DT-MRI interpolation: At what level?

This paper studies two levels of interpolation techniques for improving spatial resolution of diffusion tensor magnetic resonance imaging (DTI): a) at the level of scalar gray-level images using anisotropic interpolation and b) at the level of diffusion tensor fields using Log-Euclidean interpolation and quaternion interpolation. The performance of these different level interpolations is evaluated both qualitatively and quantitatively using criteria such as fractional anisotropy (FA), mean diffusivity (MD), fiber length, etc. We conclude that tensor field level interpolations avoids undesirable swelling effect in DTI, which is not the case with scalar gray-level interpolation, and that scalar gray-level image interpolation and Log-Euclidean tensor field interpolation suffer from decreases in FA and MD, which is however avoided by the quaternion tensor field interpolation.