Techniques for analysis of folding in the cerebral cortex

© 2017 Dr Rosita ShishengarGyrification describes the series of events through which the immature cortex develops from a smooth surface to a folded sheet. While there is considerable ongoing investigation into the biological mechanisms that underpin the formation of gyri and sulci, little is known about the events that directly lead to folding. Quantitative assessment of cortical folding is important in the definition of normal brain development and can provide insight into neurodevelopmental disorders. This thesis proposes morphometric analysis approaches for the study of cortical folding that address shortcomings of existing methods, with particular application to pre-clinical studies of fetal brain development. As the first contribution, a new measure that assesses the localised nonlinearities in cortical folding development is proposed. This measure is based on a linearised prediction of cortical shape using a spherical harmonic representation of cortical surfaces at adjoining developmental time points. Then we introduce a new gyrification index derived from the level set of the first Laplace Beltrami eigenfunction, referred to as LB-GI, and demonstrate how to perform sulcal extraction using the same Laplace Beltrami analysis. These measures combine the strengths of both the surface-based and curvature-based techniques existing in the literature. The LB-GI approach is extended to render it applicable to highly convoluted cortical surfaces, using level sets of multiple Laplace Beltrami eigenfunctions. The final contribution of the thesis is an atlas-free segmentation method that computes the inner and outer surfaces of the developing cortical plate in ex-vivo fetal brains by exploiting diffusion-weighted imaging cues, from which cortical thickness is derived. The techniques developed in this thesis enable automated analysis of fetal development in pre-clinical studies, consistent with analysis tools commonly applied in adult human studies. Furthermore, it is anticipated that the Laplace Beltrami-based cortical folding analysis tools will prove advantageous in large clinical neuroimaging studies