Huawei Cloud and Ruijin Hospital Release RuiPath 2.0 Pathology Foundation Model
Huawei Cloud and Ruijin Hospital have launched RuiPath 2.0, a 7-billion-parameter pathology foundation model covering 19 common cancer types and 205 diagnostic tasks. Alongside it, they introduced RuiPath 2.0 Edge, a lightweight 2-million-parameter model designed for local deployment on standard hospital PCs. This release addresses the high computational barriers of medical AI by offering a highly compressed edge version, enabling resource-constrained hospitals to perform local AI-assisted diagnostics. Furthermore, its focus on explainable AI and synthetic data generation for rare diseases helps bridge critical gaps in clinical trust and data scarcity. RuiPath 2.0 achieves state-of-the-art (SOTA) performance in 42 out of 59 downstream diagnostic tasks and uses multimodal generative AI to synthesize training data, achieving a 96.48% accuracy rate in lymphoma diagnosis. The edge version, RuiPath 2.0 Edge, works in tandem with the cloud to deliver minute-level inference.
## BACKGROUND
Pathology foundation models are deep learning frameworks trained on large-scale histopathology images to assist in automated cancer diagnosis and prognosis. In digital pathology, acquiring sufficient training data for rare diseases is challenging, leading researchers to use generative AI to synthesize realistic medical images and augment datasets.