Computational anatomy strategies for characterization of brain patterns associated with Alzheimer's disease

Alzheimer's disease (AD) is one of the most complex systematic malfunctions of the nervous system that are known. The clinical symptoms of this neurodegenerative disease are alterations in cognition and behaviour that can lead to the onset of a dementia syndrome. Disease mechanisms that lead to neurodegeneration and cognitive impairment in sporadic AD are not well understood yet, making it difficult to predict the clinical progression of patients at the early stages of the AD continuum. Currently, no single biomarker or exam is sufficient to diagnose AD and existing standard instruments are not sensitive enough to detect subtle changes, predict the clinical course, and recognize heterogeneous forms of AD. This thesis presents two computational anatomy strategies aiming to identify and quantify neurodegeneration patterns associated with different clinical stages along the AD continuum using two different modalities of magnetic resonance imaging. A third contribution consists of a data-driven strategy to develop a set of domain-specific scores that result useful to estimate the risk of and predict the progression from mild cognitive impairment to dementia. Evaluation of these strategies with machine-learning and statistical inference methods demonstrate the potential of the proposed quantitative tools to help patients' clinical management and monitoring and could be used to improve the evaluation of potential disease-modifying interventions.

Publication year:2022

Keywords:Doctoral thesis

Accessibility:Open