Author + information
- Received January 8, 1999
- Revision received September 10, 1999
- Accepted October 21, 1999
- Published online February 1, 2000.
- Michael L Chuang, MS∗,
- Mark G Hibberd, MD, PhD∗,
- Carol J Salton, BA∗,
- Raymond A Beaudin, MS†,
- Marilyn F Riley, BS∗,
- Robert A Parker, ScD‡,
- Pamela S Douglas, MD, FACC∗ and
- Warren J Manning, MD, FACC∗,§,* ()
- ↵*Reprint requests and correspondence: Dr. Warren J. Manning, Cardiovascular Division, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Boston, Massachusetts 02215
This study sought to determine the concordance between biplane and volumetric echocardiography and magnetic resonance imaging (MRI) strategies and their impact on the classification of patients according to left ventricular (LV) ejection fraction (EF) (LVEF).
Transthoracic echocardiography and MRI are noninvasive imaging modalities well suited for serial evaluation of LV volume and LVEF. Despite the accuracy and reproducibility of volumetric methods, quantitative biplane methods are commonly used, as they minimize both scanning and analysis times.
Thirty-five adult subjects, including 25 patients with dilated cardiomyopathies, were evaluated by biplane and volumetric (cardiac short-axis stack) cine MRI and by biplane and volumetric (three-dimensional) transthoracic echocardiography. Left ventricular volume, LVEF and LV function categories (LVEF ≥55%, >35% to <55% and ≤35%) were then determined.
Biplane echocardiography underestimated LV volume with respect to the other three strategies (p < 0.01). There were no significant differences (p > 0.05) between any of the strategies for quantitative LVEF. Volumetric MRI and volumetric echocardiography differed by a single functional category for 2 patients (8%). Six to 11 patients (24% to 44%) differed when comparing biplane and volumetric methods. Ten patients (40%) changed their functional status when biplane MRI and biplane echocardiography were compared; this comparison also revealed the greatest mean absolute difference in estimates of EF for those subjects whose EF functional category had changed.
Volumetric MRI and volumetric echocardiographic measures of LV volume and LVEF agree well and give similar results when used to stratify patients with dilated cardiomyopathy according to systolic function. Agreement is poor between biplane and volumetric methods and worse between biplane methods, which assigned 40% of patients to different categories according to LVEF. The choice of imaging method (volumetric or biplane) has a greater impact on the results than does the choice of imaging modality (echocardiography or MRI) when measuring LV volume and systolic function.
Left ventricular (LV) volume and ejection fraction (EF) are important discriminants of prognosis in both ischemic and nonischemic cardiomyopathies (1–4). Left ventricular volume also provides valuable prognostic information that may be useful in the selection of therapy (3,5)or determination of the optimal time for operation (6–8). Thus, accurate determination of these values is important for clinical assessment of cardiac status and for serial monitoring. Currently, transthoracic (surface) echocardiography is the most commonly used noninvasive method for screening andfor the serial measurement of LV volume and LVEF, as it is noninvasive and widely available. With clinical biplane echocardiography, LV volumes are determined using two tomographic views of the heart. The accuracy of these measures depends on the ability to acquire specific “standard” cross-sectional views. This strategy also requires that the standard views be representative of the heart and that three-dimensional (3D) heart shape is well described by a particular geometric model. A variety of 3D echocardiographic approaches (9–11)have been shown to provide more accurate determination of LV size and systolic function than with biplane echocardiography, as these volumetric methods do not require heart shape to conform to a geometric model to accurately determine LV volume.
Breathhold cine MRI is also noninvasive and well suited for serial assessments of LV volumes and LVEF. Volumetric MRI methods have been shown to be both highly accurate and highly reproducible for the measurement of LV volumes and LVEF when multiple thin slices are used to cover the LV in the cardiac short-axis orientation (12). However, nearly all currently implemented volumetric methods, MRI and echocardiography require multiple cross-sectional views of the heart, with time-intensive acquisition and analysis. Thus, biplane methods remain attractive, requiring only two views and minimal data acquisition and analysis times. Biplane MRI is of particular interest because, in contrast to transthoracic echocardiography, MRI is not dependent on good acoustic windows and, in principle, allows reliable acquisition of the required standard views of the heart.
