Vikas Singh’s group’s overarching research goal is to develop novel models and algorithms to solve real image (and data) analysis problems in scientific, biomedical and engineering or industrial applications. Their solutions and the systems that implement these ideas are driven by a broad spectrum of needs: evaluating neuroscience hypotheses, designing clinical trials using imaging data, statistical analysis of longitudinal images, data imputation, heterogeneous and across-sites harmonization problems, video analysis, as well as standard image understanding problems in vision involving segmentation, categorization, reconstruction and tracking. They typically focus on developing algorithms (based on extending or adapting ideas from statistics, geometry or optimization) whose mathematical properties can be analyzed; this often yields a clearer sense of the method’s behavior or failure modes beyond empirical performance on a few datasets alone. They also devote effort into providing open-source repositories and then working with collaborators to ensure that their methods address the requirements of the scientific application that initiated the work.