Summary:
Microsoft has announced a collaboration with digital pathology provider Paige to develop the world’s largest image-based artificial intelligence model for cancer identification. This AI model is being trained on an unprecedented amount of data, including billions of images, with the capability to detect both common and rare cancers that are challenging to diagnose. The goal is to support healthcare professionals dealing with staffing shortages and increasing caseloads.
Paige aims to improve pathology workflows by digitizing them, enhancing accuracy and efficiency. The collaboration with Microsoft, using their cloud infrastructure, has allowed Paige to build an extensive AI model.
One significant challenge in digital pathology is the cost associated with data storage, as a single slide can require over a gigabyte of storage. Smaller health systems often struggle with these expenses, while wealthier academic centers have been the primary adopters of digital pathology. Paige, having spun out of the Memorial Sloan Kettering Cancer Center, possesses a substantial amount of data, which enabled them to create AI-powered solutions. However, to expand and identify more cancer types, they partnered with Microsoft.
Over the past year and a half, Paige has leveraged Microsoft’s cloud storage and supercomputing infrastructure to build an advanced AI model. This model is training on 4 million slides to detect both common and rare cancers, making it the largest publicly announced computer vision model to date.
While the research is a significant step forward, it is still in the early stages of development.Ultimately, this AI model has the potential to alleviate storage challenges for health systems and expedite diagnoses, especially in overwhelmed community clinics, democratizing access to healthcare and potentially reducing waiting times for patients.
Article written by Ashley Capoot| Image by Unsplash
07/09/2023
Source:
CNBC
https://www.cnbc.com/amp/2023/09/07/microsoft-paige-building-worlds-largest-ai-model-to-detect-cancer.html