HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Remy-Jardin, M. Articles by Remy, J. Search for Related Content PubMed PubMed Citation Articles by Remy-Jardin, M. Articles by Remy, J. Social Bookmarking                      What's this?

MDCT of Right Ventricular Function: Impact of Methodologic Approach in Estimation of Right Ventricular Ejection Fraction, Part 2

¹ Department of Thoracic Imaging, Hospital Calmette, University Center of Lille, Blvd. Jules Leclerq, Lille 59037, France.
² Department of Nuclear Medicine, Hospital Roger Salengro, Lille 59037, France.
³ Department of Medical Statistics, University of Lille, Lille 59037, France.

Received July 11, 2005; accepted after revision February 1, 2006.

 
Address correspondence to M. Remy-Jardin (mremy-jardin{at}chru-lille.fr).

Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
OBJECTIVE. The purpose of this study was to evaluate the impact of the methodologic approach for MDCT estimation of right ventricular ejection fraction (RVEF).

MATERIALS AND METHODS. In 49 consecutive patients (30 men, 19 women; mean age, 59 years) known to have or suspected of having right ventricular (RV) dysfunction secondary to pulmonary disease, 16-MDCT of the heart was performed after standard CT angiographic examination of the entire thorax, with determination of RVEF by two reviewers who had limited experience in cardiac CT. The reconstruction windows were determined using the ECG tracing (reviewer 1) or using transverse test images obtained in 5% steps through the entire R-R interval showing the largest and smallest RV cavity areas (reviewer 2). After manual segmentation of the ventricular cavity on diastolic and systolic short-axis reformations by each reviewer, the end-diastolic and end-systolic RV volumes were calculated, with subsequent determination of the RVEF. CT results were compared with those of equilibrium radionuclide ventriculography.

RESULTS. Agreement between the two methods for determining the end-systolic and end-diastolic phases was observed in 61% of cases (n = 30) for the systole and 59% of cases (n = 29) for the diastole. Discordant selections were observed in 39% of cases (n = 19) for determination of the systole and in 41% of cases (n = 20) for determination of the diastole, ranging from 5% to 15% of the R-R interval, suggesting that selection of the reconstruction window on the ECG tracing does not differ significantly from that obtained by the visual analysis of transverse test images. Focusing on the 59 common selections of the reconstruction windows made by the two reviewers, no statistically significant differences were found in the determination of mean (± SD) end-diastolic volumes (reviewer 1, 176.21 ± 67 mL vs reviewer 2, 175.55 ± 71.24 mL; p = 0.98) and end-systolic (reviewer 1, 97.3 ± 26.49 mL vs reviewer 2, 96.33 ± 65.72 mL; p = 0.65), suggesting the lack of operator dependence in the manual-contour drawing process. No significant difference was found between the mean values of RVEF obtained by each reviewer with MDCT and equilibrium radionuclide ventriculography, and there was excellent interobserver agreement with MDCT (intraclass correlation coefficient, 0.86). Using a Bland-Altman approach, the limits of concordance between the two reviewers ranged between -10.2 and 10.9. The mean absolute percentage error for measuring RVEF between the two reviewers was 9.7%. A moderate agreement was found between RVEFs obtained on CT by each reviewer and scintigraphy (intraclass correlation coefficients, 0.76 for reviewer 1 and 0.64 for reviewer 2).

CONCLUSION. These results show that RVEF can be accurately assessed with ECG-gated MDCT using commercially available software.

Keywords: cardiac imaging • chest • CT • lung diseases • MDCT angiography

Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The introduction of MDCT scanners has led to new possibilities in the field of chest imaging. Previously limited to the depiction of morphologic changes at the level of the thoracic organs, MDCT examinations now offer the possibility of providing information on cardiac function by means of the availability of retrospective ECG-gating techniques and developments in dedicated cardiac analysis software. Although most attention was initially directed toward validation of MDCT for left ventricular function analysis [1-3], three clinical studies [4-6] have recently investigated the accuracy of MDCT in the assessment of right ventricular (RV) function, which is vital information in the management of patients who have a wide variety of acute or chronic respiratory disorders [7].

