Tractography based white matter tract segmentation is robust to anisotropic resolutions

Motivation: Several studies have established tractography's sensitivity to anisotropic spatial sampling; however, the direct comparison between the performance of different algorithms has only been shown quantitatively on a simulated dataset. Goal(s): We aim to characterize the effect of anisotropic spatial sampling on bundle-wise tractography across algorithms compared to a high-resolution ground truth. Approach: We down-sample high-resolution ground truth data to anisotropic sampling. We up-sample the anisotropic data to isotropic for further comparison. Results: We show that tractography is robust to anisotropy with minimal biases if the beginning and end of pathways are well-defined. Impact: We show that some algorithms, when paired with well-defined beginning and endpoints for major pathways, show stability in highly anisotropic data.