Description and purpose
The determination of the mode of delivery when an operative delivery is indicated in the second stage of labor is a critical issue in Obstetrics. Fetal occiput position and station are the main determinants of the choice between cesarean and instrumental delivery. Such parameters can be more accurately defined by means of intrapartum ultrasound, which has also been shown to predict the outcome of instrumental delivery.
Purpose
Our research group has developed algorithms allowing the automatic assessment of the fetal occiput position and station from transperineal sonographic axial acquisition. The aim of this project is to prospectively evaluate the performance of an artificial intelligence-based system allowing the integration of the information concerning occiput position and station acquired using one single axial sonographic insonation in predicting the outcome of instrumental delivery in the second stage.
Expected results
The proposed algorithm enables the automated integration of the sonographically assessed head position and station and provide a traffic light-like output in terms of delivery color code with respect to the prediction of the outcome of the operative delivery: 1)GREEN: easy fetal extraction; 2)YELLOW: challenging fetal extraction; RED: extraction procedure with high chance to fail. The proposed algorithm will undergo first a training phase and then a validation phase on a different cohort.
Achieved results
The training phase is now concluded. Preliminary results suggest the feasibility of the prediction of the outcome of operative delivery and a fair accuracy of the so far developed AI-based algorithm with 73.6% sensitivity, 95%CI (59.7-84.7) and 80.3% specificity, 95%CI (69.1-88.8), if compared to the gold standard reference represented by the decision of the clinician. The validation phase is ongoing.