Lembcke et al. [4] were the first authors to report their experience with 8- and 16-MDCT scanners to measure RV dimensions and function in a population of 25 patients before cardiac surgery. Comparing MDCT with MRI, these authors found that MDCT was an accurate and reliable noninvasive technique for evaluating RV measurements. In a population of 19 patients, Koch et al. [5] reached similar conclusions using 16-MDCT. More recently, Delhaye et al. [6] calculated the right ventricular ejection fraction (RVEF) of 49 consecutive patients using ECG-gated MDCT and compared the results with those of equilibrium radionuclide ventriculography. Their investigation led to the conclusion that a reliable estimation of RVEF can be obtained with helical 16-MDCT in unselected patients. Because MDCT appears to represent a promising noninvasive approach to investigating RV function, it is mandatory to evaluate the impact of the methodologic approach of such calculations. The purpose of this study was to investigate the variability of measurements of RV function using ECG-gated 16-MDCT in comparison with equilibrium radionuclide ventriculography, which was used as the standard of reference.

Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Data Acquisition
During a 5-month period from January through May 2004, 49 consecutive patients (30 men, 19 women) with a mean age of 59 ± 14 years (age range, 24-84 years) known to have or suspected of having RV dysfunction secondary to bronchopulmonary disease (n = 30) or pulmonary vascular disease (n = 19) underwent MDCT of the chest on a 16-MDCT scanner (Sensation 16, Siemens Medical Solutions). The CT protocol was designed to allow the radiologist to provide morphologic information on the entire thorax, complemented by information on RV function. This biphasic protocol comprised a non-ECG-gated CT angiographic examination of the entire thorax (the diagnostic scan) with the following scanning parameters: 80-120 kV; 60-100 mAs; rotation time, 0.5 second; collimation, 16 x 0.75 mm; pitch, 1.5 followed by an ECG-gated CT angiographic examination of the heart (the functional scan) with the following scanning parameters: rotation time, 0.42 second; 120 kV; 300 mAs; collimation, 12 x 0.75 mm; pitch, 0.2. The acquisition, injection, and reconstruction parameters for both scans have been detailed in a previous study, in which good agreement was found between MDCT and equilibrium radionuclide ventriculography for the assessment of RVEF in unselected patients [6]. This preliminary investigation was supervised by two reviewers with 1 year each of experience in cardiac CT at the time of the initiation of this study. The reviewers determined the necessary steps for RVEF calculation by consensus.

Image Processing Method
The calculation of RVEF with MDCT is a 5-step process: step 1, selection of the reconstruction windows; step 2, reconstruction of systolic and diastolic transverse CT scans of the cardiac cavities; step 3, reconstruction of short-axis images of the cardiac cavities during the systolic and diastolic phases; step 4, segmentation of the RV cavity; and step 5, calculation of RV volumes and RVEF. Whereas steps 2, 3, and 5 are automated procedures using the scanner's standard software, the selection made by the operator during steps 1 and 4 can potentially influence the overall calculation of RVEF. In selecting the temporal window for reconstruction of systolic and diastolic im- ages—step 1—two techniques can be used. The first is based on the identification of end-diastolic and end-systolic phases with ECG tracings. The second method is the identification of the maximal systolic contraction and diastolic relaxation phases on a series of transverse test images obtained at the midventricular level, acquired in 5% steps through the entire R-R interval. Step 4—segmentation of the RV cavity— requires manual delineation of the RV endocardial contours on short-axis images of the right ventricle at end-diastole and end-systole.

Study Design
The first objective of this study was to investigate the influence of operator technique on the determination of the reconstruction window for the diastolic and systolic phases. To reach this goal, reviewer 1 made the selection using the ECG tracing, whereas reviewer 2 used test images; each reviewer chose one interval among the 5% steps proposed through the entire R-R interval. Their respective selections, made for the entire study group, were compared to determine the number of concordant selections. In cases of discordant selections, the range of variations was systematically analyzed.

The second objective was to analyze the influence of interobserver variability in the manual segmentation of the RV cavity. This was investigated by comparing end-systolic and end-diastolic RV volumes obtained by each reviewer and focusing this analysis on the concordant selections of the reconstruction windows by both reviewers.

The third objective was to provide an overall estimation of the variability of RVEF measurements with CT. This was investigated by comparing the mean RVEF measurement obtained by each reviewer on MDCT with that obtained on equilibrium radionuclide ventriculography in the entire study group. Figure 1 summarizes the methodologic approach followed by each reviewer to evaluate this variability.

View larger version (20K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1 —Flowchart shows methodologic approach followed by each reviewer to evaluate interobserver variability in assessment of right ventricular ejection fraction (RVEF) with MDCT. First step was to investigate influence of operator technique on determination of reconstruction window for diastolic and systolic phases, assessed by comparing concordant and discordant selections between reviewer 1 and reviewer 2 in step 1, reconstruction window. Interobserver variability in manual segmentation process was evaluated by comparing end-systolic (ESV) and end-diastolic (EDV) right ventricular volumes obtained by each reviewer after concordant selections of reconstruction windows by both reviewers (step 2, segmentation process).The variability of RVEF measurements with CT was investigated by comparing each reviewer's measurement on MDCT with that obtained on equilibrium radionuclide ventriculography (step 3, MDCT vs scintigraphy). Agreements between two reviewers and between MDCT and scintigraphy were assessed by means of intraclass correlation coefficient. Agreement was also assessed by Bland-Altman method and by mean absolute percent error.

Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Selection of Systolic and Diastolic Phases by the Two Reviewers
Among the 98 selections made by the two reviewers to determine end-systolic and end-diastolic phases in our study group of 49 patients, an agreement between the two methods was observed in 59 cases (60%), and discordant selections were observed in 39 cases (40%). In 35 discordant selections (90%), the variation ranged between 5-10% of the R-R interval; in four discordant selections (10%), a 15% variation was observed between the two methods. Table 1 summarizes detailed results for the selection of each phase.

Influence of the Segmentation Process
A comparison of end-systolic and end-diastolic volumes obtained by each reviewer was made for the 59 concordant selections of the reconstruction windows. The mean end-systolic RV volume was 97.3 ± 26.49 mL (range, 44-399 mL) for reviewer 1 and 96.33 ± 65.72 mL (range, 44-417 mL) for reviewer 2; no significant difference was found between the two mean values (paired Student's t test, p = 0.65). The mean end-diastolic RV volume was 176.21 ± 67 mL (range, 93-455 mL) for reviewer 1 and 175.55 ± 71.24 mL (range, 107-487 mL) for reviewer 2; no significant difference was found between the two mean values (paired Student's t test, p = 0.98).

View larger version (9K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2 —Bland-Altman plot of mean difference of right ventricular ejection fractions between two reviewers. Broken line represents mean difference. Solid lines represent the 95% confidence intervals for mean difference.

The agreement between reviewer 1 and reviewer 2 was good (ICC, 0.86). The dispersion of differences of RVEFs assessed by each reviewer is illustrated with a Bland-Altman plot (Fig. 2). Only four values (8% of the data) were outside the limits. The limits of concordance for MDCT ranged between -10.2 and 10.9. The percentage of variability between two measurements expressed by the MAPE was 9.7%. These results revealed small interobserver variability among the MDCT measurements.

The agreement between each reviewer and equilibrium radionuclide scintigraphy was rated as moderate. Reviewer 1 showed better correlation to equilibrium radionuclide scintigraphy than reviewer 2 (ICC for reviewer 1, 0.76; ICC for reviewer 2, 0.64).

Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
A widely available automated imaging technique for assessing RV volume and ejection fraction could be important in the management of patients with respiratory disorders that can be potentially responsible for impairment of RV performance. Because MDCT potentially represents an imaging tool that can accomplish this purpose [4-6], it is important to evaluate the impact of the CT data analysis protocol. Selection of the reconstruction windows for the diastolic and systolic phases was the first practical aspect to investigate. Two methods are currently available, one based on the patient's ECG tracing and the second on the identification of transverse test images showing the largest and smallest RV cavities.

Our comparative analysis involved two reviewers, each with 1 year of experience in cardiac CT, who selected the reconstruction windows of the 49 MDCT examinations in a blinded fashion, leading to a total of 98 selections for each reviewer. We observed a similar selection by both reviewers in 59 cases (60%) and noted that among the 39 discordant selections, 90% (35/39) varied between 5-10% of the R-R interval. The impact of discordant selections was indirectly assessed by comparing the mean values of RVEF obtained by each reviewer that failed to show any significant differences. Consequently, we can suggest that the method used, which can be based either on the ECG tracing or on the analysis of test images, has a minimal impact in determining the temporal reconstruction window.

The second methodologic aspect investigated in the present study was the influence of the RV cavity segmentation process. The software evaluated in our study is the same tool as that used for left ventricular function analysis, itself adapted from MRI analysis software that has been validated in research and clinical studies for the past decade [3]. However, the main difference between MDCT estimation of left ventricular ejection fraction and RVEF relies on the use of software enabling semiautomated contour detection for the left ventricle, whereas delineation of RV contours is a manual process.

To ensure that interobserver variability in the segmentation process would not be affected by the selection of the reconstruction windows made by each reviewer, we limited this comparative analysis to the 59 concordant selections of the systolic and diastolic phases. We failed to observe any significant difference between the mean end-systolic and end-diastolic volumes measured by the two reviewers, suggesting a lack of operator dependence in the manual-contour drawing process. These results are of practical importance because of the necessity of segmenting the right ventricle on 2D images instead of 3D reconstructions [5, 10]. Assessing RV function with 16-MDCT in comparison with MRI, Koch et al. [5] have recently shown that threshold-supported 3D reconstructions revealed insufficient correlation with MRI. With regard to the delineation of RV contours on short-axis images, one should underline the importance of an injection protocol aimed at obtaining sufficient enhancement of the RV cavity without streak artifacts.

