MDCT of right ventricular function: impact of methodologic approach in estimation of right ventricular ejection...

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

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.

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).

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