducibility between studies and reported percent variabilities of <6% for end-diastolic volume,
THE AMERICAN

JOURNAL

OF CARDIOLOGY

VOLUME

67

be developed that incorporate faster imaging techniques, automated analysis algorithms and tracking of the base of the ventricle (perhaps using an isotropic 3dimensional approach). Current

tine

nuclear

magnetic

resonance

study:

The mean end-diastolic volume, end-systolic volume and ejection fraction found in the current study (70 f 9 ml/m2, 23 f 5 ml/m2 and 67 f 4%) were similar to normal values reported by Semelka et a128in a previous NMR study of 11 subjects, to data obtained by Kennedy, 3o Wynne3i and their co-workers by using contrast ventriculography, and to values found in 11 normal volunteers in a study using ultrafast tine CT by Feiring et a1.32 In the present biplane long-axis study, there was no significant change in heart rate between the 2 NMR examinations. Overall, there was a high consistency between all measurements of end-diastolic volume, endsystolic volume and ejection fraction. Paired analysis of data did demonstrate a small but significant difference between observers for the end-diastolic volume and ejection fraction measurements, so that end-diastolic volume and ejection fraction were greater by observer 1 than by observer 2. However, the mean absolute difference in both these measures between observers was low (end-diastolic volume 8.1 f 4.9 ml, ejection fraction 2.1 f 1.7%) and the largest difference in ejection fraction between observers was only 6%. There was also a tendency for end-diastolic volume determined from the first study to be slightly higher than from the second study; however, this did not reach statistical significance. All other measurements failed to reveal any consistent difference between measurements. The variance component was low (less than f7 ml for volumes and less than f2.4% for ejection fraction) for interstudy, interobserver and intraobserver paired measurements. Intraclass correlation coefficients were also high for these paired measurements. These results and the absolute differences (Table III) revealed a relatively low variability when compared with those reported by other investigators using either echocardiography, or radionuclide or radiographic cineangiography methods for the measurements of LV volumes and ejection fraction.16-23 Factors contributing to variability: The factors that might contribute to measurement variability can be divided into 2 categories: biologic and technical. Biologic factors were unlikely to cause significant variation in the present study, in which all subjects were normal volunteers, examined in a resting supine position. No significant change in heart rate occurred between studies. The technical factors involve data acquisition and analysis. Variability could occur during the set-up procedure for image acquisition. The imaging plane may JUNE 15,

1991

not necessarily be the same from study to study in a given patient. In the present study, care was taken to define precisely the 2- and 4-chamber views based on the initial scouts obtained at each examination. This imaging strategy has become routine in our laboratory and thus variation is kept to a minimum. However, it is emphasized that appropriate angulation software and expertise is needed to obtain these planes. In the present study, interstudy data (analyzed by the same observer) were not significantly different (Table II) and had low variability (Figure 2, Table III), suggesting that potential differences in acquisition procedures were not a major factor in accounting for measurement variation. Analysis of images contains a subjective factor, which is demonstrated in the small but significant difference between observers for end-diastolic volume and consequently ejection fraction. These small differences occurred in a few subjects and were thought to be due to minor differences in technique between observers when defining the mitral valve plane. During diastole, the mitral leaflets are wide open and care needs to be taken in identifying the hinge points before commencing analysis. As with all operator-influenced techniques, considerable training is required to obtain consistent results. Sheehan and Mitten-Lewis,33 in a recent article examining the accuracy of LV volumes determined by radiographic cineangiography, state that their “New technicians undergo 2 to 3 months of training before being tested on 20 ventriculograms previously analyzed by a senior technician.” It is possible that application of this principle could have removed the consistent small differences in end-diastolic volume and ejection fraction noted between observers in the current study. Potential study limitations: It could be argued that the main limitation of the present study is that only subjects without known heart disease were included. The purpose of this study was to examine the intrinsic reproducibility of the NMR technique rather than be confounded by differences due to possible real changes in LV volumes and ejection fraction, which may occur on a day-to-day basis. Volumes obtained from patients may fluctuate because of myocardial ischemia or altered loading conditions due to drug therapy or changes in fluid balance. It is recognized that it is important to consider these effects in any serial patient study; however, the technical issues affecting reproducibility need first to be addressed by studies such as this, before the biologic day-to-day variations can be examined in later studies. Second, although absolute determination of LV volumes using the tine NMR biplane technique has been validated against ventricular angiography in patients with abnormal wall motion,14 the reproducibility results of the present study strictly apply only to ventricles with normal wall motion. A further study, with careful at-

