Visualizing the Differences between Diffusion Tensor Volume Images

We present a technique for visualizing the differences between two Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) volumes. Our technique registers two DT-MRIs and then produces a 3D model that allows a user to simultaneously view structure in both volumes. The geometric model is based on one developed to visually represent a single volume with a carefully chosen set of streamtubes [6,7]. In the new model, we choose corresponding tubes from both volumes and show how they differ. Collectively, this illustrates the differences between the volumes. Saturation is used to indicate the magnitude of the difference between corresponding points on streamtubes. Streamtubes follow the orientation of the principal eigenvector of the diffusion tensor at each point in the volume. The cross-section at each point of the streamtube is an ellipse determined by the second and third eigenvectors and eigenvalues of the diffusion tensor. We apply the method to three types of cases. In the first, we compare the streamtubes generated with different integration methods to show how the choice of method can lead to different streamtubes. In the second case, we show difference models for two volumes that are identical except for noise. In the third case, we show differences between two volumes acquired from the same human subject but at different orientations. Differences are shown in all the cases and they are easily identifiable from the color coding.