D6.7 COMPARATIVE ASSESSMENT OF THE PROPOSED HARMONISATION APPROACHES

This deliverable (D6.7) aims to compare different approaches for data harmonisation to improve the performance of computational modules and provide harmonised data for clinical practice. These algorithms will be applied as a supervised-learning technique to generate a converted image where image quality has been learned from a pre-defined hand-crafted ground truth based on best state-of-the-art existing filtering, registration and normalization algorithms. The hand-crafted images can be acquired by adding some artefacts (e.g., some noise or linear/nonlinear transformation) to the original raw images.