🛠️ Updating spinal levels - feedback needed

[jcohenadad]

Dear SCT users,

After getting feedback from some of our users about the spinal levels available in SCT, we decided to make some important modifications on the spinal levels (PAM50/spinal_levels).

Please note that these modifications will introduce a break in cross-compatibility in a future SCT release. Notably:

• The spinal levels won’t be encoded as individual files per level, but as a single file for all levels. Similarly to the vertebral level file (template/PAM50_levels.nii.gz), the future spinal level files will be located under template/PAM50_spinal_levels.nii.gz and will systematically be warped with the function sct_warp_template;
• Consequently, the flag -s in sct_warp_template will be removed;
• The spinal levels won’t be encoded as probabilistic value anymore, but as integer. This is because the previously used probabilistic values were an extrapolation along the entire spinal cord, which as some of you pointed out, is likely inaccurate;
• The spinal levels will be based on the Frostell et al. paper;

At this stage, we would like to get some feedback from SCT users (you!), to make sure the proposed implementation will be welcome and not too disruptive. If you have any concerns or suggestions, please share them here.

Related issue and pull request on the PAM50 repository:

• Update spinal segments using Frostell et al. paper · Issue #16 · spinalcordtoolbox/PAM50 · GitHub
• Update spinal levels based on Frostell et al. paper by jcohenadad · Pull Request #18 · spinalcordtoolbox/PAM50 · GitHub

Related forum discussions:

• Is it possible to label all 29 levels of spinal cord?
• PAM50 - Spinal levels under T12
• Labels for lumbar spinal level
• Probabilistic spinal level masks: what is the metric? (sct_warp_template -s)

[Nawal_Kinany]

Dear Julien & SCT team,

The proposed implementation seems satisfactory to me and I have no particular concern to raise.
Thanks a lot for this very useful update!