Registering MNI image to template using mrregister

1

Hello MRtrix community,

I am encountering challenges with registering and transforming MNI images to a template space using MRtrix. Here’s a breakdown of my issue:

  1. Problem Overview:

    • I used the mrregister command to register each ROI to the template mask (template_maskPS1.mif) using an affine transformation. However, the registration did not succeed as expected. Only 6 out of 8 ROIs had output files generated, and they were incorrect.

  2. Questions:

    • How should I structure my mrregister command to ensure proper output?
    • After a successful registration, how should I structure the mrtransform command to correctly transform the files to the template space?
    • Are there specific parameters or settings I should adjust in mrregister or mrtransform for better alignment results?

  3. Data and Checks:

    • I have verified the compatibility of my ROI files (CC.nii.gz, etc.) and the template mask (template_maskPS1.mif). They appear to be correctly formatted and aligned spatially.

  4. Overall goal:

    • I want to use various ROIs defined by an MNI atlas to extract FBA based metrics (the template space i am trying to register to).

Any insights or guidance on troubleshooting or improving this registration and transformation process would be greatly appreciated.

Thank you for your assistance.

Best regards,
Paige Spencer

2

Hi Paige,

Here are two alternative solutions that don’t require mrregister but might still be able to achieve the objective…

No group-averaged T1w image in population-template space
This does rely on having a DWI image and a registered T1w image in subject space…

  1. register the subject T1w image to MNI T1w image, and store the transformation. ANTs might be helpful here. Repeat this for every subject.
  2. apply the inverse transformation to the MNI ROI (to produce a ROI in subject-space). The ANTs transformation can be converted to an mrtrix compatible format (see here), which can then be fed into the mrtrix command mrtransform. Assuming that the subject-space T1w image and the DWI image are aligned, the output ROI of mrtransform should align with the subject T1w image and the DWI (since the T1w and DWI images are aligned).
  3. The transformation from subject DWI space to population template space is generated as part of the FBA pipeline. This warp can also be applied (using mrtransform) to the subject-space ROI. This can be repeated across all subjects, and an average ROI mask can be created using a similar command that was using to create the group-average mask (i.e., mrmath <input_all_warped_masks_multiple_inputs> min <output_template_mask_intersection>

Group average T1 image in population template space
If some sort of multimodal population template has been generated (for example, see here), then it is possible to directly register the MNI anatomical image to the group-averaged anatomical image (i.e., using ANTs). Then convert the transform and use mrtransform apply the transformation to the MNI ROI (to get the ROI in the study specific population template space)

Hope this helps!

Cheers,
Arkiev