A SAM-Residual U-Net Pipeline for Segmentation and Classification of Down Syndrome in Fetal Ultrasound Images

This paper presents an automated deep learning approach using Segment Anything and Residual U-Net to detect Down syndrome markers (nuchal translucency thickness and nasal bone presence) in fetal ultrasound images. The method achieved 98% accuracy and represents a non-invasive alternative to risky invasive prenatal diagnostic techniques.