Reproducibility of fetal lung volume measurements with 3-dimensional ultrasonography

To build a nomogram of normal fetal lung volumes and to assess the reproducibility of measurements using 3-dimensional ultrasonography.

Inclusion criteria were healthy women, singleton normal pregnancies, reliable dating, and 20 to 30 weeks' gestation. Exclusion criteria were discordance between clinical and ultrasonographic dating, patients lost to follow-up, and birth weight disorders. Patients were scanned at intervals longer than 2 weeks. Three volumes were acquired for each patient; only data from the volume with the best image quality was used for analysis. Volumes were rated and measured by the manual tracing method. We recorded whether the clavicle was visualized. Only good-quality volumes were included in analysis. The best volume was chosen, and each lung was measured.

A total of 75 patients were studied over a 9-month period, from which 182 volumes were analyzed. Of the 182 volumes, 15 (8.2%) were excluded for poor quality. The remaining 167 volumes were included in the final analysis. In 83 volumes (50%), the clavicle was not visualized. The best fit for total lung volume was a second-degree polynomial regression curve. Lung volume was 10.28 mL at 20 weeks and 51.49 mL at 30 weeks. Assessment of agreement was studied by selection of 40 volumes. Intraobserver variability was 5.48 mL (10.6%) and 3.07 mL (5.96%). Interobserver variability was 7 mL.

Our findings suggest that 3-dimensional ultrasonographically derived measurements are reliable and reproducible up to 30 weeks if a standard measurement technique is used.