Hi Gabriel - The eddy-current correction method that's appropriate depends on the b-value. For a moderate b-value, the DWIs have enough features to be registered to the b=0 image. If you have something like b=2000, this won't work so well.
The reorientation of tensors is just applied to the gradient table, it doesn't do anythign to the images themselves, so nothing special there that would change with the protocol.
The intra-subject registration is from the b=0 image to the T1, so as long as there is b=0 image, the rest of the protocol doesn't have any effect.
To check the gradient vectors, you have to fit some kind of model of orientation (like the tensor) to the image data and check if the estimated orientations are correct (following the big tracts, like going left-right in the corpus callosum, inferior-superior in the corticospinal tract, etc).
Hope this helps, a.y
On Thu, 18 Sep 2014, Gabriel Gonzalez Escamilla wrote:
Dear Freesurfer and DTI experts,
I will start some analyses with DWI data, but I have some doubts about it:
- I have data with 5 different acquisition protocols, so, I'm not sure if the standard pre-processing protocol (like eddy-current compensations the
reorientation of the vectors and the intra-subject registration) can be applied to all images equally or if I do need to set any specific parameters for each acquisition protocol?
- Does your method have any tool to check that the vectors are good, before or after fitting the tensors?
Many thanks in advanced, Gabriel