Layered variable selection for multivariate Bayesian regression with density function-based features: a case study in imaging-genomics (Shariq Mohammed -- BU Biostats)

We will present a statistical framework that integrates radiological magnetic resonance imaging (MRI) and genomic data to identify the underlying radiogenomic associations in lower grade gliomas (LGG). We devise a novel imaging phenotype by dividing the tumor region into concentric spherical layers that mimics the tumor evolution process. MRI data within each layer is represented by voxel intensity-based probability density functions which capture the complete information about tumor heterogeneity. Under a Riemannian-geometric framework these densities are mapped to a vector of principal component scores which act as imaging phenotypes. Subsequently, we build Bayesian variable selection models for each layer with these imaging phenotypes as the response and the genomic markers as predictors. Our novel hierarchical prior formulation incorporates the interior-to-exterior structure of the layers, and the correlation between the genomic markers. With a focus on the cancer driver genes in LGG, I will present some biologically relevant findings. I will discuss a general theme of some ongoing work on developing statistical methods for complex-structured biomedical data with applications in cancer, dementia and Alzheimer's disease, and geo-spatial public health.