Classification Of Fetal Brain Plates In Ultrasound Images Using Convolutional Neural Networks

The rise in fetal health issues has made it crucial to find accurate ways to check the well-being of unborn babies. Ultrasound imaging is a key tool for diagnosing and monitoring fetal conditions, but it often struggles with precision. In this study, we developed a deep learning model using a Convolutional Neural Network (CNN) to identify three key areas of the fetal brain: the Thalamic, Ventricular, and Cerebellum regions. The model uses an attention mechanism and is based on the ResNet۵۰ framework, which helps it pick up on detailed patterns in ultrasound images. We tested the model on ۱,۷۱۰ images, and it achieved strong results: ۹۱.۸۳% accuracy, ۹۱.۶۶% precision, ۹۱.۴۵% recall, and an F۱ score of ۹۱.۷۰%. These findings show that our model works better than older methods at identifying fetal brain areas and could be a valuable tool for improving prenatal care and detecting problems early.

کلیدواژه ها:

deep learning ، fetal ultrasound ، neural networks ، attention mechanism ، brain area classification