Effect of different configurations of diffusion gradient directions on accuracy of diffusion tensor estimation in cardiac DTI

Diffusion tensor imaging (DTI) has now been increasingly used to study myocardial fiber structures and diagnose abnormalities of the human heart. Optimal diffusion gradient directions used to acquire data at a given angular resolution are often different for different sampling schemes or DTI machines. Different choices of diffusion gradient directions result in different diffusion weighted (DW) images and subsequently affect the accuracy of diffusion tensor estimation. This paper aims to study the effect of choices of diffusion gradient directions on estimation accuracy of diffusion tensors in cardiac DTI. To this end, various configurations of diffusion gradient directions were designed by means of a large number of rigid rotations with respect to a fixed set of diffusion gradient directions distributed uniformly on a sphere. Then cardiac DW images associated with the rotated directions were generated. Diffusion tensors, tensor-derived fractional anisotropy (FA), mean diffusivity (MD), and fiber orientations were subsequently estimated from the generated DW images for comparison and analysis. The results showed that the accuracy of diffusion tensors, FA, MD, and fiber orientation all varied depending on the rotations of diffusion gradient directions, but their ranges of variation were relatively small. The range of mean estimation errors over all the configurations of diffusion gradient directions decreased as the angular resolution increased. Besides, the mean estimation errors were also reduced when increasing the number of signal averages. These findings provide potential suggestions that can be used in optimizing trade-off in designing DTI protocols.