Artificial intelligence assistance for fetal head biometry: Assessment of automated measurement software

Purpose: To evaluate the feasibility and reproducibility of artificial intelligence software (Smartplanes (R)) to automatically identify the transthalamic plane from 3D ultrasound volumes and to measure the biparietal diameter (BPD) and head circumference (HC) in fetus. Material and methods: Thirty fetuses were evaluated at 17-30 weeks' gestation. For each fetus two three-dimensional (3D) volumes of the fetal head along with one conventional two-dimensional (2D) image of the transthalamic plane were prospectively acquired. The Smartplanes (R) software identified the transthalamic plane from the 3D volumes and performed BPD and HC measurements automatically (3D auto). Two experienced sonographers also measured BPD and HC from 2D images and from the 3D volumes. Measurements were compared using Bland-Altman plots. Interclass correlation coefficient (ICC) was used to evaluate intra and interobserver reproducibility. Results: For each series of measurements, intra-and interobserver reproducibility rates were high with ICC values > 0.98. The 95% confidence intervals between the BPD measurements were 2 mm (3D versus 2D) and 4 mm (3D auto versus 2D) and the HC measurements were 7.5 mm (3D versus 2D) and 11 mm (3D auto versus 2D). Conclusion: Fetal head measurements obtained automatically by Smartplanes (R) software from 3D volumes show good agreement with those obtained by two experienced sonographers from conventional 2D images and 3D volumes. The reproducibility of these measurements is similar to that observed by experienced sonographers. (C) 2018 Societe francaise de radiologie. Published by Elsevier Masson SAS. All rights reserved.