Search Tools Links Login

Microsoft and Paige Join Forces for Groundbreaking Cancer AI

Posted: 2023-09-11
By: dwirch
Viewed: 81

Filed Under:

News, AI

No attachments for this post


Microsoft and digital pathology firm Paige have joined forces to create the largest image-centric AI model aimed at cancer identification. Training on billions of images, this AI intends to diagnose both common and elusive cancers. This venture is expected to aid doctors overwhelmed with cases due to staffing shortages.

Pathologists, the doctors responsible for diagnosing diseases via tissue and fluid testing, have a crucial yet often overlooked role. The conventional method, inspecting tissue on a glass slide, remains unchanged for over a century. But mistakes can be catastrophic. To modernize this process, Paige has developed digital tools to enhance accuracy and efficiency.

Having already received FDA approval for its FullFocus viewing tool, which digitizes the pathology slide viewing process, Paige's AI can now help detect certain cancers. But to expand the AI's capabilities, Microsoft's cloud and computing prowess was sought.

With Microsoft's assistance, Paige's new AI model is now training on a whopping 4 million slides. Described as the largest computer vision model ever publicly revealed, it will revolutionize cancer care, likened to ChatGPT's impact on daily lives.

The duo is also considering predictive modeling for a more comprehensive diagnosis in the future. While AI's potential is tremendous, it aims to assist rather than replace pathologists. As Desney Tan of Microsoft emphasizes, "AI is a tool that is to be wielded by a human."

The partnership will soon release a paper detailing the model's advancements, outdoing current academic creations. However, making this model a market-ready product requires extensive testing and regulatory collaboration. Paige aims to address storage challenges for health institutions and expedite diagnostic timelines, ensuring patients in all care settings receive timely and accurate care.


Comments on this post

No comments have been added for this post.

You must be logged in to make a comment.