Previous studies have shown that LV volume and LVEF measured using biplane MRI correlate well with the results obtained using other established methods (13), but the concordance between biplane and volumetric methods using MRI and echocardiography on the same patients has not been systematically examined. Therefore, we sought to determine 1) the limits of agreement between biplane and volumetric methods for the measurement of LV volumes and LVEF in healthy adults and in patients with dilated cardiomyopathy to assess the relative impact of imaging method (biplane vs. volumetric) as compared with imaging modality (MRI vs. echocardiography) and 2) how each of the imaging strategies would classify individual patients according to LVEF.
Thirty-five subjects (10 women and 25 men, 24 to 76 years old) were evaluated by both MRI and echocardiography after providing written, informed consent. The study was approved by the hospital’s Committee on Clinical Investigation. Patients with contraindications to MRI (pacemaker, implantable cardioverter-defibrillator, intraoccular implants, claustrophobia, atrial fibrillation) were excluded. Subjects were not screened for echocardiographic image quality. For each subject, biplane and volumetric echocardiographic and MRI examinations were all performed within 3 h. There were no changes in the patients’ clinical status during this period. The 35 subjects included 25 patients with dilated cardiomyopathy (10 ischemic with rest focal wall motion abnormalities, 7 with severe valvular regurgitation [2 mitral, 5 aortic], 8 with nonischemic cardiomyopathy) and 10 clinically healthy adult volunteers.
Magnetic resonance imaging
MRI examination was performed using an electrocardiogram (ECG)-triggered segmented k-space breathhold cine sequence (12)(repetition time = 11 ms, echo time = 5.2 ms, flip angle = 25°, 112 × 256 matrix, 224 × 320 mm2field of view) with the subjects supine in a 1.5-tesla whole-body system (Gyroscan NT, Philips Medical Systems, Best, The Netherlands) using either an anteriorly placed 20-cm diameter surface coil (C1) or a five-element cardiac synergy coil for radiofrequency signal reception. Effective temporal resolution ranged from 50 to 70 ms for both biplane and volumetric MRI, performed at end-tidal breathhold.
After a series of scout images to determine the position and orientation of the LV within the thorax, two cine loops, each spanning the cardiac cycle, were obtained in the two-chamber and four-chamber LV long-axis orientations. A slice thickness of 5 mm was used.
Images were acquired using 16 contiguous slices (stack) in the double-oblique LV short-axis orientation, covering the LV from the apex to just above the level of the base. Slice thicknesses ranged from 5 to 7.5 mm and were selected to enable coverage of the LV within 16 total slices.
Echocardiographic examinations were performed using commercially available scanners (Sonos 1500 or Sonos 2500, Hewlett Packard Medical Products, Andover, Massachusetts) using a 2.5-MHz phased-array cardiac probe with subjects in the left lateral recumbent position. Cine loops, triggered on the ECG QRS complex and spanning the cardiac cycle, were acquired at a 30-Hz frame rate and recorded digitally on optical disk for both biplane and 3D echocardiographic examinations. The biplane echocardiographic examination used apical two-chamber and four-chamber data acquired at end-tidal breathhold.
Three-dimensional cardiographic examinations were performed using the Sonos 1500 or 2500 scanner and 2.5-MHz probe in conjunction with custom scanner software and a Flock of Birds magnetic location system (Ascension Technologies, Burlington, Vermont). The “Bird” system performs 6 degrees-of-freedom tracking, allowing “freehand” scanning so that the sonographer can select from all available acoustic windows during a single examination. The position and orientation of the transducer (14,15)were recorded at a 30-Hz rate to match the image frame rate and digitally encoded in the header of each image to allow temporal and spatial registration of images during analysis. Twenty ECG-triggered cine loops widely surveying the LV were acquired from a combination of apical and parasternal views at end-tidal breathhold. As for the biplane echocardiographic examination, cine loops extended from the QRS complex to beyond peak systole.
Three-dimensional echocardiographic data were analyzed by an observer who had no knowledge of the MRI data. Volumetric MRI data were similarly analyzed without knowledge of 3D echocardiographic results. Biplane echocardiographic and MRI measurements were performed without knowledge of volumetric data at a separate time point after all volumetric data analyses. Ten subjects were randomly selected for assessment of intraobserver and interobserver variabilities. Observer variability analyses were performed without knowledge of patient identity or previous results. In addition, the intraobserver analyses were performed several months after primary analyses.
