A Novel Clustering Algorithm for Application to Large Probabilistic Tractography Data Sets

Robert Elton Smith^1,2, Jacques-Donald Tournier^1,2, Fernando Calamante^1,2, Alan Connelly^1,2

Current clustering methodologies are not able to process very large data sets, such as those generated using probabilistic tractography. We propose a novel clustering algorithm designed specifically to handle a very large number of tracks, which is therefore ideally suited for processing whole-brain probabilistic tractography data. A hierarchical clustering stage identifies major white matter structures from the large number of smaller clusters generated. The method is demonstrated on a 1,000,000 track whole-brain in-vivo data set.