The research plan includes four main work packages:
1) the analysis of radiomics reliability;
2) the diagnosis-related machine-learning classifications performed retrospectively based on imaging studies previously performed;
3) the prognosis-related machine-learning classifications performed prospectively;
4) data sharing of experiments, classification models and datasets through an online platform.
The preliminary reliability analysis is a milestone for machine-learning classification tasks performed both retrospectively and prospectively. In turn, the development and validation of the classification models is essential to share these models with other physicians and scientists.