Magnetic resonance imaging
Endocardial borders at end-diastole (ED) and end-systole (ES) were manually traced using standard system software analysis tools. For the biplane examination, LV ES and ED volumes (ESV and EDV) were determined using the long-axis biplane ellipsoid formula: V = 8A2A4/(3pLmin), where A2and A4are the planimetered two- and four-chamber LV cross-sectional areas, respectively; and Lminis the shorter of the apex-to-base lengths of the two views. In the volumetric examination, volumes were computed by a summation of the disks method (“Simpson’s rule”) where the sum of cross-sectional areas was multiplied by the slice thickness. For both methods, LVEF was computed as 100%∗(EDV − ESV)/EDV.
For biplane echocardiography, EDV and ESV were computed using manually traced endocardial contours and the same long-axis biplane ellipsoid formula as for biplane MRI. Three-dimensional echocardiographic data were analyzed using a Unix-based workstation (Hewlett Packard Series 735) by manually tracing endocardial contours on each of the 20 component cine loops, first at ED and then at ES (14). During tracing, contours were also displayed in another window as a 3D wire frame model, helping the operator visualize the 3D relation between component image planes and to ensure 3D consistency of the complete set of contours. Volumes were determined using a minimal energy deformable shell model (16–18). Echocardiographic borders were traced in accordance with the recommendations of the American Society of Echocardiography (19).
Results were compared by repeated measures analysis of variance. The Ftest was used to test the null hypothesis that the four imaging strategies used produce identical estimates of LV volume and LVEF for each subject. The paired Student-Newman-Keuls test was used for multiple pairwise comparisons between imaging strategies. All tests were two-tailed and p ≤ 0.05 was considered significant. Additional pairwise comparisons were made by limits of agreement according to the method described by Bland and Altman (20). Correlation between imaging strategies was summarized using standard (Pearson) correlation. Summary data are presented as the mean value ± SD. Interobserver and intraobserver variations were computed as the root of the mean squared differences between corresponding observations, divided by the average of all observations.
The practical impact of differences between imaging strategies was further assessed by using each method to classify each patient according to LVEF into one of three categories: 1) normal or compensated systolic function (LVEF ≥55%); 2) depressed (LVEF >35% to <55%); and 3) severely depressed (LVEF ≤35%). Pairwise comparisons between the various imaging strategies showed the number of subjects assigned to different functional categories for each pair of strategies.
Finally, we classified the results of the two imaging strategies as consistent if the measured LVEF values were no more than 5% different, and we used McNemar’s test to assess whether one method (or modality) was more consistent than the other.
Echocardiographic and MRI examinations were successfully completed in all subjects without incident. Left ventricular EDV, as determined by volumetric MRI, ranged from 96 to 540 ml (196 ± 91 ml), ESV ranged from 27 to 482 ml (105 ± 91 ml) and LVEF ranged from 11% to 73% (52 ± 17%). Biplane echocardiography significantly underestimated EDV (p < 0.01) and EDS (p < 0.01) relative to the other three strategies (Table 1). Left ventricular volumes, as measured by volumetric echocardiography, volumetric MRI and biplane MRI, did not differ significantly for EDV or ESV. There were no significant differences in LVEF between any of the four imaging strategies. Further comparisons by patient subgroup (those with rest focal wall motion abnormalities [n = 10] and without [n = 15]) did not reveal additional differences between imaging strategies in the patient cohort studied.
Correlations (r2) between each of the imaging strategies for LV volume ranged from 0.88 (comparing biplane echocardiography and biplane MRI) to 0.99 (comparing volumetric echocardiography and volumetric MRI) for LV volumes. Intermodality and intramodality comparisons between volumetric and biplane methods yielded correlations between the low extreme of biplane versus biplane and the high extreme of volumetric versus volumetric. Similarly, correlations ranged from 0.73 (volumetric MRI vs. biplane echocardiography) to 0.98 (volumetric MRI vs. volumetric echocardiography) for LVEF. The limits of agreement (Fig. 1)showed that the mean bias, or systematic overestimation or underestimation, of LVEF is small when comparing the results of biplane and volumetric methods for both MRI and echocardiographic modalities, but the 95% confidence intervals are wide (±9.8 to ±10.6 EF units) when comparing biplane MRI results with volumetric methods, and very wide (±17.7 to ±19.2 EF units) when comparing biplane echocardiography with any of the other three imaging strategies. In contrast, the limits of agreement are over threefold tighter (±5.2 EF U) when comparing LVEF measured by volumetric echocardiography with LVEF by volumetric MRI. Similarly, for LV volumes, intermodality and intramodality limits of agreement between biplane and biplane or biplane and volumetric methods were approximately threefold wider than those between volumetric MRI and volumetric echocardiography. Analyzing patient subgroups separately (according to normal, depressed and severely depressed LVEF subgroups) produced consistent results, with best agreement between volumetric MRI and echocardiography and poorest agreement between biplane techniques.
