Reproducibility of Right Ventricular Volumes and Ejection Fraction Using Real-time Three-Dimensional Echocardiography*

Transcription

1 Original Research PULMONARY VASCULAR DISEASE Reproducibility of Right Ventricular Volumes and Ejection Fraction Using Real-time Three-Dimensional Echocardiography* Comparison With Cardiac MRI Carly Jenkins, MS; Jonathan Chan, MD; Kristen Bricknell, MS; Mark Strudwick, PhD; and Thomas H. Marwick, MD, PhD Objectives: The nongeometric nature of the right ventricle (RV) makes it difficult to measure. We sought to determine whether real-time three-dimensional echocardiography (RT3DE) is superior to two-dimensional echocardiography (2DE) for the follow-up of RV function by validation vs cardiac MRI. Methods: RV volumes and ejection fraction (EF) were studied with 2DE (including area-length [A-L], the modified two-dimensional subtraction [2DS] method, and the Simpson method of discs), RT3DE, and MRI in 50 patients with left ventricular wall motion abnormalities, the results of which suggested possible RV infarction. Test-retest variation was performed by a complete restudy using a separate sonographer within 24 h without the alteration of hemodynamics or therapy. Interobserver and intraobserver variations were noted in a subgroup of 20 patients. Results: EF estimations were similar using each technique. The mean ( SD) MRI end-diastolic volume (87 22 ml) was only slightly underestimated by RT3DE (mean difference, 3 10; p < 0.05), with a greater mean difference for 2DE A-L ( 29 10; p < 0.05), and the Simpson method of discs ( 29 23; p < 0.05), and was greatly overestimated by 2DS (mean difference, 26 23; p < 0.05). Similarly, the mean MRI end-systolic volume (46 17 ml) was only slightly underestimated by RT3DE ( 4 7; p < 0.05), compared with 2DE A-L ( 16 8; p < 0.05) and the Simpson method of discs ( 16 8; p < 0.05), and was overestimated by 2DS (14 13; p < 0.05). RT3DE findings had a higher correlation with each parameter than any 2DE technique. There was also good intraobserver and interobserver correlation between RT3DE by two sonographers. RT3DE had less test-retest variation of RV volumes and EF than any 2DE measure. Conclusions: RT3DE is more accurate than two-dimensional approaches and reduces the test-retest variation of RV volumes and EF measurements in follow-up RV assessment. (CHEST 2007; 131: ) Key words: ejection fraction; real-time three-dimensional echocardiography; right ventricle; two-dimensional echocardiography; volumes Abbreviations: A-L area-length; EDV end-diastolic volume; EF ejection fraction; ESV end-systolic volume; LV left ventricle, ventricular; RT3DE real-time three-dimensional echocardiography; RV right ventricle, ventricular; 3D three-dimensional; 3DE three-dimensional echocardiography; 2DE two-dimensional echocardiography; 2DS two-dimensional subtraction The degree of right ventricular (RV) involvement is an important prognostic determinant after myocardial infarction. 1 Moreover, sequential assessments of RV volumes and ejection fraction (EF) are potentially important in the follow-up of patients with congenital heart disease such as corrected tetralogy of Fallot, 2 and the evaluation of RV systolic function is of value in assessing prognosis and treatment response in patients with pulmonary hypertension. 3 A reliable noninvasive technique for RV eval Original Research

2 uation would therefore be useful in patients with ischemic and other heart diseases. Two-dimensional echocardiography (2DE) remains the most widely available technique, but the nongeometric nature of the RV and its heavy trabeculations makes the measurement of RV volume difficult. Consequently, evaluation of the RV by 2DE is most commonly qualitative, rather than quantitative, and this is of limited value in the assessment of the RV over time. Cardiac MRI is an alternative for RV volume and function assessment, 4 but it is expensive, of limited availability, and cannot be performed in the increasing number of heart failure patients with implantable devices. Real-time three-dimensional echocardiography (RT3DE) has been validated 5 as a reliable technique for the assessment of left ventricular (LV) size and function. Previous