An Automated Pipeline for Phenotyping Embryonic Lethal Mice using 3D Imaging

With the completion of the Human Genome Project, the chemical sequence and location of every gene is now known. However, little is known about the function of each of the protein-coding genes in the human genome. The International Mouse Phenotyping Consortium (IMPC) is phenotyping single gene knockout mouse lines in an attempt to elucidate gene function and gain insight into genetic diseases. The goal of the IMPC is to provide phenotype information for knockout mouse lines for each of the 20,000+ genes in the mouse genome by 2021. A primary screen needed to be developed for phenotyping mouse embryos because approximately thirty percent of these knockout lines are predicted to be embryonic or perinatal lethal. This thesis describes the implementation of three-dimensional imaging as a method to identify morphological phenotypes in the developing mouse embryo. The IMPC has incorporated imaging as the primary screen for phenotyping embryonic lethals, but the classical method of qualitative annotation of individual two-dimensional sections is low-throughput and prone to observer bias. This thesis presents computer automated image analyses to identify statistically significant morphological phenotypes from knockout mouse embryo images. The generation of a 3D segmented atlas of wild-type anatomy provides baseline volume measurements of 48 mouse embryo organs and a template for automatic volume calculation of these structures for a given embryo image. Image registration methods combined with intensity, deformation, and atlas-based analyses are described, which automatically flag statistical abnormalities in anatomical volume and shape. A novel staging system is described that uses a computed four-dimensional developmental atlas to evaluate the stage of development of any mouse embryo at the global, structural, or voxel level. The described methods will be incorporated into the IMPC embryonic phenotyping pipeline to provide standardized image analyses that report mutant phenotypes in an automated and unbiased fashion.Ph.D