FOCUS: Four-chamber Ultrasound Image Dataset for Fetal Cardiac Biometric Measurement

During perinatal care, the four-chamber view ultrasound is a crucial imaging plane used in second-trimester screenings and fetal echocardiography, providing direct visualization of the fetal heart’s chambers. The biometrics in this plane, such as the cardiothoracic diameter ratio, is often measured for diagnosing congenital heart disease. However, manual measurement remains a time-consuming and labor-intensive task, and publicly accessible datasets are scarce, which limits the development and performance evaluation of machine learning algorithms. In this work, we therefore introduce FOCUS, a new dataset of four-chamber view ultrasound images to assist investigators in artificial intelligence. This dataset includes 300 images that have been manually annotated with cardiac and thoracic regions for fetal biometric estimation with the help of an expert sonographer. We hope this dataset can contribute to various research areas, particularly towards fetal diagnosing congenital heart disease. Furthermore, we report a series of deep learning experiments to demonstrate the potential utility of this dataset.  As part of open science, the FOCUS dataset is made publicly available under the CC-BY 4.0 license.