Infrared Nanoprobes and TissuE mimickinG tumoR shApe identificaTION

Objective

Cancer surgery often comes down to a crucial question: where does the tumor end, and healthy tissue begin? One wrong cut can leave malignant cells behind or damage vital functions. INTEGRATION addresses this challenge by combining three cutting-edge approaches into a single preclinical platform: near-infrared-emitting nanoprobes, 3D-bioprinted brain-like models, and machine learning algorithms. By operating in the underexplored third biological transparency window (NIR-III, 1550-1850 nm), where tissue scattering is reduced and autofluorescence disappears, INTEGRATION will deliver sharper, higher-contrast images of tumor boundaries than currently possible. To validate this strategy, custom fluorescent nanoprobes will be engineered for stability, biocompatibility, and strong NIR-III emission. These will be tested in lifelike tissue phantoms created with 3D bioprinting, where healthy and tumoral regions are reproduced with tunable shape and optical properties. Fluorescence images collected from these models will train a convolutional neural network (U-Net) to precisely segment tumor margins in three dimensions. Finally, the approach will be validated ex vivo in mouse brain tissues, generating a proof-of-concept pipeline that is rigorous, ethical, and sustainable. Expected outcomes include the development of reliable NIR-III contrast agents, protocols for brain-mimicking phantom fabrication, and a machine-learning-based method for accurate tumor segmentation. INTEGRATION thus paves the way for safer surgeries, fewer relapses, and faster translation toward real-time fluorescence image-guided operations. Beyond healthcare, it reduces dependence on animal models, strengthens Europe’s leadership in nanomaterials, bioprinting, and machine learning, and supports the long-term vision of precision medicine and robotic-assisted surgery.