studies of RV volumes and EF with three-dimensional echocardiography (3DE) have been performed using three-dimensional (3D) reconstruction, 6 sparse array transducers, 7 9 and RT3DE, but none have addressed test-retest reliability. We hypothesized that RT3DE would prove a reliable, reproducible technique for the follow-up of RV function, with results that would be comparable to those of cardiac MRI, and superior to those of 2DE. Therefore, the objectives of this study were as follows: (1) to compare RV volumes and EF by 2DE, RT3DE, and MRI; (2) to determine whether the test-retest variability of RT3DE; and (3) to compare the results of RT3DE to similar results from 2DE techniques. Study Design Materials and Methods We prospectively recruited patients who had been referred to the echocardiography laboratory for the assessment of cardiac structure and function after experiencing an acute myocardial infarction, and who were scheduled to undergo 2DE, RT3DE, and MRI. In 54 patients (47 men; mean [ SD] age, years), RV involvement was suspected from standard imaging, and an additional medial RV view was obtained using RT3DE *From the University of Queensland, Brisbane, QLD, Australia. Supported in part by a grant-in-aid from the National Health and Medical Research Council of Australia. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received August 29, 2006; revision accepted January 28, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Thomas H. Marwick, MD, PhD, University of Queensland, Department of Medicine, Princess Alexandra Hospital, Ipswich Rd, Brisbane Q4102, QLD, Australia; tmarwick@soms.uq.edu.au DOI: /chest (see RT3DE section below). The investigations were approved by the Human Research Ethics Committee of Princess Alexandra Hospital, and all patients gave informed consent. Data for test-retest variability were obtained by discharging patients from the laboratory and repeating their imaging within 24 h with no intervening therapy. A subgroup (n 20) was studied for interobserver variability, which was determined by using the same set of 3D and two-dimensional images measured by two separate sonographers. The same group was tested for intraobserver variability, with volume measurements repeated on the same data set by the same sonographer. Intraobserver repeated measures were performed on average 1 week apart; repeated analysis was performed in random order. 2DE An experienced sonographer acquired apical views of the RV, using harmonic imaging with a transthoracic 3-MHz phased array transducer (Sonos 7500; Philips Medical Systems; Andover, MA). Measurements of RV end-diastolic volume (EDV), RV endsystolic volume (ESV), and EF were obtained using the software installed on the ultrasound machine, with EDV measured at the time of tricuspid valve closure and ESV measured on the image with the smallest RV cavity. The papillary muscles were excluded from the volumes; however, the moderator band and trabeculations were included. The area-length (A-L) method used was the monoplane ellipsoid method, in which the RV volume is 3 / 8 {(Area-apical 4chr) 2 /Length-apical 4chr} (Fig 1, top left, A). 13 The Simpson method of discs is based on the division of the RV into a series of slices of equal thickness, the volume of each of which is calculated from the following formula: volume area length, so that the RV volume is calculated from the sum of these slices (Fig 1, center left, B). Both measurements were obtained from apical four-chamber views by tracing the endocardial border of the RV. 14 The 2DE subtraction method 15 was obtained from the apical four-chamber view by tracing the volume of the LV with the inclusion of the interventricular septum and subtracting it from the total volume of the LV and RV (Fig 1, bottom left, C). RT3DE RT3DE images were obtained from a modified apical window (to open up the RV and to situate it in the middle of the sector, apical images were slightly more medial than the standard view) with the patient in the same position as for 2DE. Images were also gathered over four cardiac cycles using a matrix array transducer ( 4 transducer, Philips Sonos 7500; Philips Medical Systems). Measurements of RT3DE volumes