Image-based multi-omics data integration : Exploring whole-body PET/MRI, -omics data and body composition

Abstract: Advanced body composition analysis with whole-body imaging could uncover novel associations between regional tissue composition and metabolic disease. Imiomics is an automated image analysis framework that enables large-scale integration of magnetic resonance imaging (MRI) data and orthogonal technologies such as metabolomics and genomics for the detailed study of body composition. The Imiomics method is based on spatial normalisation to attain voxel-to-voxel correspondence in large cohorts of volumetric MR images. The spatially normalised data is then further used to generate voxel-wise statistical inference volumes for analysis. In this thesis, Imiomics was integrated with metabolomics for the first time, providing a detailed map of the relationship between the metabolome and regional body composition in T2D. Furthermore, Imiomics was integrated with genomics for the first time, exposing detailed associations between single nucleotide polymorphisms (SNPs) and sex-stratified body composition. A rapid and intuitive visual framework was developed for the analysis of volumetric Imiomics maps, and further applied to study the relationship between body composition and clinical variables in T2D. Whole-body positron emission tomography (PET)/MR was used to study detailed insulin-stimulated glucose metabolism and its associations with tissue volume and tissue fat fraction. This thesis has contributed to the field of advanced body composition research, primarily through the integration of Imiomics with additional -omics platforms.