Oblique acquisitions and scanner space v.s. image space for gradient directions

Hi all,

this may be an old topic, but I couldn’t find the answer in this forum:

I have diffusion data acquired using oblique acquisitions, and gradient directions specified in image space (i.g. FSL style). Am I correct in in saying that mrtrix would need the gradient directions in scanner space? Or is there a way to tell mrtrirx that the gradient vectors are in image space?

I know I can go back to the DICOMs and extract gradient directions using mrconvert, however this may not be an option (i.e. the DICOMS may not be accessible).



All MRtrix3 commands that need access to the DW gradient table will offer a few options to provide this table if it isn’t already in the image headers. One of these options is:

which is designed specifically for use with FSL-style gradient tables. Have a look at the documentation for e.g. dwi2fod, under the DW gradient table import options section for all the other options.


thanks for the quick reply!

I had seen the -fslgrad option, however I was worried that this would only cater for file format an not for coordinate rotations in the case of oblique acquisition.

So if I understand correctly, I can use a gradient table which is in image coordinate system (and not in scanner coordinate system - the two will differ with oblique acquisitions) simply using the -fslgrad option?

Thanks again!



Absolutely. We’ve taken great care to ensure the gradient table is interpreted correctly with reference to the DWI image provided. Just make sure the DWI is exactly as it was when the bvecs/bvals were originally produced (i.e. no regridding etc. has been applied) - otherwise this can introduce problems. But generally, it should just work.

1 Like