Abstract
The RINO project represents a groundbreaking integration of artificial intelligence and molecular imaging to address critical challenges in glioblastoma multiforme (GBM) diagnosis and monitoring. Our innovative approach employs trained agentic AI models to revolutionize the development pipeline for positron emission tomography (PET) imaging probes specifically designed for neuro-oncological applications beyond LAT1. The AI system operates through two primary functions: intelligent target selection for radiopharmaceutical development and optimization of engineered antibody fragment design and radiochemistry. Our computational models analyze vast datasets of molecular targets to identify the most promising antigenic sites for GBM-specific imaging, while simultaneously designing novel radiolabeled antibody fragments with enhanced properties for central nervous system penetration. These engineered fragments have been specifically modified to optimise their biology for nuclear imaging and maximize blood-brain barrier transport, addressing a fundamental limitation in neuroimaging probe development. The probes are designed to enable precision molecular antigenic imaging that could significantly improve diagnostic molecular profiling and treatment monitoring in GBM patients. This presentation will detail the architecture and training methodologies of our agentic AI models, demonstrating how machine learning algorithms can accelerate and optimize the traditionally lengthy process of radiopharmaceutical development. We will present our comprehensive pre-clinical translation pathway, outlining the systematic progression from AI-guided molecular design through pre-clinical translation. The RINO project establishes a new paradigm for AI-assisted molecular imaging probe development, with implications extending beyond GBM to other neurological disorders requiring precision diagnostic imaging approaches.