Reliability of fetal cardiac volumetry using spatiotemporal image correlation: assessment of in-vivo and in-vitro measurements.

OBJECTIVE

To assess the reliability of measurement of fetal cardiac ventricular volume, stroke volume, and ejection fraction with four-dimensional ultrasound using spatiotemporal image correlation (STIC).

METHODS

Volume datasets were collected from two sources: 24 from fetuses over a range of gestational ages and 12 from a miniature balloon model. Datasets were analyzed by three observers, repeatedly in 12 fetal datasets and all balloon datasets. Volume calculations were obtained by manually tracing multiple parallel slices (three-dimensional (3D) slice method). Measurement error was assessed by calculating standard errors of measurement (SEM) and coefficients of variation (CV). Reliability was assessed by calculating interobserver and intraobserver intraclass correlation coefficients (ICC).

RESULTS

Measurement errors of balloon volumes were small and reliability was good (SEM <or= 0.07 mL, ICC 0.98-1.00). Fetal ventricle volume measurement error ranged from 0.09 to 0.20 mL and CV from 14.6 to 28.3%. Ventricular volume reliabilities for intra- and interobserver comparisons were greater than or equal to 0.94 and 0.75, respectively. Fetal stroke volume measurement error (SEM 0.17 mL), CV (21.9%) and reliability were measured (intraobserver ICC: left ventricle stroke volume (LVSV), 0.93 vs. right ventricle stroke volume (RVSV), 0.88; interobserver ICC: LVSV, 0.75 vs. RVSV, 0.86). The measurement error decreased with increasing operator experience. The reliability of ejection fraction calculations was poor (ICC < 0.7) for intra- and interobserver comparisons.

CONCLUSIONS

Volume measurements obtained with STIC and 3D slice methods using a balloon model were reliable. In the fetus, measurement errors decreased with increasing operator experience, and reliability was better for stroke volume than for ejection fraction.