and EF were performed off-line (4D Analysis; Tomtec Gmbh; Unterschlessheim, Germany). Frames for EDV and ESV measurements were identified by the same method as that for 2DE, and endocardial contours were marked in 12 slices (ie, 15 per slice). Contour tracing was performed with semiautomatic border detection; after first identifying the apex and tricuspid annulus on each slice, a preconfigured ellipse was fitted to the endocardial borders of each frame and was adjusted as required. The EDV and ESV were measured from the resulting 3D volume (Fig 1, top right, D). RV measurements were obtained by applying the LV model to fit the RV, with semiautomated border detection and manual editing of the borders to give a 3D model from which volumes can be measured without geometric assumptions. The mean scanning time for 3DE taken for one full volume was s; however, the acquisition time requiring a breathhold was approximately 10 s. The time required for the calculation of RV volumes ranged from 5 to 10 min. CHEST / 131 / 6/ JUNE,

3 Figure 1. Top left, A: analysis of RV volume using 2DE A-L; the left panel shows EDV, and the right panel shows ESV. Center left, B: analysis of RV volume using 2DE Simpson method of discs, the left panel shows EDV, and the right panel shows ESV. Bottom left, C: analysis of RV volume using the 2DS method, the top panels show the volume calculated from the entire LV and RV volume, while the bottom panels show the volume from the LV and intraventricular septum; left panels show EDV, and the right panels show ESV. Top right, D: analysis of RV volume using RT3DE, demonstrating the selection of one image (top right), an automated contour tracing (lower left, the superimposition of all contours in the 3D space (top left), and the resulting time-volume curve (bottom right). Bottom right, E: analysis of RV volume using MRI; automated tracings are shown for EDV (left panel) and ESV (right panel). MRI MRI was performed using a 1.5-Tesla scanner (Sonata; Siemens; Erlangen, Germany). A true free induction, steady-state precession sequence (recovery time, 47.1 ms; echo time, 1.57 ms; flip angle, 60 ; bandwidth, 930 Hz/pixel) was used to examine RV anatomy and function. Images were acquired during breath-hold in short-axis planes (voxel size, mm) perpendicular to the tricuspid valve annulus. Between six and eight short-axis images of the RV were obtained with slice thicknesses of 8 mm. RV volumes were measured using offline software (Argus; Siemens). After the identification of the end-systolic and enddiastolic phases of the cardiac cycle, the semiautomated detection of endocardial borders was optimized with fine manual adjustment, and RV volumes and EF were calculated. Statistical Analysis The results for RV EDV, RV ESV, and EF are represented as the mean and SDs. Correlations were performed between echocardiography and MRI measurements, and agreement was expressed according to the method of Bland and Altman Ap value of 0.05 was considered to be significant. Z transformations were performed between each group to see whether there was any significant difference between correlations. 19 Data analyses were performed using a statistical software package (SPSS, version 10; SPSS; Chicago, IL). Results Patient Characteristics Of the 54 original patients, 2 were excluded from RT3DE as they were unable to hold their breath during acquisition, and another 2 were excluded from the MRI validation due to claustrophobia. Therefore, the results of the MRI, RT3DE, and 2DE were analyzed in 50 patients (43 men; mean 1846 Original Research

