Towards Large-Scale Training of Pathology Foundation Models
This work addresses the need for large-scale, reproducible foundation models in medical imaging, particularly for pathology, by providing open-source tools and models to advance research and clinical applications.
The authors tackled the challenge of building foundation models for pathology images by developing a scalable training pipeline and releasing the first publicly available models trained on TCGA whole slide images, achieving state-of-the-art performance on patch-level tasks like breast cancer subtyping and colorectal nuclear segmentation.
Driven by the recent advances in deep learning methods and, in particular, by the development of modern self-supervised learning algorithms, increased interest and efforts have been devoted to build foundation models (FMs) for medical images. In this work, we present our scalable training pipeline for large pathology imaging data, and a comprehensive analysis of various hyperparameter choices and training techniques for building pathology FMs. We release and make publicly available the first batch of our pathology FMs (https://github.com/kaiko-ai/towards_large_pathology_fms) trained on open-access TCGA whole slide images, a commonly used collection of pathology images. The experimental evaluation shows that our models reach state-of-the-art performance on various patch-level downstream tasks, ranging from breast cancer subtyping to colorectal nuclear segmentation. Finally, to unify the evaluation approaches used in the field and to simplify future comparisons of different FMs, we present an open-source framework (https://github.com/kaiko-ai/eva) designed for the consistent evaluation of pathology FMs across various downstream tasks.