(718) Artificial Intelligence for Fetal Sex Determination

Determination of fetal sex is one of the cardinal requirements of an anatomic ultrasound. It provides important clinical information that could profoundly affect both fetal and neonatal care and, in addition, is one of the most requested questions providers are asked during an ultrasound visit. Recognizing the different genders requires an element of skill and training which is not always available in more rural areas, especially when noninvasive prenatal testing (NIPT) is not an option. A viable artificial intelligence (AI) model for fetal sex determination could be of significant value to providers in these settings. Our novel study assessed how confidently a prediction model can determine the sex of a fetus from an ultrasound image.

Study Design:

Analysis was performed using 25,000 ultrasound image slices from a high-volume fetal sex determination practice. This dataset was then split into a training set (17,500) and holdout test set (7,500). A computer vision model was trained using a transfer learning approach with EfficientNetB4 architecture as base. The performance of the computer vision model was evaluated on the hold out test set. Accuracy, Cohen’s Kappa and Multiclass Receiver Operating Characteristic AUC were used to evaluate the performance of the model.

Results:

The AI model achieved an Accuracy of 88.27% on the holdout test set and a Quadratic Cohen’s Kappa score 0.843. The Multiclass ROC AUC score for Male was calculated to be 0.896, for Female a score of 0.897, Unable to Assess a score of 0.916 and for Text Added score of 0.981 was achieved.

Conclusion:

This novel AI/ML model proved to have a high rate of fetal sex determination that could be of significant use in areas where ultrasound expertise is not readily available.