Register DW to T1

Dear community,

I’m new here and trying to figure out some MRtrix 3 commands and functions.
Following some tutorial or discussion I only found how to register T1 images to DW images but I would like to try the opposite registration (fit the DW to T1) because another software which I use, is doing it. I have 2x2x2 mm DW images and 1x1x1 mm T1 anatomical.

Thank you very much in advance for your help and support.

Tommaso

1 Like

Welcome Tommaso!

Since MRtrix3 still does not yet have an appropriate image similarity metric for inter-modal registration, you are most likely going to be using some other software for the actual registration step anyway. In that case, whether or not doing the registration in one direction vs. another is of consequence will depend on the particular implementation of that software. For a truly symmetric registration, there should be no consequence (apart from the fact that the transformation will be in the opposite direction). For non-symmetric registration, it’s not a domain of familiarity for me and so I don’t know if one direction is preferred over the other.

Also bear in mind that transformations can simply be inverted. So for instance, you can register T1w to DWI, invert the transformation, and then apply that transformation to the DWI in order to align it with the T1w image. For linear transformations, which is what one would typically expect for an inter-modal, intra-subject registration, this inversion is relatively trivial (see transformcalc).

Cheers
Rob

Dear Robert,

Thank you very much for your help. Actually for this registration I was using another software but then, to transfer this images to MRtrix3 it was a bit complicated that’s why I asked whether I can do everything in MRtrix. I will try the suggestions you proposed and transformcalc.

Thanks a lot

Tommaso

Dear Tommaso,

if you have FSL installed, you can use the epi_reg script, it works well. If you want to use the output transformation matrix in MRtrix, you need to convert it to MRtrix format using transformconvert.

Best regards
Samuel

Dear Samuel,

Thank you very much! I will try also that.

All the best

Tommaso