Abstract #0442

Laidlaw D, Correia S, Tate D, Zhang S

We evaluated the accuracy of automated DTI fiber clustering results for representing known white matter features in four normal subjects. Two expert raters provided confidence ratings for identifying 11 white matter fiber tracts bilaterally in the clustered models. Large coherent fiber tracts were identified with high confidence, whereas some smaller tracts were missing. Some of the missing tracts were identifiable in an unclustered model with prior anatomical knowledge. The results demonstrate that the clustering algorithm can produce anatomically-relevant results, but dropout of smaller fiber pathways indicate the need for improvements in DTI imaging and modeling.