The overarching goal of this work has been that of devising novel methods for building functional neuromarkers from resting-state fMRI data to describe healthy and pathological human behaviour. Observing spontaneous uctuations of the BOLD signal, resting-state fMRI allows to have an insight into the functional organisation of the brain and to detect functional networks that are consistent across subjects. Studying how patterns of functional connectivity vary both in healthy subjects and in subjects a ected by a neurodegenerative disease is a way to shed light on the physiological and pathological mechanisms governing our nervous system. The rst part of this thesis is devoted to the description of fully data-driven feature extraction techniques based on clustering aimed at supporting the diagnosis of neurodegenerative diseases (e.g., amyotrophic lateral sclerosis and Parkinson's disease). The high-dimensional nature of resting state fMRI data implies the need of suitable feature selection techniques. Traditional univariate techniques are fast and straightforward to interpret, but are unable to unveil relationships among multiple features. For this reason, this work presents a methodology based on consensus clustering, a particular approach to the clustering problem that consists in combining di erent partitions of the same data set to produce more stable solutions. One of the objectives of fMRI data analysis is to determine regions that show an abnormal activity with respect to a healthy brain and this is often attained with comparative statistical models applied to single voxels or brain parcels within one or several functional networks. Here, stochastic rank aggregation is applied to identify brain regions that exhibit a coherent behaviour in groups of subjects a ected by the same disorder. The proposed methodology was validated on real data and the results are consistent with previous literature, thus indicating that this approach might be suitable to support early diagnosis of neurodegenerative diseases... [edited by Author]

2016 - 2017

XXX ciclo