4 age, years). All patients had LV regional wall motion abnormalities, with 70% of patients undergoing angiography. The majority had multivessel coronary artery disease (defined by 70% stenosis in more than one major epicardial vessel). Interest in RV dysfunction was based on either inferior involvement in 41 patients (raising suspicions about RV infarction) or LV dysfunction (Table 1). RV systolic dysfunction (RV EF, 40%) was defined by the visual assessment of an experienced cardiologist. RT3DE slice planes showed that the greater the number of slices, the greater the possibility of capturing the entire structure, although there is a tradeoff with measurement time (Table 2); hence, the use of 12 slices for RV volume measures. Test-Retest Variation of 2DE and RT3DE Test-retest variation was assessed in all 50 patients (Table 3). RT3DE had less test-retest variation of RV volumes and EFs than any 2DE measurements. Volume measurements were more reproducible with RT3DE compared with all 2DE parameters for EDV. The least reproducible was the Simpson method (EDV: r 0.55; p 0.01; ESV: r 0.57; p 0.01), and two-dimensional subtraction (2DS) method showed the greatest absolute difference between measures (EDV: r 0.63; p 0.01; mean difference, 1 22 ml; ESV: r 0.64; p 0.01; mean difference, 1 13 ml). The correlation between sequential EF measurements was more significant for RT3DE than for any 2DE measurements, and the absolute differences were least with RT3DE. Interobserver and Intraobserver Variation of RT3DE and 2DE Table 2 Difference in EDV Using a Different of Cut Planes for RT3DE and Corresponding Time To Complete the Measurement* Planes Difference From MRI-EDV, ml Measurement Time, min 15 cuts cuts cuts cuts *Values are given as the mean SD. Table 4 summarizes the interobserver agreement of volume and EF measurements for each echocardiographic technique. For EDV and EF, the interobserver agreement of RV measurements with RT3DE was similar to that of all 2DE measurements. However, for ESV the agreement with RT3DE was superior to all 2DE parameters (A-L: z 3.24; p 0.01; 2DS: z 2.72, p 0.01; Simpson method: z 4.03; p 0.01). For intraobserver variation (Table 4), RT3DE performed significantly better than the 2DE measurements for all measurements apart from the Simpson method assessment of EDV. Validation With MRI The correlations and mean differences of the echocardiographic techniques and the MRI technique are summarized in Table 5. All 2DE techniques underestimated both RV EDV and RV ESV, except for the 2DS method, which greatly overestimated the RV volumes. RT3DE showed less of a difference from MRI than any of the 2DE techniques. RV EF estimations were similar with each technique; of the echocardiographic methods for EF measurement, the use of RT3DE gave the smallest measurement error and the closest correlation with MRI. Characteristics Table 1 Clinical Characteristics of the Study Patients* Test-Retest Correlation/ MRI Validation (n 50) Interobserver/Intraobserver Reproducibility (n 20) Age, yr Male gender, No Etiology LV regional wall motion abnormalities 50 (100) 20 (100) Global LV dysfunction 2 (4) 1 (2) Inferior wall motion abnormality 41 (82) 15 (30) RV systolic dysfunction (RV EF, 40%) 8 (16) 3 (6) Coronary angiography 35 (70) 20 (100) Left anterior descending artery 22 (63) 15 (75) Left circumflex artery 18 (53) 8 (40) Right coronary artery 25 (71) 15 (75) *Values are given as mean SD or No. (%). CHEST / 131 / 6/ JUNE,

5 Table 3 Correlation and Mean Difference Between the Sequential (Test-Retest) Studies With 2DE Measures and RT3DE (n 50)* Variables RT3DE A-L p Value 2DS p Value Simpson Method p Value EDV Correlation r 0.91, p 0.01 r 0.60, p , z 5.72 r 0.63, p , z 5.39 r 0.55, p , z 6.23 Mean , F , F , F 2.02 difference ESV Correlation r 0.90, p 0.01 r 0.60, p , z 5.34 r 0.64, p , z 4.90 r 0.57, p , z 5.65 Mean p 0.01, F , F , F 4.21 difference EF Correlation r 0.78, p 0.01 r 0.13, p , z 6.27 r 0.33, p , z 4.82 r 0.17, p , z 5.99 Mean difference , F , F , F 3.88 *Values are given as the mean SD, unless otherwise indicated. The correlations between sequential measures were greater with RT3DE than 2DE techniques. A-L vs RT3DE. 2DS vs RT3DE. Simpson method of discs vs RT3DE. Table 4 Interobserver and Intraobserver Agreement With 2DE and RT3DE (n 20) Showing Correlations and Mean Difference Between the Studies* Variables RT3DE A-L p Value 2DS p Value Simpson Method p Value Interobserver agreement EDV Correlation r 0.76, r 0.72, 0.72, z 0.37 Mean difference , F 6.36 ESV Correlation r 0.91, r 0.63, 0.01, z 3.24 Mean difference , F 7.97 EF Correlation r 0.70, r 0.53, 0.27, z 1.14 Mean difference , F 1.62 Intraobserver agreement EDV Correlation r 0.94, r 0.66, 0.01, z 3.90 Mean difference , F 7.96 ESV Correlation r 0.96, r 0.66, 0.01, z 4.75 Mean difference , F 5.47 EF Correlation r 0.83, r 0.59, 0.05, z 2.10 Mean difference , F 1.14 *Values are given as the mean SD, unless otherwise indicated. A-L vs RT3DE. 2DS vs RT3DE. Simpson method vs RT3DE. r 0.75, 0.93, z , F 3.7 r 0.70, 0.01, z , F 3.91 r 0.51, 0.23, z , F 2.07 r 0.82, 0.03, z , F 3.17 r 0.82, p 0.01, z , F 3.73 r 0.59, 0.05, z , F 1.91 r 0.73, 0.78, z , F 4.77 r 0.50, 0.01, z , F 7.69 r 0.48, 0.17, z , F 1.47 r 0.87, 0.11, z , F 6.99 r 0.56, 0.01, z , F 4.86 r 0.52, 0.02, z , F Original Research

