Shape Diffeomorphometry of Brain Structures in Neurodegeneration and Neurodevelopment

Large-scale structural neuroimaging datasets of neurodegeneration and neurodevelopment are emerging. It is now possible to analyze localized changes in shapes of brain structures at the 1 mm(3) scale of the magnetic resonance imaging (MRI) scans from these studies. These analyses provide neuroscientists and clinicians insight into how a network of brain structures can be affected by disease. Specifically, atrophied subregions of brain structures can be examined across populations. This is achieved by shape diffeomorphometry methods that map brain structures to population templates and use statistical models to estimate rates and change-points (onsets) of atrophy of these structures cross-sectionally and longitudinally. This chapter describes a pipeline for shape diffeomorphometry of brain structures that is publicly available via www.mricloud.org with applications to Alzheimer's disease and Huntington's disease as examples of neurodegeneration and schizophrenia and attention deficit/hyperactivity disorder as examples of neurodevelopmental diseases. The main steps are as follows: multi-atlas parcellation of brain structures from population samples, constructing templates, longitudinal geodesic mapping, and shape analysis for analyzing atrophy rates and change-points via linear mixed-effects models. Discussion of potential pitfalls and limitations concludes the chapter.