Fslgrad option with dwinormalise group

Hi, I’m trying to run dwinormalise group as part of FBA preprocessing. So far in the pipeline, I’ve been using the option fslgrad for my bvecs and bvals, but that’s not an option here.
I’m getting the following error:
dwi2tensor: [ERROR] no valid diffusion gradient table found

How can I get dwinormalise group to pipe my gradient tables to dwi2tensor? Or is there another workaround?


Best wishes,

1 Like

Hi Erin,

The reason why the -fslgrad option is not implemented for dwinormalise group as it is for other commands is because that command specifically operates on data from multiple subjects. To provide the gradient table information in this context would be to assume that the gradient table is identical across every single subject; which means that:

  1. The acquired gradient table and number of volumes need to be identical across all subjects;

  2. There would need to not have been any rotation of gradient vectors to account for subject-specific motion during pre-processing;

  3. The application of diffusion sensitisation gradients relative to the image voxel grid would need to be identical across participants (e.g. my scanner applies the diffusion sensitisation gradients relative to its own physical axes, so if two subjects have the imaging FoV aligned to their respective positions in the head coil, their corresponding bvec tables will differ).

I suppose that technically dwinormalise group could operate with a -fslgrad option, provided that the number of volumes is equivalent across all participants (and a healthy warning message is issued). However the technically preferable solution is to use the MRtrix image format, which can embed a subject-specific diffusion gradient table within the image header. So even if the gradient table is indeed identical for all subjects, for now you’ll nevertheless need to use an mrconvert call to embed the gradient table for each subject individually before running dwinormalise group.