EP01.45: Repeatability and reproducibility of measurements of fetal central nervous system obtained using artificial intelligence‐based SonoCNS in the third trimester of pregnancy

The aim of the study was to evaluate the intra- and interobserver variability of biometric measurements of the fetal head (HC, BPD, OFD, CM, TCD, Vp) obtained using SonoCNS software. Two sonographers conducted manual and then twice-automated measurements using SonoCNS software, capturing 3D volume post-transventricular plane display. Intraobserver variability was gauged by intraclass correlation coefficients between manual and automated averages. Interobserver variability was assessed via ICC between automated measurements. The study involved third trimester women aged 28-33 weeks. The study included 111 patients. The results were presented in table 1. We achieved good and excellent reproducibility and repeatability for HC, BPD, and OFD, with all obtained values within the range of 0.75-1. For TCD measurements, the results were worse and fell within the range of 0.75-0.9. For Vp and CM measurements, the ICC was in the moderate range (0.5 – 0.75), although for some measurements, it could not be proven that the ICC was different from 0.5 (confidence intervals overlap with 0.5). The automated system for performing biometric measurements of the fetal head and CNS structures demonstrates good performance in the range of BPD, HC, OFD, and TCD measurements of the fetus. However, it requires improvements in the area of Vp and CM measurements.