Fractal Dimension Maps of brain MRI and decentralized deep learning for prediction in Friedreich ataxia and SCAs using the ENIGMA-Ataxia meta-dataset

Friedreich (FRDA) and spinocerebellar ataxias (SCAs) are slowly developing and severely disabling conditions. Finding stratification and treatment monitoring indices for pharmacological and gene therapy trials is vitally necessary. In the previous FLEX-AI project, we proved that the fractal dimension (FD), a quantitative index of structural brain complexity derived from magnetic resonance imaging (MRI) data, provides further insights into brain changes in these diseases. However, the FD does not give spatial localization of brain changes. Thus, in this project, we propose a novel FD brain map to identify the brain regions with altered brain complexity and evaluate it as a novel imaging index in inherited ataxias. Finally, using an artificial intelligence (AI) decentralized privacy-preserving approach, we will assess the added value of the FD map to predict a clinical decline in FRDA and SCAs. We will use the brain MRI, genetic and clinical data of 786 patients with inherited cerebellar disorders and 1013 healthy subjects collected through the ENIGMA-Ataxia international consortium in the FLEX-AI project. This project will provide a cutting-edge computational approach – the FD map – to understand these diseases’ pathophysiology better, identify strategies to improve clinical trial design, and make available to the inherited ataxia international community the decentralized privacy-preserving technologies for AI training.