Visual navigation and classification of datasets in feature similarity space
Ian Bowman, Shantanu H. Joshi, John D. Van Horn · 2011
Discovering and understanding relationship patterns within a feature-rich archive (e.g. quantifying the degree of neuroanatomical similarity between the scanned subjects of a Magnetic Resonance Imaging (MRI) repository) is a nontrivial task. Scientists and expert users employ a variety of commodity algorithms for automated statistical analysis of feature patterns within a collection. But such analysis assumes the user is an experienced statistician, and disregards human visual processing capability. In this work we define a visual process for exploring the structure, relationships and patterns within a neuroimaging archive. Through dataset placement, our three-dimensional environment expresses similarity among the data. The application facilitates further analysis via two-stage exploratory clustering and classification.