Development of statistical methods for neuroimage data analysis towards early diagnostic of neurodegenerative diseases

Abstract

This doctoral thesis introduces a novel statistical framework to improve early detection of Alzheimer’s disease (AD) using neuroimaging data, particularly Positron Emission Tomography (PET). The proposed method is grounded in Functional Data Analysis (FDA), a statistical paradigm that treats data as continuous spatial surfaces rather than collections of independent voxel values. This shift allows for more powerful and spatially coherent inference, which is especially important in neuroimaging contexts and in early disease stages where subtle and diffuse changes are the norm.