Graphical Data Mining Approaches in Human Neuroimaging
John Darrell Van Horn, Ph.D.
As archives of primary neuroimaging data grow in size, conventional approaches for statistical analysis may not always be appropriate for identifying trends and relationships present across the collection of data. Neuroinformatics approaches are needed but tend to suffer from an “image problem” or their own. Moreover, user interaction with the information contained therein requires new thinking to eliminate the burdens placed upon it by text-only queries. In this presentation, I will critically review the informatics approaches needed for “big data” neuroimaging and explore graphical ways by which users might dynamically interact with large-scale neuroimaging datasets and thereby lead to new hypothesis formation. Such methods present one vision for the future of “discovery science” in brain imaging fueling the next generation of neuroimaging research.