EMory BrEast Imaging Dataset (EMBED)

Description

EMBED is a racially diverse mammography dataset containing 3.4M screening and diagnostic images from 110,000 patients collected from 2013-2020, with an equal representation of black and white women. The dataset is comprised of 2D, synthetic 2D (C-view), and 3D (digital breast tomosynthesis, i.e. DBT) images. It contains 60,000 annotated lesions linked to structured imaging descriptors and ground truth pathologic outcomes grouped into six severity classes. This release represents 20% of the total 2D and C-view dataset and is available for research use. DBT, US, and MRI exams will be added at a later date.

Acknowledgements - We would like to thank Glendor, Inc and MD.ai for assistance with image de-identification.

Update Frequency

New data to be added annually

License

Custom License

Documentation

https://github.com/Emory-HITI/EMBED_Open_Data

Managed By

See all datasets managed by Health Innovation and Translational Informatics Lab at Emory University (hitilab.com).

Contact

hari.trivedi@emory.edu

How to Cite

EMory BrEast Imaging Dataset (EMBED) was accessed on DATE from https://registry.opendata.aws/emory-breast-imaging-dataset-embed. In addition, please cite the following publication - https://pubs.rsna.org/doi/epdf/10.1148/ryai.220047

Usage Examples

Tutorials

• Sample Notebook by Emory-HITI
  

Publications

• The EMory BrEast imaging Dataset (EMBED): A Racially Diverse, Granular Dataset of 3.4M Screening and Diagnostic Mammograms by Jiwoong J. Jeong, Brianna L. Vey, Ananth Reddy, Thomas Kim, Thiago Santos, Ramon Correa, Raman Dutt, Marina Mosunjac, Gabriela Oprea-Ilies, Geoffrey Smith, Minjae Woo, Christopher R. McAdams, Mary S. Newell, Imon Banerjee, Judy Gichoya, Hari Trivedi