Visualization framework for voxel-wise adjacency matrices for graph-theorical based analysis methods on fMRI data

Motivation: Graph-theoretical methods to analyze fMRI data can be powerful and flexible. Common to such methods is the construction of a graph adjacency matrix, which cannot be intuitively understood when visualized in itself. Goal(s): We aim to develop a framework to visualize adjacency matrices intuitively while retaining intuitive spatial localizability in relation to the brain. Approach: By stratify voxel-wise functional connectivity adjacency matrices through agglomerative clustering to form edge bundles, we 3D-render them with their end-point locations in a brain contour to ease localization. Results: 3D rendered, color-coded, edge bundles and their end-points can be distinctly identified in relationship to the brain. Impact: Our visualization framework allows both scientist and clinicians to employ graph theoretical analysis methods on fMRI data in a intuitive manner while retaining spatial localizability in the brain.