Paige has announced a significant expansion of its PanCancer Detect AI application, extending its cancer detection capabilities from 17 to more than 40 tissue and organ types. This advancement is powered by Virchow V2, the company's foundation model trained on 3 million digitized slides and 1.8 billion parameters.
Siqi Liu, Director of AI Science at Paige, outlined the platform's innovative approach: "By leveraging the tissue-agnostic capabilities of Virchow V2, Paige PanCancer Detect transforms cancer detection into a broad, scalable AI solution that generalizes the concept of 'cancer' across tissues. This unique approach, originally demonstrated in our research published in Nature Medicine, has shown in our studies to be effective in detecting both rare cancer variants, low-prevalence cancers, and even precancerous lesions—challenges that traditional task-specific AI models often struggle to address."
The technology aims to enhance pathology lab efficiency by automatically screening and prioritizing cancer cases, enabling pathologists to focus on critical diagnoses. The system also serves as a quality control measure, screening all cases to identify potential diagnostic discrepancies and minimize errors.
The platform's effectiveness has been validated through independent testing. In a study led by Dr. Catarina Eloy, head of the Pathology Department at Ipatimup (Porto, Portugal), researchers evaluated 62 challenging cases across 16 tissue types. The study revealed the system's capability to accurately flag suspicious cases, with one instance where the AI's assessment proved correct in a discrepant result.
PanCancer Detect is currently available through Paige's FullFocus image viewer and is expanding its accessibility through established platform partners, including PathPresenter, Aiforia, Indica Labs, PathAI, and Gestalt. This integration aims to provide broader access to advanced cancer detection capabilities across the pathology ecosystem.
Click here for the original news story.