tention to the aforementioned problems of myocardial ischemia, is needed to evaluate reproducibility in serial studies of patients with regional wall motion abnormalities. Last, although subjective analysis is not a limitation of the present study, the small but consistent interobserver errors do highlight the need for careful standardization of the analysis technique and the training of observers, and confirm the importance of continued efforts to develop automated analysis and edge-detection algorithms. Acknowledgment: We wish to acknowledge Cindy Comeau, BS, CNMT, for technical assistance in acquiring the NMR data, Shirley Nolen for secretarial support, Mary Lou Camp, BS, for organizational support, Bruce Carter, ASEE, for maintenance of the NMR system, and Mohammed Benomar, MD, FACC, of the Department of Cardiology, University Mohammed V, for support of Dr. Benjelloun during her fellowship at the University of Alabama.

REFERENCES 1. Nelson GR, Cohn PF, Gorlin R. Prognosis in medically-treated coronary artery disease. Influence of ejection fraction compared to other parameters. Circuhtion 1975;52:408-412. 2. Cohn PF, Gorlin R, Cohn LH, Collins JJ Jr. Left ventricular ejection fraction as a prognostic guide in surgical treatment of coronary and valvular heart disease. Am J Cardiol 1974;34:136-141. 3. Kennedy J, Dotes J, Stewart D. Left ventricular function before and following aortic valve replacement. Circulation 1977;56:944-950. 4. Stratemeir EJ, Thompson R, Brady TJ, Miller SW, Saini S, Wismer CL, Okada RD, Dinsmore RE. Ejection fraction determination by MR imaging: comparison with left ventricular angiography. Radiology 1986;158:775-777. 5. Dilworth LR, Aisen AM, Mancini J, Lande I, Buda AJ. Determination of left ventricular volumes and ejection fraction. by nuclear magnetic resonance imaging. Am Heart J 1987;113:24-32. 6. Edelman RR, Thompson RT, Kantor H, Brady TJ, Leavitt M, Dinsmore R. Cardiac function: evaluation with fast-echo MR imaging. Radiology 1987; 162:611-615. 7. Underwood SR, Gill CRW, Klipstein RH, Mohiaddin RH, Rees RSO, Longmore DB. Left ventricular volume measured rapidly by oblique magnetic resonance imaging. Er Heart J 1988;60:188-195. 8. Sechtem U, Pflugfelder PW, White RD, Gould RG, Holt W, Lipton MJ, Higgins CB. Cine MR imaging: potential for the evaluation of cardiovascular function. AJR 1987;148:239-246. 9. Utz JA, H&kens RJ, Heinsimer JA, Bashore T, Califf R, Glover G, Pelt N, Shimakawa A. Cine NMR determination of left ventricular ejection fraction. AJR 1987;148:839-843. 10. Sechtem U, Pflugfelder PW, Gould RG, Cassidy MM, Higgins CB. Measurement of right and left ventricular volumes in healthy individuals with tine MR imaging. Radiology 1987;163:697-702. 11. Pettigrew RI, Ziffer JA, Churchwell AL, Parks WJ, Baron M. Fast gradient echo imaging at 0.5T: assessment of cardiac function and valvular dysfunction. Dynamic Cardiouasc Imaging 1987;1:220-226. 12. Buser PT, Aufferman W, Holt WW, Wagner S, Kircher B, Wolfe C, Higgins CB. Noninvasive evaluation of global left ventricular function with use of tine nuclear magnetic resonance. J Am CON Cardiol 1989; 13: 1294- 1300. 13. Cranney GB, Lotan CS, Pohost GM. Evaluation of aortic regurgitation by magnetic resonance imaging. Curr Probl Cardiol 1990,15:87-l 14. 14. Cranney GB, Lotan CS, Dean L, Baxley W, Bouchard A, Pohost GM. Left ventricular volume measurement using cardiac axis NMR imaging-validation by calibrated ventricular angiography. Circulation 1990;82:154-163. 15. Keller AM, Peshock RM, Malloy CR, Buja LM, Nunnally R, Parkey RW, Wilerson JT. In viva measurement of myocardial mass using nuclear magnetic REPRODUCIBILITY

