LG AI Research 574

LG AI Research is leading bio-healthcare innovation by releasing EXAONE Path 2.0, a next-generation cancer diagnostic AI model, as an open-source model for research use.

EXAONE Path 2.0 is a model specialized in analyzing high-resolution pathology images (WSI) to identify cellular structures and abnormalities. In particular, through technology that integrates learning from the patch level to the whole slide level, it solves the feature collapse problem and achieves world-class gene mutation prediction accuracy.

Key features and achievements are as follows:

• Reduced genetic testing time: Through large-scale multimodal data training combining over 10,000 pathology images with RNA genetic information, testing time that previously took more than 2 weeks has been reduced to just a few seconds.
• Precision medicine support: Gene expression prediction is now possible using only tissue images without expensive RNA testing, helping to reduce patients' medical costs and enable customized treatment proposals.
• Clinical applicability: By providing near real-time speed and reliable results, it is expected to be utilized not only for cancer diagnosis but also for prognosis prediction and treatment response analysis.

Through this model release, LG AI Research plans to build an open research ecosystem and continue expanding AI-based precision medicine innovation in collaboration with global partners.