Microsoft is working with Paige a company dedicated to creating digital and AI solutions for pathologists, in developing the world’s largest AI model intended for identify and recognize different types of cancer.
The model, capable of identifying common and rare cancers, was trained on a large data set, including billions of images. It can also detect other tumors whose diagnosis is often difficult without the help of technology. The researchers who worked on its initiation hope that it will help doctors who have difficulties due to the lack of staff and the accumulation of cases in the diagnosis of these diseases.
The role of pathologists is important in medicine, because they are responsible for performing tests on tissues and body fluids in laboratories to reach a diagnosis. They are in charge, among other things, of determining what a cancer patient has, and in many cases their work is essential to create a treatment plan for patients.
Therefore, having a model that can facilitate their work is very helpful not only for them, but for the group of professionals who are responsible for the diagnosis and treatment of cancer, as well as, of course, for those who suffer from the disease . pain.
Due to the increase in cases, their work has increased, but their work flow has not changed much in recent decades. Usually, they will examine a piece of tissue under a microscope and give their diagnosis. This is a tried and true method, but as in many cases, pathologists can miss something, with dire consequences for patients. That’s why Paige has long digitized the pathologists’ workflow to improve the accuracy and efficiency of these professionals.
That’s why Paige teamed up with Microsoft to develop the world’s largest image-based Artificial Intelligence model. They are intended for pathology and oncology. Paige created the first large-scale foundation model with nearly a billion images from half a million pathology slides of various cancer types.
Now, together with Microsoft, it is developing a new, much bigger than the other existing image-based AI model, which is configured with billions of parameters. This model will help to capture small complications of cancer, and will be the basis of clinical applications and computational biomarkers of the next generation. With them, pathology and oncology will be more developed.
In the next phase of development, Paige will incorporate up to four million digitized microscope slides of different types of cancer into the model. All from the company’s extensive clinical data archive. To do this, Paige will use Microsoft’s supercomputing infrastructure. This way you can train the technology on a large scale, and deploy it through Azure in hospitals and laboratories around the world.
Thomas Fuchs, Founder and Chief Scientific Officer of Paige remember that “Paige’s technology goes beyond what humans can do today and helps doctors provide better cancer care with the support of AI. By harnessing the potential of generative AI on an unprecedented scale, Paige’s collaboration model with Microsoft is a milestone in the history of oncology. It opens a window into the microscopic world with incredible fidelity, enabling not only greater precision but entirely new capabilities.“.
As in Desney Tan, Vice President and General Manager of Microsoft Health Futures emphasizes that «By combining Microsoft’s global research and cloud infrastructure with Paige’s deep experience and big data, we are creating new AI models that will provide unprecedented insight into cancer pathology. Harnessing the power of AI is a game-changer for advanced healthcare to improve lives“.