Paige and Microsoft Build World's Largest AI Model to Detect Cancer

“By realizing the potential of generative AI at unprecedented scale, the Paige collaboration with Microsoft is a milestone in the history of oncology. It opens a window into the microscopic world with extraordinary fidelity, allowing for not only much higher accuracy but completely novel capabilities.”

Thomas Fuchs, Dr. Sc., Founder and Chief Scientist, Paige

On September 7, 2023, Paige announced a collaboration with Microsoft to build the world’s largest image-based AI model to detect cancer. The new AI model will be orders-of-magnitude larger than any other image-based AI model that exists today. Paige is the global leader in generative AI for pathology, leveraging exclusive access to the largest and most diverse dataset in cancer diagnostics. The company builds AI powered computational pathology products that help doctors make more precise and informed treatment decisions. Paige will leverage over a billion images from half a million pathology slides to build a new Foundation Model to enable next-generation applications in oncology and pathology.

“Until ChatGPT was released, no one really understood how this was going to impact their lives. I would argue this is very similar for cancer patients going forward. This is sort of a groundbreaking, land-on-the-moon kind of moment for cancer care.”

Andy Moye, CEO, Paige

With the number of cancer cases worldwide projected to increase by 60% over the next two decades, and a decrease in the number of pathologists relative to demand, there is an urgent need for new approaches like this. Many hospitals would like to use digital pathology but it's very expensive to collect and store huge amounts of image data. This new AI model will solve this storage problem and make it possible for smaller hospitals and clinics to use digital pathology. Paige will leverage Microsoft’s advanced supercomputing infrastructure to train the AI model at scale and deploy it to hospitals around the world using Microsoft's Azure cloud computing platform.

This new technology also helps pharma companies to more effectively evaluate treatment options for patients and design new biomarkers for drug development so that every patient gets precise treatment options. Paige is working to incorporate predictive modeling to give pathologists and patients access to information about their biomarkers and genomic mutations. This is a big milestone for AI in healthcare.

Highlights from the Virchow pre-print on arXiv

A preprint entitled Virchow: A Million-Slide Digital Pathology Foundation Model is available on arXiv showing the impact of the new model compared with existing models. The new model outperforms anything that has ever been built.

• For many computational pathology tasks the amount of data is inadequate for development.
• To address this challenge Dr. Fuchs and his colleagues created Virchow, a 632 million parameter deep neural network foundation model for computational pathology.
• Using self-supervised learning, Virchow is trained on 1.5 million hematoxylin and eosin stained whole slide images from diverse tissue groups, which is orders of magnitude more data than previous works.
• When evaluated on downstream tasks, Virchow outperforms state-of-the-art systems both on internal datasets drawn from the same population as the pretraining data as well as external public datasets.
• Virchow achieves 93% balanced accuracy for pancancer tile classification, and AUCs of 0.983 for colon microsatellite instability status prediction and 0.967 for breast CDH1 status prediction.
• The gains in performance highlight the importance of pretraining on massive pathology image datasets, suggesting pretraining on even larger datasets could continue improving performance for many high-impact applications where limited amounts of training data are available, such as drug outcome prediction.

History & Milestones

Paige was founded in 2017 by Thomas Fuchs and colleagues at Memorial Sloan Kettering Cancer Center. At MSK, Fuchs invented novel machine-learning algorithms in computational pathology. Prior to MSK, Dr. Fuchs worked at NASA Jet Propulsion Laboratory. Fuchs notes that the images captured by telescopes orbiting the solar system are the same size as the slides used in modern pathology and the pattern recognition techniques used by the Mars Curiosity Rover to navigate different types of terrain (sand vs gravel) are also used to differentiate types of tumors (malignant vs benign).

• In July 2020 the FDA cleared FullFocus viewer for the primary diagnosis of digital pathology slides. Paige was the first company to receive both FDA clearance and FDA breakthrough designation (December 2019) for digital pathology products.
• In October 2021 the FDA approved Paige Prostate for clinical use. Paige Prostate was the first AI-based pathology product to receive FDA clearance for in vitro diagnostic use in detecting cancer in prostate biopsies.?Paige Prostate is a clinical-grade AI solution designed to identify foci that could indicate cancer, enabling fast, accurate in vitro?diagnosis.
• In June 2022 Paige?announced a collaboration with Janssen to deploy a new AI based biomarker test for advanced bladder cancer. With this AI based biomarker test, patients can get their results in less than one hour. This means that patients can enroll in clinical trials and start treatment in days instead of waiting weeks or months to receive results from traditional gene sequencing tests.

Products Overview

1) Paige Platform powers a lab’s complete digital pathology workflow and includes cloud storage.

2) FullFocus? whole-slide image viewer delivers advanced digital slide review tools.

• FullFocus is FDA and CE cleared for clinical use with Philips Ultrafast Scanner and for use with H&E-stained?slides only.

3) Paige Prostate includes Paige Prostate Detect, Paige Prostate Grade & Quantify, Paige Prostate Perineural Invasion, Paige Prostate Biomarker.

• Paige Prostate significantly?improve pathologist sensitivity and specificity in the diagnosis of prostate cancer with an average?70% reduction in detection errors?in a cohort of 16 first-time users.
• Paige Prostate Detect is FDA and CE cleared for clinical use with Philips Ultrafast Scanner and for use with H&E-stained?slides only.
• All other Paige Prostate applications and use cases are limited to Research Use Only and not for use in diagnostic procedures.

4) Paige Breast includes Paige Breast Detect, Paige Breast Neoplasm, Paige Breast Mitosis, Paige Breast Lymph Node, HER2Complete.

• Paige Breast Lymph Node detects metastatic lesions that are not easily detected by the human eye. The tool has shown a 25% improvement in pathologist sensitivity.
• Paige currently holds CE-IVD and UKCA marks for HER2Complete designed to identify patients with breast cancer whose tumors have expressions of HER2 protein on H&E-stained tissue samples.
• Paige Breast Suite applications and use cases are limited to Research Use Only and not for use in diagnostic procedures.

Subscribe, Comment, Join Group

I'm interested in your feedback - please leave your comments.

To subscribe to the AI in Healthcare Milestones newsletter please click here .

To join the AI in Healthcare Milestones Group please click here .

Copyright ? 2023 Margaretta Colangelo. All Rights Reserved.

This article was written by Margaretta Colangelo. Margaretta is a leading AI analyst who tracks significant milestones in AI in healthcare. She consults with AI healthcare companies and writes about some of the companies she consults with. Margaretta serves on the advisory board of the AI Precision Health Institute at the University of Hawai?i?Cancer Center @realmargaretta