6 Table 5 Absolute Measures and Mean Differences Between Echocardiographic and MRI Measurements (n 50) for EDV, ESV, and EF* Variables RT3DE A-L 2DS Simpson Method EDV (87 22 ml) Difference r Value 0.60 (p 0.01) 0.36 (p 0.01) 0.40 (p 0.01) 0.34 (p 0.02) F statistic 0.91 (p 0.37) 3.78, ( 0.01) 5.10 (p 0.01) 3.44 (p 0.01) ESV (46 17 ml) Difference r Value 0.55 (p 0.01) 0.42 (p 0.01) 0.54 (p 0.01) 0.48 (p 0.01) F statistic 1.81 (p 0.02) 6.36 (p 0.01) 3.14 (p 0.01) 7.10 (p 0.01) EF (49 12%) Difference r Value 0.72 (p 0.01) 0.08 (p 0.60) 0.13 (p 0.35) 0.06 (p 0.72) F statistic 0.69 (p 0.10) 2.10 (p 0.01) 2.32 (p 0.01) 1.55 (p 0.07) *Values are given as the mean SD, unless otherwise indicated. Discussion The results of this study of patients with ischemic LV dysfunction and suspicion of RV involvement indicate that RV volume measurements using RT3DE imaging are comparable with MRI. RT3DE also provides low test-retest variation and high reproducibility of RV measurements between observers. RV Assessment by Echocardiography Echocardiography is the most widely used technique for imaging the RV, but assessment can be challenging because of the location of the RV behind the sternum and its crescent shape, wrapped around the LV. 20 Endocardial tracing and the calculation of volumes may be hindered by trabeculations and the presence of the moderator band within the volume. Geometric shapes are too simplistic to be applied as models for RV volume calculation; the Simpson method for the calculation of EFs is more accurate in patients with cor pulmonale. 21 RV size and function may be difficult to assess in patients with pulmonary disease, but assessment plays an important role in clinical decision making for therapy and prognosis. Because of these difficulties in RV volumetric assessment, other echocardiographic modalities have been used to assess RV function. These include tissue Doppler and strain rate imaging, which are sensitive and noninvasive, although the clinical experience with both is limited. Longitudinal strain rate imaging has been shown to be feasible in a clinical setting, but radial imaging is hindered by artifacts and has not been proven to be of clinical value. 22 Pressure-volume loops are the gold standard approach to the load-independent assessment of RV systolic and diastolic function, but they require an invasive procedure. 22 Measurements not only must be accurate but also reproducible. RT3DE has been found to be reproducible for the assessment of LV systolic function in the clinical laboratory 5,23 and might, therefore, be useful for the assessment of RV function. Table 6 summarizes previous validation studies using 3DE of the RV that have used various techniques such as reconstructed array 6 and sparse array, 7 9 and, although some studies used RT3DE, the method for analysis differs. Clearly, reproducibility is limited not only by measurement error but also by extrinsic variables such as preload, afterload, pericardial constraint, and medical therapy. 24 RV Assessment With RT3DE In this study, volumes were underestimated by echocardiography, but less with RT3DE than with any other technique. This systematic underestimation may partly reflect the role of trabeculations (with the use of a particular MRI technique corresponding to ventriculography, thereby filling the space between trabeculations) and may partly reflect the failure of the 2DE methods to depict the RV outflow tract. 2DE uses simplified methods to calculate RV volumes; single-plane approaches do not take into account the true shape of the RV and its irregularities, such as trabeculations and outflow tract. A previous RT3DE comparison showed RV volumes and EF to have only a fair correlation to MRI findings, 10,11 which is probably due to the use of a disk summation approach with manual tracing of the endocardial borders. In contrast, the current study involved modified apical views using a semiautomated approach, which may account for the small difference and good correlation between RT3DE and MRI. Previous studies of volumetric 3DE 8 and reconstructed 3DE 6 have shown that this technique ex- CHEST / 131 / 6/ JUNE,

