Raidium Publishes Foundation Model for 3D Tumor Segmentation in npj Precision Oncology
Raidium in France 🇫🇷, developing a foundation model for precision medicine in radiology, has just published a study about its foundation model for 3D tumor segmentation, offering radiologists a promptable tool based on CT scans (A promptable CT foundation model for solid tumor evaluation). The ONCOPILOT is an interactive CT-based foundation model designed for semi-automatic segmentation of solid tumors. It streamlines longitudinal tumor evaluation and unlocks the potential of advanced volumetric analyses in oncology. It’s a game changer for what they like to call 📣 “precision radiology”.
ONCOPILOT is significantly refining RECIST 1.1 evaluations with active radiologist engagement. Trained on more ⬆️ than 8000 CT scans, ONCOPILOT employs intuitive visual prompts, including point-click, bounding boxes, and edit-points. It attains segmentation accuracy that matches or exceeds state-of-the-art methods, provides radiologist-level precision for RECIST 1.1 measurements, reduces inter-observer variability, and enhances workflow efficiency.
By integrating clinical expertise with interactive AI capabilities, ONCOPILOT facilitates widespread access to advanced biomarkers, notably volumetric 🎛️ tumor analyses, thereby supporting improved clinical decision-making, patient stratification, and accelerating advancements in oncology research.
Since their pre-seed round in June 2023, Raidium’s founders Paul Herent, M.D., Msc, and Pierre Manceron have been focused on building out an exceptional team, training the first 3D Large Vision Language foundation model for CT scans & MRIs on a huge proprietary dataset from their partner CENTRE IMAGERIE DU NORD, and prototyping a pioneering user interface for radiologists. Five months ago (January 2025) they proudly announced Raidium's €16M seed round, co-led by Newfund and Kurma Partners, along Galion.exe, Founders Future, Debiopharm, Techmind, and the European Innovation Council and SMEs Executive Agency (EISMEA).