The root causes of various neurological and psychiatric disorders are not identified yet. While there is a consensus that alterations in the central nervous system must be involved in these disorders, the exact nature and location of these alterations is often unknown. When those properties are discovered, a novel biomarker may be defined, enabling early and objective diagnoses of the disease, eventually improving the patient outcome. In this context, brain imaging is often used to characterize brain tissues non-invasively and in vivo. In particular, recent developments in diffusion-weighted imaging have led to the advent of models of the brain microstructure. These models provide an unprecedented insight into the complex organization of cellular structures in the brain. They are therefore of strong interest in population studies to characterize neurological and psychiatric disorders. Conducting population studies from models of the brain microstructure, however, raises numerous challenges that pertain to the required image acquisition, the estimation, registration and statistical analysis of those models. This thesis addresses these challenges and defines a comprehensive framework that harnesses multi-fascicle models of the brain microstructure in population studies. In particular, this framework includes methods to estimate multi-fascicle models from widely available clinical data, to reliably select an appropriate model of the brain microstructure, to register multi-fascicle models and to perform statistical analyses of microstructural properties. This framework is tested in population studies where it unravels alterations of the brain microstructure associated with autism spectrum disorders. The presented framework opens opportunities for new investigations of the brain microstructure in normal development and in disease and injury. It paves the way to the definition of microstructure-based biomarkers of neurological and psychiatric disorders.
Taquet, M. (2014). Multi-fascicle models of the brain microstructure for population studies : acquisition, estimation, registration and statistical analysis. https://hdl.handle.net/2078.5/51914