Image Type Recommendations for SCT DWI Registration

MRI Processing & SCT
1

Hi SCT Team! I hope that you’re doing well!

Apologies in advance, but I’m a new SCT user who has two rather naïve questions to ask regarding SCT registrations.

My colleagues and I are interested in examining the DWI of spinal cord data for people living with stroke and noticed from the scanning sequence of our pilot participant that our T2 data looks much better than our T1. We unfortunately don’t have a lot of scan time, and as such, are in the process of troubleshooting our protocol to make sure that it fits well with the requirements for the SCT.

As such, while going through the tutorials for data processing, however, we became curious about whether there is a certain preference for the type of scanning protocol (i.e., T1, T2, or T2*) to use for anatomical data registration to the PAM50 template?

Likewise, when looking up the information from the SCT Toolbox’s stroke pipeline (GitHub - sct-pipeline/levin-stroke: Repository for levin-stroke project.), the script appears to be primarily relying on T1 version of the PAM50 template for DWI registration, which is completely sensible as this follows the recommendation for the sct_register_multimodal function (Registering dMRI data to the PAM50 template - Spinal Cord Toolbox documentation), however, we were wondering if other types of participant scans (namely T2 or T2*) could be used to get to this step or if the use of a participant’s T1w image is an essential requirement?

Many thanks in advance for your guidance on these matters!

Kind regards,

Alex

2

Hi Alex,

Thanks for reaching out.

There is no single way of doing things. While the tutorials and other examples could be used as inspiration, in the end, the most optimal solution depends on your data. Notably: scanner, model, sequence, parameters, contrast, resolution, field of view, presence of artifacts, etc.

In your case, if you have a good T2w scan, you could use it for registration to the PAM50. You have a couple of solutions:

  • PAM50_T2 → T2 → b=0 scan from the DWI
  • PAM50_T1 → mean DWI scan
  • PAM50 → SC segmentation from the DWI scan

If you decide not to go with the anat scan, you need to get disc levels on the DWI scan directly. It’s doable if you can label vertebrae directly from the DWI scans (depends on data quality), or if the MR tech centered the FOV at a systematic anatomical landmark (eg: FOV centered at C3-C4 disc).

etc.

1 Like
3

Hi Dr. Cohen-Adad,

Thank you for your kind reply and for your guidance on these matters! This makes complete sense!

Kind regards,

Alex