Improving image quality in small animal diffusion tensor imaging at 7T

Diffusion tensor imaging is being increasingly used as a means to elucidate the brain's fiber structure. High spatial resolution is needed to capture details of the anatomy for tractography. However, image deteriorating factors such as low contrast-to-noise ratio, partial volume effects and subject's displacements affect the analysis. We introduce a procedure that uses single subject acquisitions at multiple b-values followed by an image processing procedure that consists of: a tensor estimation in the log-Euclidean space, a tensor registration and rotation that corrects misalignment, and averaging through acquisitions aiming to diminish the effect of the image deteriorating factors mentioned above. By applying this pipeline in an animal model that reproduces our size-limiting factors, we obtain contrast-to-noise ratio improvements of 40.9%, 110.5% and 70.8% in isometric voxels of 0.2mm3, 0.3mm3 and 0.4mm3 respectively.