Development of Artificial Intelligence-powered prediction models of disease progression in facioscapulohumeral muscular dystrophy: DEAL-FSHD

The possibility to predict disease progression in FSHD is very limited at the moment. The presence of hyperintense lesions in short tau-inversion recovery (STIR) muscle MRI sequences has been linked to a higher risk of increased fatty replacement at single muscle level after 1 year, although with large and unexplainedvariability. In the present project we will use radiomics, a technologically innovative approach applied in medical imaging for the discovery of attributes unidentifiable by naked eyes, to deeply characterize all the muscles and the STIR positive lesions in a longitudinal cohort of MRI scans from FSHD patients. This approach could allow the identification of new biomarkers usable for better prognostication and potentially exploitable as surrogate endpoints in clinical trials. We will also integrate these new pieces of information with clinical and genetic data,and apply artificial intelligence algorithms to predict in which muscles there will be an higher likelihood of disease progression, opening new opportunities to tailor medical counselling and rehabilitation interventions.