Diffusion tensor distribution imaging (DTD) is a versatile technique enabling to retrieve nonparametric intra-voxel diffusion tensor distributions from tensor-valued diffusion-encoded data. While DTD owes its versatility to the minimal set of assumptions on which it relies, such minimal constraints induce a high sensitivity to noise hindering DTD's potential clinical translation. In this work, we demonstrate within a brain-like numerical phantom that generalized singular-value shrinkage (GSVS) denoising of the data prior to DTD analysis drastically improves DTD's accuracy, mitigating the aforementioned issue.
This abstract and the presentation materials are available to members only; a login is required.