We are thrilled to share the public release of the Maternal-Fetal Ultrasound Video Dataset for End-to-End Intrapartum Biometry and Multi-task Learning (Published in Scientific Data, DOI: 10.1038/s41597-026-06900-5
)! This marks the first-ever multi-center, multi-device, and multi-category labeled intrapartum ultrasound video dataset, designed to drive progress in AI-based maternal and fetal health solutions.
📅 Key Highlights:
774 ultrasound videos and 68,106 images collected across three leading institutions.
Annotations for standard plane classification, segmentation of pubic symphysis and fetal head, and ultrasound-derived biometric parameters (Angle of Progression, Head-Symphysis Distance).
Designed to enhance multi-task learning and support the development of fully automated intrapartum ultrasound parameter measurement.
This dataset represents a crucial leap forward in improving the accuracy and efficiency of labor monitoring systems and empowering automated decision-making in obstetrics.
💡 Why it Matters:
AI-driven clinical decision support for real-time intrapartum assessments.
Streamlines automated workflows for monitoring fetal head descent and labor progression.
Vital for advancing multi-task learning in medical imaging.
🔗 Access the Dataset: https://lnkd.in/gRR7PhFm
📄 Published Article: Maternal-Fetal Ultrasound Video Dataset