7 Table 6 Previous RV Volume Validation Studies Using 3DE* Study/Year 3D Technique Correlation Variability Chen et al 12 /2006 RT3DE, disc method 16-plane volume: r 0.97, p Interobserver, r 0.97 Intraobserver, r 0.98 Prakasa et al 10 /2006 RT3DE, disc method EDV: r 0.50, ml Interobserver 95% CI for EDV, 12 to 7 ESV: r 0.72, ml 95% CI for ESV, 5 to4 EF: r 0.88, p 0.01 Intraobserver 95% CI for EDV, 1 to3 95% CI for ESV, 1 to2 Kjaergaard et al 11 /2006 RT3DE, disc method EDV: r 0.56; SEE, 24 ml EF, 8% ESV: r 0.62; SEE, 19 ml EF: r 0.42; Clark et al 25 /2006 Reconstruction EDV, 2 ml; ESV, 2 ml; EF, 8% Schindera et al 9 /2002 Volumetrics, in vitro r 0.99; SEE, 0.93 ml Interobserver, r 0.99 Ota et al 26 /1999 Volumetrics, in vitro SEE, 3.26 ml Interobserver, 4.2% Heusch et al 7 /1999 Volumetrics, animal model r 0.93; SEE, 1.57 ml Interobserver, 4.6% Shiota et al 8 /1998 Volumetrics, animal model r 0.80; 1.6 Interobserver, r 0.87 Pini et al 6 /1997 Reconstruction, animal model r 0.98; SEE, 2.09 ml Intraobserver, 5.2% Interobserver, 8.2% Jiang et al 4 /1994 Reconstruction spark gap in vitro EDV: r 0.99; SEE, 1.8 ml Interobserver, 4.0% ESV: r 0.98; SEE, 2.5 ml Intraobserver, 2.6% *CI confidence interval. hibits higher reproducibility than 2DE. In this study, interobserver variation was optimized by the use of RT3DE, which could reflect the use of a semiautomated edge-detection technique. In contrast to the more widely reported parameters of intraobserver and interobserver variability, test-retest variation is not commonly reported but assumes particular importance in the use of a test in follow-up. RT3DE was found to have less variation and higher correlation to MRI than all 2DE techniques. The major contributor to the volume underestimation of 2DE is its single-plane approach to volumetric measurement. The large variation in the measurement is due to the multiplication of the error in the calculation of the volumes. Limitations The main limitation of this study was the relatively small population (n 50) in whom interest in RV function pertained to ischemic heart disease. These changes are more subtle than those associated with more overt RV pathology such as tetralogy of Fallot, arrhythmogenic RV dysplasia, and pulmonary hypertension. The second limitation was the current shortcoming of RT3DE relating to the quality of images, which are compromised by both line density and frame rate. Difficulty in the discrimination of the endocardial border is an important potential contributor to inaccuracies in the measurement of RV volumes, and image quality was a predictor of discrepancy between 3D and MRI measurements of the change in RV volume. Both limitations of image quality and acquisition volume may be alleviated by technical improvements. Finally, MRI may itself be imperfect for RV assessment. The failure to visualize the tricuspid valve may lead to the overestimation of RV volumes. Most MRI software derives volumes from serial short-axis slices of the RV, which may pose problems because of RV geometry, and gathering a 3D data set may be more effective in measuring such a nongeometric structure. Conclusions In this study, RT3DE was the most accurate echocardiographic technique used for the measurement of RV volumes in subjects with LV dysfunction. It also appears to be a feasible follow-up imaging tool for sequential measurements of the RV volumes and function. References 1 Pfisterer M, Emmenegger H, Soler M, et al. Prognostic significance of right ventricular ejection fraction for persistent complex ventricular arrhythmias and/or sudden cardiac death after first myocardial infarction: relation to infarct location, size and left ventricular function. Eur Heart J 1986; 7: van Straten A, Vliegen HW, Hazekamp MG, et al. Right ventricular function late after total repair of tetralogy of Fallot. Eur Radiol 2005; 15: Original Research

