Correlating DTI fiber clusters with white matter anatomy

S. Zhang, S. Correia, D. F. Tate, D. H. Laidlaw Department of Computer Science, Brown University, Providence, RI, United States, Department of Psychiatry and Human Behavior, Brown Medical School, Providence, RI, United States, Department of Psychiatry and Human Behavior, Centers for Behavioral and Preventive Medicine, Brown Medical School, Providence, RI, United States, Department of Medicine, Center for AIDS Research, Brown Medical School, Providence, RI, United States Introduction – We evaluated automated tractography clustering results via expert characterization of white-matter features present in the clustering results, in the tractography results, and in the normal brain.. The results show that automated clustering algorithms can identify fiber clusters that correlate with known white matter anatomy and that preserve most features in tractography results. Improvements in data acquisition, tractography, and clustering algorithms have the potential to improve the results. DTI tractography methods trace sufficiently large coherent neural pathways based on diffusion characteristics in cerebral white matter [1]. These pathways can be grouped into spatially-related bundles that appear to represent underlying white matter anatomy. These DTI fiber bundles have the potential to improve our understanding of white matter anatomy by identifying white matter tracts of interest (TOI) for studying the functional, behavioral, and cognitive impact of white matter disease, injury or development. However, to realize these potentials, the clustering results must be consistent with the underlying white matter anatomy. We conducted an initial assessment of the anatomical accuracy of an automated DTI fiber clustering algorithm.