Prediction of Fetal Brain Anomalies Using Deep Learning Models

Fetal brain anomaly prediction is important for fetal medicine, as well as for prenatal health care. Fetal anomalies are classified into two types, anomalies in the fetus' body parts including heart, lung, and kidney anomalies in facial features. Thus, to predict the anomalies in the fetal brain, various deep-learning models are used in this paper. The LeNet, AlexNet, ResNet, VGGNet, GoogLeNet and ZFNet were used. From the results obtained, the ResNet method achieved 99% accuracy with the given 3D ultrasound images of the fetal brain compared with other models and benchmark algorithms like SVM, Logistic regression, and KNN. The results of this paper are useful to identify fetal anomalies. Hence, this approach will be useful for improving outcomes for individuals and society as much as possible.