Artificial intelligence and machine learning applications in Nuclear Medicine for the identification of predictive and prognostic factors in oncological and endocrinological diseases

Recent technological advances, combined with the increasing volume of data available in electronic health records, have paved the way for the application of artificial intelligence (AI) and machine learning (ML) in the medical field. ML is a branch of AI that uses algorithms trained on available observations in order to predict the output of future input data. Nuclear medicine, which has grown to play a key role in the diagnostic work-up and treatment of various cancers, represents an ideal setting for leveraging the new capabilities of AI and ML. This is due to several factors such as the inherently quantitative and multimodal nature of molecular imaging and the potential of AI/ML algorithms to enhance image interpretation and support personalized treatment strategies. The primary aim of this project was to explore the feasibility of applying AI and ML techniques in Nuclear Medicine to identify predictive and prognostic factors in oncological and endocrinological diseases. To establish a versatile and effective strategy that could be applied across multiple oncological disease models, the project was divided into several sub-projects, each addressing critical challenges aligned with the primary aim: - Development of a robust data storage and analysis pipeline to support AI/ML applications, ensuring efficient and scalable data management; - Investigation of the in-vivo biological correlates of PET/CT tracer distribution within an oncological disease model, emphasizing the potential translation of molecular imaging into reproducible and quantitative data with predictive and prognostic significance; - Creation of a reproducible ML-based solution to extract standardized quantitative data from PET/CT imaging in an oncological disease model, enhancing data consistency, reliability and scalability; - Development and testing of ML models using clinical data to predict the response to radiometabolic therapy within a tumor model, aiming to optimize therapeutic outcomes; - Identification of predictive and prognostic imaging-based radiomic biomarkers for molecular radiotherapy response in a tumor model. This structured approach provided a comprehensive framework for addressing key challenges and advancing the use of AI/ML in Nuclear Medicine. The results of the project demonstrated the feasibility of leveraging AI and ML in Nuclear Medicine to identify predictive and prognostic markers in oncological and endocrinological disease models. Nonetheless, further research is needed to validate the robustness, reproducibility, and clinical utility of such approaches. Large-scale, multicentric studies, standardized protocols, and regulatory frameworks will be essential to ensure reliable integration into clinical workflows. Only through rigorous validation and continuous collaboration between clinicians, data scientists, and regulatory bodies can these technologies be safely and effectively translated into routine medical practice.