Article Information
History
- March 12, 2023.
Copyright
The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.
Author Information
- Chinmayee Athalye1,2,
- Amber van Nisselrooij3,
- Sara Rizvi1,4,
- Monique Haak3,
- Anita J. Moon-Grady5 and
- Rima Arnaout1,5,6,*
- 1Division of Cardiology, Department of Medicine, University of California, San Francisco, San Francisco, California, United States of America
- 2Department of Bioengineering, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America
- 3Dept. of Obstetrics, division of fetal medicine, Leiden University Medical Center, Leiden, the Netherlands
- 4AstraZeneca, South San Francisco, California, United States of America
- 5Department of Pediatrics, Division of Cardiology, University of California San Francisco, San Francisco, California, United States of America
- 6Bakar Computational Health Sciences Institute; Department of Radiology; UCSF Berkeley Joint Program in Computational Precision Health; Center for Intelligent Imaging; Biological and Medical Informatics, University of California, San Francisco, San Francisco, California, United States of America
- ↵*corresponding author: rima.arnaout{at}ucsf.edu
Contribution: What are the novel findings of this work?
A deep learning model designed to screen for normal hearts in fetal surveys out-performed human experts in a cohort in which over 50 percent of CHD cases were initially missed clinically. Notably, the model performed well on community-acquired images in a low-risk population, including lesions it had not been previously trained on.
What are the clinical implications of this work?
These findings support the proposition that deep learning models can improve prenatal detection of CHD.
Posted March 12, 2023.
Deep learning model for prenatal congenital heart disease (CHD) screening generalizes to the community setting and outperforms clinical detection