Classification of patients according to LVEF
By classifying subjects as having either normal (LVEF ≥55%), depressed (LVEF >35% to <55%) or severely depressed (LVEF ≤35%) systolic function, volumetric MRI indicated 17 subjects with normal, 12 patients with depressed and 6 patients with severely depressed EF. All 10 healthy adults were classified as having a normal EF by all four imaging strategies. For the 25 patients, there were differences in functional category depending on the imaging strategy used. The number of patients whose classification changed between each pair of strategies is shown in Table 2. When comparing volumetric echocardiographic and volumetric MRI (Fig. 2a), 2 (8%) of 25 patients were assigned to different functional categories; for these patients, the mean (±SD) absolute difference between imaging strategies was small at 2.9 ± 0.8 EF U. Comparing biplane and volumetric methods (Fig. 2b)revealed a greater proportion of patients with different classifications, ranging from 6 (24%) of 25 patients to 11 (44%) of 25 patients; the mean absolute difference was 9.5 ± 6.5 EF U. Finally, functional classification differed for 10 (40%) of 25 patients when comparing biplane echocardiography with biplane MRI (Fig. 2c); the mean absolute difference was 13.6 ± 6.9 EF U.
Consistency of imaging strategies
When comparing imaging modalities, there were no significant differences in consistency between echocardiography and MRI (p = 0.637); the number of inconsistencies between biplane and volumetric MRI methods was comparable to the number of inconsistencies between biplane and volumetric echocardiographic methods. However, comparison of imaging methods indicated that volumetric echocardiography and MRI were significantly more consistent (p = 0.001) with respect to one another than were biplane echocardiography and biplane MRI.
A subgroup of 10 subjects (8 patients, 2 healthy volunteers) was analyzed to determine observer reproducibilities, which are presented in this order: volumetric MRI, biplane MRI, volumetric echocardiography, biplane echocardiography. Interobserver EDV: 3.5%, 9.4%, 4.0% and 17.5%; interobserver ESV: 4.8%, 12.5%, 5.6% and 24.3%; intraobserver EDV: 2.6%, 8.1%, 3.2% and 17.3%; and intraobserver ESV: 3.5%, 10.5%, 4.2% and 22.4%. Observer variabilities for LVEF are presented in Table 3. With regard to the classification of patients by volumetric MRI, all patients were assigned to the same functional category regardless of observer or observation. Intraobserver classifications did not differ for volumetric echocardiography; for interobserver classification, one patient was assigned to a different category by the second echocardiogram reader. For biplane MRI, two patients showed intraobserver differences and two interobserver differences. Finally, interobserver biplane echocardiographic classification differed for two patients and intraobserver classification differed by one patient, whose interobserver classification also differed.
To our knowledge, this is the first study comparing volumetric with biplane assessments of LV volume and LVEF by MRI and echocardiography on the same patient groups. Currently, quantitative measurements of LV size and function are most commonly obtained using two-dimensional (2D) transthoracic echocardiography, which is highly dependent on operator technique and which can be limited by poor acoustic windows and the need for geometric assumptions about heart shape. Volumetric methods of image acquisition, whether by 3D echocardiography or cardiac short-axis stack MRI, have demonstrated substantial improvements in accuracy and reproducibility over 2D echocardiography (9,14). However, these methods require additional imaging and data analysis time. Thus, biplane MRI is attractive, as it only requires two views and is unhindered by the need for good acoustic windows and the uncertainty of imaging plane positioning; therefore, it is likely less operator dependent than biplane echocardiography. However, biplane MRI is still subject to the limitations of geometric assumptions.
