Federated learning (FL) is an approach to machine learning (ML) where model training data is not stored in a central location and computation is distributed, enabling secure and privacy-preserving multi-party collaboration. FL can be used in healthcare to rapidly and securely develop new ML models from heterogeneous data sets and diverse sources, all while avoiding privacy and data ownership concerns. FL has been adopted for medical imagery research in radiology, radiation oncology, and medical oncology.