T2-based registration of DWI

SCT
1

Hello,

I hope you are doing well.
I wanted to do T2-based registration for the DWI scan. Here is the code I have used in registration part. I am not sure if changing the PAM50_T1 to PAM_T2 in sct_register_multimodal is correct or not? Also, does it have any effect on the final results of DTI metrics or not?

# Register template->dwi (using template-T2w as initial transformation)
sct_register_multimodal -i $SCT_DIR/data/PAM50/template/PAM50_t2.nii.gz -iseg $SCT_DIR/data/PAM50/template/PAM50_cord.nii.gz -d ${file_dwi_mean}.nii.gz -dseg ${file_dwi_seg}.nii.gz -param step=1,type=seg,algo=centermass:step=2,type=im,algo=syn,metric=CC,iter=5,gradStep=0.5 -initwarp ../anat/warp_template2T2w.nii.gz -initwarpinv ../anat/warp_T2w2template.nii.gz
# Rename warping field for clarity
mv warp_PAM50_t22${file_dwi_mean}.nii.gz warp_template2dwi.nii.gz
mv warp_${file_dwi_mean}2PAM50_t2.nii.gz warp_dwi2template.nii.gz
# Warp template
sct_warp_template -d ${file_dwi_mean}.nii.gz -w warp_template2dwi.nii.gz -qc ${PATH_QC} -qc-subject ${SUBJECT}

Thanks and regards,
Maryam

2

Hi,

It’s all a question of contrast: the mean DWI scan has bright cord, dark CSF.

  • The PAM50_t1 contrast has bright cord / dark CSF
  • The PAM50_t2 contrast has dark cord / bright CSF.

So in this case it makes more sense to use the PAM50_t1 contrast (unless you have good reasons to do otherwise).

Also, does it have any effect on the final results of DTI metrics or not?

I’m not sure I fully understand your question (ie what is the “it” in “does it have any effect”), but in general, any differences in registration performance will be reflected in the extracted metrics (sometimes it will be negligible). Bottomline is: your registration needs to be as good as possible.

3

So the only change is in: -initwarp ../anat/warp_template2T2w.nii.gz, is that right?
and by it, I meant the registration result based on T2 compared to the default (T1). I see better segmentation and resolution in T2w images, that’s why I thought maybe this change in registration, could result in more accurate DTI metrics.

4

the only change compared to what?

I see better segmentation and resolution in T2w images, that’s why I thought maybe this change in registration, could result in more accurate DTI metrics.

I think there is a fundamental misunderstanding here. The decision to choose PAM50_t1 or PAM50_t2 has nothing to do with quality of segmentation. Your destination image is the DWI scan. You do the segmentaiton on the DWI scan, not on the PAM50 images.

5

Compared to default code for registration of template to DWI.

Yes you’re right, I understand that T1w is the most similar contrast to DWI.

But as you mentioned:

How can one decide about choosing T1 or T2 for transformation in -initwarp flag part of the code (to increase the registration performance)?

6

What do you mean by the “default” code? There are a lot of tutorials and pipelines with SCT, so it is difficult for me to know what code you are referring to. Please add a link to the exact line of code you are referring to.

What is indicated by the -initwarp refers to the previous registration done (ie on another contrast). So if you did PAM50 registration using a T2 contrast, then you would point to that warping field. If you did a registration using PD, or T1w or T2*w or any other contrast, then you would point to the associated warping field.

I strongly recommend you go through the tutorial section, which will help you understand the core principles behind SCT (registration across contrasts, etc.).

7

This link (I was inspired by the T1, T2 and DWI sections).

Yes, I understood what you explained. Thanks very much, Julien.

1 Like