In this study we found no significant mean (group) differences between any of the imaging strategies in the determination of LVEF. However, pairwise comparisons between strategies revealed excellent agreement between volumetric echocardiography and volumetric MRI; modest agreement between volumetric methods and biplane MRI; and poor correspondence between biplane echocardiography and biplane MRI (with over threefold wider limits of agreement than those between the volumetric methods). Biplane echocardiography significantly underestimatedLV volume relative to the other three imaging strategies. There were no significant differences between volumetric MRI, volumetric echocardiography and biplane MRI, although the limits of agreement between biplane MRI and either of the volumetric methods were substantially wider than the limits of agreement between the two volumetric methods.
Previous comparisons of volumetric with biplane MRI
Several investigators have compared volumetric and biplane MRI measures of LV volume and LVEF. Dulce et al. (21)used nonbreathhold cine MRI to examine 10 normal subjects and 10 patients with LV hypertrophy using a variety of mono and biplane geometric formulas to determine LV volume and LVEF, which they compared with short-axis stack volumetric results. The hemisphere-cylinder model (⅚ area-length or “bullet” formula [22,23]) overestimated LV volumes in both normal subjects and those with LV hypertrophy and underestimated EF in normal subjects. For patients with LV hypertrophy, differences in LV volume or LVEF were described between bullet, single-plane ellipsoid and Teichholz (22)models. Standard errors of the estimate for LV volume were as high as 15 ml (EDV) and 11 ml (ESV). Limits of agreement analyses were not reported. Van Pol et al. (24)recently presented preliminary data comparing volumetric and biplane MRI in 174 examinations of patients with severely depressed LVEF (≤35%) and found correlations of ∼0.90 for EDV, ESV and LVEF. Limits of agreement analysis indicated systematic differences (mean bias) of ≤3% for EDV, ESV and LVEF. These studies suggest that aggregate volumetric and biplane MRI measures of LV volume and LVEF differ only slightly. Our results indicate that the variability in individual measures can be large, so that biplane and volumetric MRI cannot be considered interchangeable for a given patient.
Previous comparisons of volumetric imaging with biplane echocardiography
In contrast to MRI, the accuracy and reproducibility of volumetric (3D) relative to biplane (2D) echocardiography have been more extensively studied. Although there are hardware and software differences between 3D echocardiographic implementations (25), volumetric echocardiography appears superior to biplane echocardiography when the results are compared with MRI (14,26), equilibrium radionuclide angiography (9,10)or postmortem anatomic results (27). The greater differences between volumetric and biplane echocardiography, relative to differences between biplane and volumetric MRI, can be attributed to several factors. First, neither the absolute nor relative positions of imaging planes are known in biplane echocardiography, although standard cross-sectional views of the heart must be obtained. Selection of imaging planes and accuracy of biplane echocardiography are critically dependent on operator experience (28), yet standard views can be difficult to recognize in diseased ventricles and may not be representative even if obtained. In addition, limited acoustic windows may not allow acquisition of the needed standard views of the heart (29).
Reichek et al. (30)recently presented preliminary results comparing volumetric MRI and biplane echocardiography to determine LVEF in 28 patients within 9 days (mean 4 days) after an acute myocardial infarction. Echocardiographic EF was assessed visuallyby two observers and quantitatively using the apical biplane Simpson’s rule. There was poor agreement (r = 0.43 0.46) between MRI and any of the biplane echocardiographic methods. At least 10 (35%) of 28 patients were misclassified (LVEF ≥40% vs. <40%) by echocardiography relative to volumetric MRI, a value similar to ours. Ray et al. (31)compared mono or biplane 2D echo to radionuclide ventriculography performed at two centers among 70 patients within 36 h of acute infarction. Differences between LVEF by radionuclide ventriculography and echocardiography were −8 ± 10% at one center and −14 ± 11% at the other. In addition, radionuclide ventriculographic EF classified 67 (96%) of 70 patients with LVEF <40%, whereas echocardiography identified only 39 (56%) of 70 patients with LVEF <40%, so that 40% of the patients would have been classified differently depending on the imaging test used.
