Reproducibility of echocardiographic measurements of human fetal left ventricular volumes and ejection fractions using four-dimensional ultrasound with the spatio-temporal image correlation modality

Abstract

Objectives: To determine the reproducibility, both reliability and agreement, of measurements of fetal left ventricular parameters from volumes obtained by spatio-temporal image correlation (STIC) acquisition applying virtual organ computer-aided analysis (VOCAL) and Simpson's rule (method of discs). Furthermore the success rate of STIC acquisition was determined.

Study design: In 84 pregnancies between 20 and 34 weeks of gestation the fetal heart was scanned using the STIC modality. An optimal four-chamber view in end-diastole and end-systole was obtained. Left ventricular end-diastolic volume, left ventricular end-systolic volume, stroke volume and ejection fraction were determined. For calculations based on Simpson's rule only one plane was traced, whereas for VOCAL six planes were traced. To quantify the reliability intraclass correlation coefficients were calculated for both intra- and inter-observer measurements. Agreement of measurements was evaluated by Bland-Altman plots.

Results: The STIC volumes of 54 women (64%) were excluded from the study because of poor quality, leaving 30 volumes for further analysis. Intraclass correlation coefficients for intra-observer reliability for VOCAL and Simpson were 0.99 and 0.99 for left ventricular end-diastolic volume, 0.95 and 0.92 for left ventricular end-systolic volume, 0.98 and 0.97 for stroke volume, 0.76 and 0.77 for ejection fraction, respectively. Intraclass correlation coefficients for inter-observer reliability for VOCAL and Simpson were 0.97 and 0.86 for left ventricular end-diastolic volume, 0.97 and 0.86 for left ventricular end-systolic volume, 0.95 and 0.81 for stroke volume, 0.68 and 0.63 for ejection fraction, respectively. According to Bland-Altman plots, the mean percentage difference and 95% limits of intra- and inter-observer agreement for left ventricular stroke volume measurements using VOCAL were -0.2 (-25.1, 24.7)% and 2.8 (-34.2, 39.8)%, respectively. For left ventricular stroke volume measured with Simpson versus VOCAL the mean percentage difference and 95% limits of agreement were -1.8 (-22.1, 18.5)%.

Conclusions: 4D STIC enables reproducible measurements of left ventricular volumes. Reliability of the VOCAL mode is not essentially different from the single-plane method used in Simpson's rule. The large percentage of poor quality STIC volumes and the wide limits of inter-observer agreement would create obstacles for the clinical applicability of this technique.