Google has released Health AI Developer Foundations (HAI-DEF) to expand on its earlier research use only model and enable health artificial intelligence (AI) development from early research to commercial ventures.
The initial release of HAI-DEF includes three embedding models focused on supporting development of medical imaging AI applications: Path Foundation for digital pathology, CXR Foundation for chest x-rays, and Derm Foundation for skin images.
Google said that the aim of the project is to provide a starting point for developers to build "high performing AI models for their own use cases, with a very small amount of additional data and compute."
HAI-DEF models are developed from extensive, self-supervised training on large amounts of diverse, deidentified data for their respective modalities, according to Google in a November 25 blog post.
Google described Path Foundation as an embedding model, trained from a ViT-S architecture, specialized on hematoxylin and eosin (H&E) stained images, adding that Path Foundation accepts 224 x 224 pixel patches from H&E slides to produce embeddings that can be used for data-efficient classification for applications like grading or identifying tumors, classifying tissue or stain type, and determining image quality.
