The quantitative analysis of imaging via machine learning methods for high dimensional data (“radiomics”) is defining the new frontier for diagnostic imaging research. Radiomic analyses generate a vast array of imaging - based markers for diagnosis and prediction. Each marker carries claims of potential utility in clinical care and is presented as a potential candidate for inclusion in clinical trials. The volume of the new radiomics - based markers and their potential for fast evolution, even without formal learning, poses a new set of challenges for researchers and regulators. In this presentation we will survey statistical machine learning methods used in the development of new markers, discuss published examples, and examine the statistical and regulatory challenges they create. We will also discuss studies in “radiogenomics”, which examine the correlation between imaging and genomic information.