First-Trimester Fetal Cardiac Function Measurements Using Spatio-Temporal Image Correlation and Two Ultrasound-Related Post-Processing Methods: A Feasibility and Reproducibility Study.

Abstract

Objective: To study the feasibility and reproducibility of two ultrasound (US) related post-processing methods for first-trimester fetal cardiac function assessment by ventricle volume measurements.

Method: First-trimester transvaginal Spatio-Temporal Image Correlation (STIC) US datasets acquired between 11⁺⁰-13⁺⁶ weeks gestational age (GA) were used to perform fetal cardiac ventricle volume (FCVV) measurements in the end-diastolic (EDVV) and end-systolic (ESVV) phases using two methods: the manual segmentation method Virtual Organ Computed-Aided AnaLysis (VOCAL) and (semi-)automated volume measuring method Virtual Reality (VR). Reproducibility was assessed by calculating the intra-, interobserver and intersystem agreement using intraclass correlation coefficients (ICCs) followed by Bland-Altman plots.

Results: 25 STIC US datasets were selected. The mean GA was 13⁺⁰ weeks (SD 2.3 days) and mean crown-rump length was 68.0 mm (range 61.0-75.6 mm). The intra- and inter-observer agreement for both methods resulted in good to excellent agreement (ICCs > 0.85). Mean relative differences for all FCVV measurements were < 10.0%, except for the inter-observer agreement of the VOCAL ESVV measurement (40.6%). The inter-system agreement showed poor to moderate agreement (ICCs 0.32-0.75) and moderate to good agreement (ICCs 0.62-0.78) in terms of absolute agreement and consistency, respectively.

Conclusion: FCVV measurements performed in STIC US datasets using VR are feasible and reproducible, specifically when compared to VOCAL.