Imagene Partners with Oracle to Launch CanvOI on Oracle Cloud Infrastructure

Oracle and Imagene, a pioneer in AI-based precision oncology, today announced CanvOI, a state-of-the-art pan-cancer foundation model designed to facilitate innovative research and development in the digital pathology and oncology space. Running on Oracle Cloud Infrastructure (OCI), CanvOI captures the complex features and patterns within biopsy images, providing valuable oncology intelligence that could improve the way cancer is diagnosed and treated around the world.

CanvOI delivers a robust vision data backbone for the development of downstream applications in oncology research, including predicting treatment responses, discovering biomarkers, assessing disease prognosis, and identifying pathological features with state-of-the-art results, outperforming leading foundation models built for digital pathology.

For AI inference computation and to process data at scale, Imagene partnered with Oracle to run CanvOI on OCI, taking advantage of OCI Supercluster and OCI AI Infrastructure technology.

"To transform cancer research, Imagene did what countless other leading AI startups have done-it turned to Oracle's Gen2 AI infrastructure, which delivers the scale and performance required to power the world's largest AI models and help solve humanity's most difficult challenges," said Larry Ellison, Oracle Chairman and CTO. "With CanvOI running on Oracle Cloud Infrastructure, Oracle and Imagene are providing scientists with new weapons in the fight against mankind's deadliest adversary, cancer."

"The next decade holds immense potential for breakthroughs in cancer research driven by Artificial Intelligence," said Dean Bitan, Co-founder and CEO, Imagene. "We designed CanvOI to inspire researchers to dare to explore beyond. This is a pivotal milestone in the fight against this complex and devastating disease. By opening doors for innovative research, we aim to fundamentally change the future of cancer care."

CanvOI, a 1.1 billion parameters foundation model, was pre-trained on a diverse dataset including more than 630,000 tissue images from over 100 sites, covering more than 40 major body organs and tissue types. By applying a new approach to digital pathology foundation models and pushing computational demands differently, CanvOI has achieved industry-leading performance across various cancer-related tasks. Its ability to perform effectively, even with minimal labeled data, is crucial in the biomedical space, where data availability can be challenging.

CanvOI is the cornerstone of Imagene's new OISuite, a platform that supports researchers and diagnostics developers to explore a wide range of questions to conduct in-depth research independently, alleviating the need for AI expertise and extensive data acquisition. OISuite enables rapid discoveries while delivering the highest standards of data privacy and security.

"CanvOI transforms our approach to healthcare and disease understanding," said Dr. David Agus, Founding Director and Co-CEO, Ellison Institute of Technology and Imagene board member. "By providing access to comprehensive oncology intelligence in the cloud, it enables researchers to develop answers quicker and more efficiently. EIT is committed to leveraging CanvOI to propel groundbreaking innovations in our research that improve patient lives. This technology marks a new era of cancer research that pushes us closer to a future where cancer is a manageable condition."

Additional Resources

• CanvOI scientific paper, including benchmarks