Because our objective was to evaluate the possibility of providing functional information from an MDCT angiogram of the chest, we favored a biphasic administration of contrast medium. The first phase was similar to that of a standard CT angiogram of the chest, namely, the administration of 90 mL of a 30% contrast agent at 3 mL/s followed by the administration of a limited amount of contrast medium—30 mL at a low flow rate of 1.5 mL/s—during the ECG-gated acquisition over cardiac cavities. When the MDCT examination is exclusively dedicated to the assessment of RV function, one can logically recommend the use of the system's bolus tracking option with the region of interest positioned in the pulmonary artery [4] or ascending aorta [5].

The third aspect investigated by the present study was the overall interobserver variability in the calculation of RVEFs using the scanner's standard software. This objective was assessed by comparing RVEF measurements obtained by each reviewer using MDCT in the entire study group to those of equilibrium radionuclide scintigraphy. We found no significant differences in the mean values of RVEF obtained by each reviewer with MDCT and those of equilibrium radionuclide ventriculography, and there was excellent interobserver agreement with MDCT. A Bland-Altman approach was also used to visualize the differences between reviewers. The limits of concordance ranged between -10.2 and 10.9. The percentage of variability between two measurements was also expressed by the mean absolute percent error, which was 9.7% between the two reviewers. These results show a high level of reproducibility of RVEF measurements with MDCT and, subsequently, the limited impact of the two operator-dependent steps of the CT data analysis protocol.

To our knowledge, Lembcke et al. [4] were the first authors to document the interobserver variability for RV measurements with MDCT, comparing MDCT measurements with MRI. Evaluating this variability with different parameters from those used in our study, they reported an interobserver variability of 5% for calculation of the RVEF with MDCT, which was quite similar to that obtained with MRI, a technique known to provide accurate and reproducible measurement results even in cases with substantially deformed right ventricles [11-13]. However, Grothues et al. [14] have recently pointed out that the reproducibility of RV parameters with cardiac MRI was lower than for left ventricular volumes.

The concordance between MDCT and scintigraphy was rated as moderate, and this finding should be interpreted in light of the well-known limitations of equilibrium cardiac scintigraphy. As previously emphasized [15], the main disadvantage of radionuclide angiography is that it is a projection method, not a tomographic one, which means that it provides much less anatomic information than competing methods. Therefore, it is not surprising to observe limitations in the RVEF calculations due to unclear separation between the right atrium and right ventricle [16-18]. For RVEF assessment, observers draw RV outlines, primarily guided by the visual impression of the RV shape seen at end-diastole [19]. If the cardiac chambers overlap in the equilibrium technique, the calculated ejection fraction will be erroneously low [17]. Moreover, whereas MDCT enables precise identification of the RV valvular borders [5], the second obstacle to equilibrium radionuclide scintigraphy is the sometimes inadequate delineation of the pulmonic valve plane [16]. These reasons account for the underestimation of RVEFs with equilibrium radionuclide ventriculography compared with SPECT equilibrium radionuclide angiography or with MRI [18]. Despite these limitations, equilibrium radionuclide ventriculography remains a useful gauge of global RV performance [20].

In conclusion, our study shows that RVEF can be accurately assessed with ECG-gated MDCT using commercially available software.

