Hi SCT Team,
I am reaching out to seek guidance on how to estimate the MTR across different spinal cord tracts particularly focusing on areas at, above, and below the regions of compression.
The region(s) of compression are not as readily identified on these images as they are on the T2W.
I would greatly appreciate your recommendation for accurately identifying and delineating these ROIs on MT images. (please email link for shared data)
For a bit of context for others: the data are acquired using the spine-generic protocol and consist of T2w iso (0.8mm) and axial T2star and MT scans. T2star and MT scans cover C2-C6/C7 vert levels.
Since the images are acquired using the spine-generic protocol, you can get inspiration from the process_data.sh processing script designed specifically for the spine-generic data.
A few modifications are needed in your case:
instead of registering T1w image to the PAM50, you will be registering T2w
if you are interested only in MT, you can remove the DWI and T2star parts from the script
Regarding the areas at, above, and below the regions of compression. There are a few considerations on how you define these areas.
The main consideration is that the compression site cannot be (currently) identified automatically by SCT. A possible solution would be to label compression site(s) manually in the T2w image (as shown in this tutorial), then load the compression label during analysis, bring it from the T2w space to MT space, and compute metrics at the level of compression(s). HOWEVER, interpretation of MT metrics at the compression site must be taken with a grain of salt because MT data might be biased by low SNR and partial volume effect.
You can compute metrics from the level above the compression automatically if you choose a reference level, for example, vert level C2 – compression usually does not appear at this level.
The level below the compression is more tricky. Since compression(s) occurs predominantly between the C4/5 and C6/7 cervical levels (see for example Introduction in this review), the level below the compression would be C7 or even T1. These levels are not covered by your MT data, though.
TLDR: you can automatically compute metrics above the compression. I drafted a processing script (see the attachment). Note that the script requires JSON sidecars for MT files for the sct_compute_mtsat calculation.
Please let me know if you have any further questions.
Thank you for the guide. Please could you share information on how to generate the JSON sidecars?
The JSON sidecars for all images (including MT) are usually generated during the DICOM to NIfTI/BIDS conversion. For example, the popular dcm2bids conversion tool generates these JSON sidecars automatically based on DICOM headers.
Thank you Jan,
To clarify my question, yes, we have few all DICOM data with their JSON files, however most of our MRI data are the HEAD and BRIK files (AFNI). I am not sure which ones are the JSON files. files.E18463.zip - Google Drive
The link above is an example of the Raw data and the converted NIFTI files.
I will appreciate any guidance to identify/create the JSON files.
Thanks for the clarification and example data! Indeed, your source data are in the AFNI format (HEAD and BRIK files). Unfortunately, I have never worked with the AFNI format, so I do not have insights on how to convert the AFNI format to BIDS properly while obtaining JSON sidecars. I did a quick search on the NeuroStars forum and found this relevant thread, which recommends the usage of DICOM, though.
Thank you Jan for your help and reaching out to your colleagues on our behalf. The good news is we have a way to retrieve all the Dicom files for all the AFNI files. I will keep you posted once I get them.