Microsoft has partnered with digital pathology company Paige to develop the largest image-based artificial intelligence (AI) model for cancer detection. This AI model will be used in digital pathology and oncology, and will be significantly larger than any existing model. According to Dr. Thomas Fuchs, Paige’s founder and chief scientist, the amount of data utilized in this model is “orders of magnitude” larger than what has been publicly released by Google or Facebook. This scale is crucial in accurately assisting pathologists and oncologists, as millions of digitized histology slides are required for reliable results.
Paige has been digitizing millions of slides to identify cancers for the past seven years, which demands vast amounts of data and ever-growing AI models. To handle this scale, Microsoft’s involvement becomes essential. The collaboration between Paige and Microsoft includes partnerships with Nuance, Azure, and Microsoft Research, as well as access to Microsoft’s supercomputer infrastructure. This partnership will enable Paige to train the AI model using Azure and deploy it across laboratories and hospitals.
The upcoming development phase will incorporate up to 4 million digitized microscopy slides from a clinical data archive. Fuchs describes the AI model as “ChatGPT for the microscopic world” and emphasizes that the data used to build this model is ten times more than the entire Netflix library. The benefits of this model include a better understanding of cancer morphology, which can serve as a foundation for rare cancer models, mutation prediction, and response prediction. It is expected that the model will significantly reduce misdiagnosis in medicine, which currently contributes to numerous deaths and disabilities in the United States each year.
In the future, Fuchs aims to expand the model to cover all rare cancers and build detection systems for various types of cancer. Paige has gained recognition in the healthcare industry for developing the first Large Foundation Model, utilizing over a billion images from pathology slides containing multiple types of cancers. They have received FDA approval for this clinical AI application in digital pathology. Additionally, Paige’s AI technology assists pharmaceutical companies in evaluating effective treatment plans for patients.
With Microsoft’s collaboration and expertise, Paige believes they can advance the state-of-the-art in cancer imaging and ultimately improve the lives of millions affected by cancer.