8 3 Nath J, Demarco T, Hourigan L, et al. Correlation between right ventricular indices and clinical improvement in epoprostenol treated pulmonary hypertension patients. Echocardiography 2005; 22: Jiang L, Siu SC, Handschumacher MD, et al. Three-dimensional echocardiography: in vivo validation for right ventricular volume and function. Circulation 1994; 89: Jenkins C, Bricknell K, Hanekom L, et al. Reproducibility and accuracy of echocardiographic measurements of left ventricular parameters using real-time three-dimensional echocardiography. J Am Coll Cardiol 2004; 44: Pini R, Giannazzo G, Di Bari M, et al. Transthoracic threedimensional echocardiographic reconstruction of left and right ventricles: in vitro validation and comparison with magnetic resonance imaging. Am Heart J 1997; 133: Heusch A, Koch JA, Krogmann ON, et al. Volumetric analysis of the right and left ventricle in a porcine heart model: comparison of three-dimensional echocardiography, magnetic resonance imaging and angiocardiography. Eur J Ultrasound 1999; 9: Shiota T, Jones M, Chikada M, et al. Real-time threedimensional echocardiography for determining right ventricular stroke volume in an animal model of chronic right ventricular volume overload. Circulation 1998; 97: Schindera ST, Mehwald PS, Sahn DJ, et al. Accuracy of real-time three-dimensional echocardiography for quantifying right ventricular volume: static and pulsatile flow studies in an anatomic in vitro model. J Ultrasound Med 2002; 21: Prakasa KR, Dalal D, Wang J, et al. Feasibility and variability of three dimensional echocardiography in arrhythmogenic right ventricular dysplasia/cardiomyopathy. Am J Cardiol 2006; 97: Kjaergaard J, Petersen CL, Kjaer A, et al. Evaluation of right ventricular volume and function by 2D and 3D echocardiography compared to MRI. Eur J Echocardiogr 2006; 7: Chen G, Sun K, Huang G. In vitro validation of right ventricular volume and mass measurement by real-time three-dimensional echocardiography. Echocardiography 2006; 23: Helbing WA, Bosch HG, Maliepaard C, et al. Comparison of echocardiographic methods with magnetic resonance imaging for assessment of right ventricular function in children. Am J Cardiol 1995; 76: Panidis IP, Ren JF, Kotler MN, et al. Two-dimensional echocardiographic estimation of right ventricular ejection fraction in patients with coronary artery disease. J Am Coll Cardiol 1983; 2: Tomita M, Masuda H, Sumi T, et al. Estimation of right ventricular volume by modified echocardiographic subtraction method. Am Heart J 1992; 123: Bland JM, Altman DG. Correlation, regression, and repeated data. BMJ 1994; 308: Bland JM, Altman DG. Calculating correlation coefficients with repeated observations: part 2. Correlation between subjects. BMJ 1995; 310: Bland JM, Altman DG. Calculating correlation coefficients with repeated observations: part 1. Correlation within subjects. BMJ 1995; 310: Bland M. An introduction to medical statistics. 2nd ed. Oxford, UK: University Press Inc, Ho SY, Nihoyannopoulos P. Anatomy, echocardiography, and normal right ventricular dimensions. Heart 2006; 92(suppl): i2 i13 21 Vitolo E, Castini D, Colombo A, et al. Two-dimensional echocardiographic evaluation of right ventricular ejection fraction: comparison between three different methods. Acta Cardiol 1988; 43: Bleeker GB, Steendijk P, Holman ER, et al. Assessing right ventricular function: the role of echocardiography and complementary technologies. Heart 2006; 92(suppl):i19 i26 23 Jacobs LD, Salgo IS, Goonewardena S, et al. Rapid online quantification of left ventricular volume from real-time threedimensional echocardiographic data. Eur Heart J 2006; 27: Hines R. Right ventricular function and failure: a review. Yale J Biol Med 1991; 64: Clark TJ, Sheehan FH, Bolson EL. Characterizing the normal heart using quantitative three-dimensional echocardiography. Physiol Meas 2006; 27: Ota T, Fleishman CE, Strub M, et al. Real-time, threedimensional echocardiography: feasibility of dynamic right ventricular volume measurement with saline contrast. Am Heart J 1999; 137: CHEST / 131 / 6/ JUNE,