Reports of the Deliverables of 22HLT05 MAIBAI Project

D1: Technical specification document describing the characteristics of the clinical image collection framework and providing guidelines for its utilisation D2: Good practice guide on the established centralised and common metadata indexes and middleware, made available as open-source software for translation of metadata from different clinical imaging databases D3: Technical report on the categorisation of clinical data into subsets based on identified population subgroups and image acquisition key factors, including training and validation case separation D4: Technical report on sensitivity analysis to assess the robustness of the physics-informed modelling results to variability in input parameters of irradiation process, and in anatomical properties and pathology features D5: Technical report on the results of the AI tool training and benchmarking with respect to prediction performance, robustness, fairness, and, potentially, uncertainty quantification, under three scenarios D6: Technical report on the developed and validated methods for the explainability and interpretability of the trained AI tools, and on the results of the XAI tool training and benchmarking with respect to explanation performance D7: Technical report describing the use and features of a public AI validation toolbox for diagnostic imaging and providing recommendations for the design of an assessment framework for explainable and traceable AI tools