OF CINE NMR

MEASUREMENTS

1419

resonance imaging. J Am Co11 Cardiol 1986;8:113-117. 16. Florentine MS, Grosskreutz CL, Chang W, Hortnett JA, Dunn JD, Ehrhardt JC, Fleagle SR, Collins SM, Marcus ML, Skorton DJ. Measurement of left ventricular mass in viva using gated nuclear magnetic resonance imaging. J Am Co11 Cardiol 1986$:107-l 12. 17. Maddahi J, Crws J, Berman DS, Mericle J, Becerra A, Garcia EV, Henderson R, Bradley W. Non-invasive quantitation of left ventricular myocardial mass by gated proton nuclear magnetic resonance imaging. J Am CON Cardiol 1987;10:682-692. 16. Lotan CS, Cranney GB, Bouchard A, Bittner V, Pohost GM. The value of tine NMR for assessing regional ventricular function. J Am CON Cardiol 1989;14:1721-1729. 19. Dodge HT, Sandier H, Ballew DW, Lord JD. The use of biplane angiocardiography for the measurement of left ventricular volume in man. Am Heart J

1960;60:762-776. 20. Cohn PF, Levine

JA, Bergeron GA, Gorlin R. Reproducibility of the angiographic left ventricular ejection fraction in patients with coronary artery disease. Am Heart J 1974;88:713-720. 21. McAnulty JH, Kremkau EL, Hattenhauer MT, Rahimtoola SH. Spontaneous changes in left ventricular function between sequential studies. Am J Cardiol 1974;34:23-28. 22. Rogers WJ, Smith R, Hood WP, Mantle JA, Rackley CE, Russell RO. Effect of filming projection and interobserver variability on angiographic biplane left ventricular volume determination. Circulation 1979;59:96-104. 23. Chaitman BR, DeMots H, Bristow JD, Rosch J, Rahimatoola SH. Objective and subjective analysis of left ventricular angiograms. Circulation 1975;52:

420-425. 24. Gordon EP, Schinittger I, Fitzgerald PJ, Williams P, Popp RL. Reproducibility of left ventricular volumes by two-dimensional echocardiography. J Am Co11

1420

THE AMERICAN

JOURNAL

OF CARDIOLOGY

VOLUME

67

Cardiol 1983;2:506-513. 25. Marshall RC, Berger HJ, Reduto LA, Gottschalk A, Zaret BL. Variability in sequential measures of left ventricular performance assessed with radionuclide angiography. Am J Cardiol 1978;41:531-526. 26. Wackers FJT, Berger H, Johnstone DE, Goldman L, Reduto LA, Langou RA, Gottschalk A, Zaret BL. Multiple gated cardiac blood pool imaging for left ventricular ejection fraction: validation of the technique and assessment of variability. Am J Cardtol 197%43:1159-l 166. 27. Upton MT, Rerych SK, Newman GE, Bounous EP, Jones RH. The reproducibility of radionuclide angiographic measurements of left ventricular function in normal subjects at rest and during exercise. Circulation 1980;3:126-132. 26. Semelka RC, Tomei E, Wagner S, Mayo J, Kondo C, Suzuki J, Caputo GC, Higgins CB. Normal left ventricular dimensions and function: interstudy reproducibility of measurements with tine MR imaging. Radiology 1990;174:763-768. 29. ierhouni EA, Parish DM, Rogers WJ, Yang A, Shapiro EP. Human heart: tagging with MR imaging-a method for non-invasive assessment of myocardial motion. Radiology 1988;169:59-63. 30. Kennedy JW, Baxley WA, Figley MM, Dodge H, Blackman J. Quantitative angiocardiography I. The normal left ventricle in man. Circulation 1966;

34272-278. 31. Wynne

J, Greene LH, Mann T, Levin D, Grossman W. Estimation of left ventricular volumes in man from biplane cineangiograms filmed in oblique projections. Am J Cardiol 1978;41:726-732. 32. Feiring AJ, Rumberger JA, Reiter SJ, Collins SM, Skorton DJ, Rees M, Marcus ML. Sectional and segmental variability of left ventricular function: experimental and clinical studies using ultrafast computed tomography. J Am Co11 Cardiol 1988;12:415-425. 33. Sheehan HF, Mitten-Lewis S. Factors influencing accuracy in left ventricular volume determination. Am J Cardiol 1989;64:661-664.

JUNE 15, 1991