Issue with Negative Intensity Values in Left ROI After Transformation Using ANTs and MRtrix

1

Hi everyone,

I’m working on warping ROIs from MNI space to a subject’s native FA space using ANTs and MRtrix. After applying transformations, I noticed that while the right ROI looks normal (intensity values ranging from 0 to 1), the left ROI has intensity values ranging from 0 to -1, which is unexpected. I cannot view the left ROI despite the right looking great and co-registered.

Here are the steps I followed:

  1. Registered subject’s FA image to MNI152 template using antsRegistrationSyN.sh.
  2. Applied inverse transformations to warp the ROIs from MNI to FA space.

Commands used:

  • For registration: antsRegistrationSyN.sh -d 3 -f [MNI] -m [FA] -o output
  • For warping: antsApplyTransforms -d 3 -i [ROI] -r [FA] -o [output_ROI] -n NearestNeighbor -t [Affine,1] -t [InverseWarp]

I’ve double-checked the original ROI files, and both appear normal in MNI space. The issue only arises after transformation. I’ve also tried re-running the transformation steps, ensuring that all parameters are identical for both ROIs.

Has anyone encountered a similar issue, or can anyone suggest what might be causing this discrepancy in intensity values for the left ROI?

I’m using the most recent ANTs version and MRtrix 3.0.4 on a macOS M1 system.

Thanks in advance for any insights or suggestions!

2

Hi @Shai_P,

Try reversing the order of the two transforms to put the InverseWarp first, and [Affine, 1] second. Also add the verbose flag, to have a little more feedback, then remove once working properly:
antsApplyTransforms -d 3 -i [ROI] -r [FA] -o [output_ROI] -n NearestNeighbor -t [InverseWarp] -t [Affine,1] -v 1

If that doesn’t work, try applying the same transform to another image in MNI-space, brain or brain_mask and confirm that the transform is working properly.

Cheers,
Carl

3

Hi @Shai_P, did this solve your problem?