A Semantic Contour Tree Approach for Visual Comparison of Brain White Matter Connectivities in Cohorts

We present a semantics-driven contour tree visualization approach to support exploring and comparing cohort-level brain white matter connectivities. A contour tree is a topological method that stores the nesting relationships of the contours in a scalar field, here a brain water molecule movement measure of factional anisotropy (FA) values. Previous contour tree visualizations have lacked the capability to effectively relate two-dimensional (2D) tree structures to the three-dimensional (3D) counterparts reflecting the semantic structures. Our approach is semantics-driven in that contour trees are labeled with brain anatomical regions, thus are stable in their structures for comparative studies. We further explore the contour tree approach as a tool for interactive exploration and for comparative studies of patient and normal cohorts. Our approach is novel not only in summarizing the 3D topological structures but also in showing spatial attributes associated with brain connectivity represented in fiber tracts, thus allowing brain scientists to examine and compare critical differences between cohorts.