Genetics and brain imaging offer unique and complementary tools to examine the causes and clinical progress of neurologic and psychiatric diseases, such as Alzheimer's or schizophrenia. Today, neuroimaging measurements are being increasingly used as quantitative endophenotypes in discovering novel genetic variants associated with disease, and functionally characterizing and/or confirming established genetic variants. To date, the genetic analysis of brain morphology has largely been conducted via basic, univariate (scalar) neuroimaging endophenotypes, such as the volume, size, or thickness of anatomical regions of interest (ROIs). These measurements are often assumed to quantify the structural integrity of the corresponding ROI. For example, the volume of the hippocampus, an index of regional atrophy, is widely used as an AD- relevant endophenotype in elderly subjects. Yet hippocampal volume is merely a global quantification of the anatomical structure and can miss subtle, complex morphological variation, for example due to disease pathology localized in hippocampal subfields. Shape measures- multidimensional geometric descriptions of the ROI - provide an alternative, rich set of neuroimaging variables. Early and subtle variations in shape can be expected to presage more global atrophy patterns and will allow for a more sensitive and detailed study of potential genetic associations. Yet, largely due to the lack of appropriate computational tools, until now shape measures have not been employed as endophenotypes. The proposed project will develop a novel computational bioinformatics pipeline to conduct ROI-focused genetic association analyses with structural neuroimaging data, particularly in the context of disease, using multi-dimensional shape measures that capture the brain's complex morphology. Furthermore, we will dedicate significant resources to implement a dedicated website that will disseminate the software and results the project produces. We will develop methods to extract clinically-relevant (Aim 1) and heritable (Aim 2) phenotypes of neuroanatomical shape.
Aim 3 will then implement tools to further dissect and localize genetic effects on promising shape endophenotypes.
The proposed project will produce a set of freely available, user friendly and validated computational tools that will enable a novel analytic approach to use neuroimaging data in studying genetic effects on brain diseases and morphology. Instead of employing the volume/size of an anatomical structure, the most widespread quantitative measurement examined in association with genetic data, we propose to extract and analyze multidimensional geometric descriptors of the shape of the structure. We expect that these alternative measurements yield novel discoveries that link genomic variables to brain morphology.
