Enhancing Fetal Anomaly Detection in US Images: A Review of Machine Learning-Based Approaches

Ramin Yousefpour Shahrivar¹, Fatemeh Karami², Ebrahim Karami³

Abstract

Fetal development is a critical phase, and timely identification of anomalies in ultrasound images is vital for the health of the unborn child and the mother. Medical imaging has been crucial in detecting fetal abnormalities and malformations during prenatal diagnosis. Despite advances in ultrasound technology, accurately detecting irregularities in prenatal images remains challenging, often requiring significant time and expertise from medical professionals. To address this, we comprehensively review recent developments in machine learning (ML) approaches applied to fetal ultrasound image analysis. By leveraging principles inspired by nature’s adaptability and efficiency, these ML techniques aimed to enhance the precision of prenatal anomaly detection. Various ML algorithms in their application to fetal ultrasound images, including image classification, object recognition, and segmentation, were reviewed. It was highlighted how these innovative approaches can improve ultrasound-based fetal anomaly detection and offer insights for future research and clinical implementations.

Keywords: fetal anomaly, prenatal diagnosis, machine learning, deep learning, ultrasound imaging