References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Juergens KU, Grude M, Fallenberg EM, et al. Using ECG-gated multidetector CT to evaluate global left ventricular myocardial function in patients with coronary artery disease. AJR2002; 179:1545 -1550[Abstract/Free Full Text]
2. Dirksen MS, Bax JJ, de Roos A, et al. Usefulness of dynamic multislice computed tomography of left ventricular function in unstable angina pectoris and comparison with echocardiography. Am J Cardiol 2002; 90:1157 -1160[CrossRef][Medline]
3. Juergens KW, Grude M, Maintz D, et al. Multi-detector row CT of left ventricular function with dedicated analysis software versus MR imaging: initial experience. Radiology 2004;230 : 403-410[Abstract/Free Full Text]
4. Lembcke A, Dohmen PM, Dewey M, et al. Multislice computed tomography for preoperative evaluation of right ventricular volumes and function: comparison with magnetic resonance imaging. Ann Thorac Surg 2005; 79:1344 -1351[Abstract/Free Full Text]
5. Koch K, Oellig F, Oberholzer K, et al. Assessment of right ventricular function by 16-detector-row CT: comparison with magnetic resonance imaging. Eur Radiol 2005;15 : 312-318[CrossRef][Medline]
6. Delhaye D, Remy-Jardin M, Teisseire A, et al. MDCT of right ventricular function: comparison of right ventricular ejection fraction estimation and equilibrium radionuclide scintigraphy, Part 1. AJR 2006; 187:1605 -1609[Abstract/Free Full Text]
7. Oldershaw P. Assessment of right ventricular function and its role in clinical practice. Br Heart J 1992;68 : 12-15[Medline]
8. Fleiss JL. Design and analysis of clinical experiments. New York, NY: Wiley, 1986:1 -32
9. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet1986; 1:307 -310[CrossRef][Medline]
10. Alfakih K, Plein S, Bloomer T, Jones T, Ridgway J, Sivananthan M. Comparison of right ventricular volume measurements between axial and short axis orientation using steady-state free precession magnetic resonance imaging. J Magn Reson Imaging 2003;18 : 25-32[CrossRef][Medline]
11. Helbing WA, Bosch HG, Malipaart C, et al. Comparison of echocardiographic methods with magnetic resonance imaging for assessment of right ventricular function in children. Am J Cardiol1995; 76:589 -594[CrossRef][Medline]
12. Niwa K, Uchishiba M, Aotsuka H, et al. Measurement of ventricular volumes by cine magnetic resonance imaging in complex congenital heart disease with morphologically abnormal ventricles. Am Heart J1996; 131:567 -575[CrossRef][Medline]
13. Lorentz CH, Walker ES, Morgan VL, et al. Normal human right and left ventricular mass, systolic function, and gender differences by cine magnetic resonance imaging. J Cardiovasc Magn Reson1999; 1:7 -21[Medline]
14. Grothues F, Moon JC, Bellenger NG, Smith GS, Klein HU, Pennell OJ. Interstudy reproducibility of right ventricular volumes, function, and mass with cardiovascular magnetic resonance. Am Heart J2004; 147:218 -223[CrossRef][Medline]
15. Araoz PA. Computed tomography for functional evaluation of the heart. Semin Roentgenol 2003;39 : 303-308
16. Schulman DS. Assessment of the right ventricle with radionuclide techniques. J Nucl Cardiol 1996;3 : 253-264[CrossRef][Medline]
17. Vizza CD, Lynch JP, Ochoa LL, Richardson G, Trulock EP. Right and left ventricular dysfunction in patients with severe pulmonary disease. Chest 1998; 113:576 -583
18. Nichols K, Saouaf R, Ababneh AA, et al. Validation of SPECT equilibrium radionuclide angiographic right ventricular parameters by cardiac magnetic resonance imaging. J Nucl Cardiol2002; 9:153 -160[CrossRef][Medline]
19. Maddahi J, Berman DS, Matsuoka DT, et al. A new technique for assessing right ventricular ejection fraction using rapid multiple-gated equilibrium cardiac blood pool scintigraphy: description, validation and findings in chronic coronary artery disease. Circulation 1979;60 : 581-589
20. Zaret BL, Wackers FJ. Nuclear cardiology (second of two parts). N Engl J Med 1993;329 : 855-863[Free Full Text]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				J. M. Kerl, J. G. Ravenel, S. A. Nguyen, P. Suranyi, C. Thilo, P. Costello, W. Bautz, and U. J. Schoepf Right Heart: Split-Bolus Injection of Diluted Contrast Medium for Visualization at Coronary CT Angiography Radiology, May 1, 2008; 247(2): 356 - 364. [Abstract] [Full Text] [PDF]

				D. Delhaye, M. Remy-Jardin, A. Teisseire, C. Hossein-Foucher, S. Leroy, A. Duhamel, and J. Remy MDCT of Right Ventricular Function: Comparison of Right Ventricular Ejection Fraction Estimation and Equilibrium Radionuclide Ventriculography, Part 1. Am. J. Roentgenol., December 1, 2006; 187(6): 1597 - 1604. [Abstract] [Full Text] [PDF]

				M. Remy-Jardin, D. Delhaye, A. Teisseire, C. Hossein-Foucher, A. Duhamel, and J. Remy MDCT of Right Ventricular Function: Impact of Methodologic Approach in Estimation of Right Ventricular Ejection Fraction, Part 2. Am. J. Roentgenol., December 1, 2006; 187(6): 1605 - 1609. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Remy-Jardin, M. Articles by Remy, J. Search for Related Content PubMed PubMed Citation Articles by Remy-Jardin, M. Articles by Remy, J. Social Bookmarking                      What's this?