Biplane echocardiographic measurements were done using the apical four-chamber and two-chamber views, as recommended by the American Society of Echocardiography (19). However, the two-chamber view does not depict the aorta or outflow tract, and the LV apex may be foreshortenend. These factors may contribute to the systematic underestimation of LV volume by biplane echocardiography in the present study. Use of the apical three-chamber or long-axis view may in part address this issue, but the three-chamber view was not obtained for all patients in this study.
Volumetric MRI and volumetric (3D) echocardiographic methods agree well for LV volume and LVEF and give similar results when used to stratify healthy subjects and patients according to systolic function. Agreement is poor between biplane and volumetric methods regardless of imaging modality, and worse when comparing biplane MRI with biplane echocardiography, which assigned 40% of patients to different functional categories. The choice of imaging method (volumetric vs. biplane) appears to have a greater impact than does the choice of imaging modality (MRI vs. echocardiography) in measures of LV volume and LVEF. The excellent agreement between volumetric echocardiography and MRI suggests that volumetric methods are strongly preferred.
☆ The work of Drs. Chuang, Hibberd and Douglas is supported in part by an unrestricted research grant from the Hewlett Packard Company, Andover, Massachusetts. Dr. Manning is supported in part by an Established Investigator Award (no. 9740003N) from the American Heart Association (Dallas, Texas). The study was also supported in part by Grant MO1-RR01032, from the National Institutes of Health, Bethesda, Maryland, to the General Clinical Research Center of the Beth Israel Deaconess Medical Center, Boston, Massachusetts.
- end-diastolic volume
- ejection fraction
- end-systolic volume
- left ventricle or ventricular
- left ventricular ejection fraction
- magnetic resonance imaging
- Received January 8, 1999.
- Revision received September 10, 1999.
- Accepted October 21, 1999.
- American College of Cardiology
- ↵Wong M, Johnson G, Shabetai R, et al., for the V-HeFT Veterans Affairs Cooperative Studies Group. Echocardiographic variables as prognostic indicators and therapeutic monitors in chronic congestive heart failure: Veterans Affairs Cooperative Studies V-HeFT I and II. Circulation 1993;87 Suppl VI:VI-65–70.
- Volpi A,
- DeVita C,
- Franzosi M,
- et al.
- Cohn JN, Johnson GR, Shabetai R, et al., the V-HeFT VA Cooperative Studies Group. Ejection fraction, peak exercise oxygen consumption, cardiothoracic ratio, ventricular arrhythmias, and plasma norepinephrine as determinants of prognosis in heart failure. Circulation 1993;87 Suppl VI:VI-5–16.
- Bonow R.O,
- Lakatos E,
- Maron B.J,
- Epstein S.E
- Hwang M.H,
- Hammermeister K.E,
- Oprian C,
- et al.
- Carabello B.A
- Gopal A.S,
- Shen Z,
- Sapin P.M,
- et al.
- Nosir Y.F,
- Fioretti P.M,
- Vletter W.B,
- et al.
- Siu S.C,
- Rivera J.M,
- Guerrero J.L,
- et al.
- Cook LT, Cook PN, Lee KR, et al. An algorithm for volume estimation based on polyhedral approximation. IEEE Trans Biomed Eng 1980;BME-27:493–500.
- Young AA, Hunter PJ, Smaill BH. Epicardial surface estimation from coronary angiograms: computer vision, graphics and image processing. 1989;47:111–27.
- Schiller N.B,
- Shah P.M,
- Crawford M,
- et al.
- Folland E.D,
- Parisi A.F,
- Moynihan P.F,
- et al.
- Wyatt H.L,
- Heng M.K,
- Meerbaum S,
- et al.
- ↵van Pol P, Foster R, Davis N, et al. Optimal clinical evaluation of left ventricular function using MRI (abstr). Circulation 1997;96 Suppl I:I-514.
- Gopal A.S,
- Schnellbaecher M.J,
- Shen Z,
- et al.
- ↵Reichek N, Theobald T, Malkowski M, et al. MRI and echo LV ejection fractions after recent acute myocardial infarction (abstract). Circulation 1997;96 Suppl I:I-189.
- ↵Ray SG, Metcalfe MJ, Oldroyd KG, et al. Do radionuclide and echocardiographic techniques give a universal cutoff value for left ventricular ejection fraction that can be used to select patients for treatment with ACE inhibitors after myocardial infarction? Br Heart J 1995;73:466–9.