CSA at vertebral levels


I am finding CSA and I am wondering whether finding CSA at each vertebral level includes either its upper or lower disc as well or not? because as I see the contraction happens at disc level and I want to know which level has lesser area, and without inclusion of disc level it won’t be meaningful. In other words, in CSA area calculations, do discs belong to upper or lower vertebral levels? Thank you.


Vertebral levels are defined based on the disc levels. So, let’s say, the interface between vertebral levels C3 and C4 would be at the C3-C4 disc. So one way to look at it, is that both C3 and C4 levels include some parts of the C3-C4 disc level. I hope that’s clear.

Yes, thank you. It’s now clearer. C3/C4 belongs to both C3 and C4 and if there is any contraction, it’s effect exist on both C3 and C4. Correct? Thank you again.

That is correct, yes, but you need to keep in mind that if the spinal cord deformation only occurs at the disc level and that is what you’d like to capture, then it is suboptimal to aggregate CSA values across the entire vertebral level, because it will average out the effect you’d like to capture.

Alternatively, it might be more appropriate to only compute CSA at the disc level. In that case, you simply need to select the labeled disc file (output of sct_label_vertebrae) when running sct_process_segmentation. Example below:

sct_process_segmentation -i t2_seg.nii.gz -vertfile t2_seg_labeled_discs.nii.gz -vert 2:5 -perlevel 1 -o csa_disc.csv

Here’s the output: csa_disc.csv (1.6 KB)

Where each line gives you the CSA at a specific disc level (notice the single slice selection). For example, a value of ‘3’ under the row ‘VertLevel’ corresponds to disc C2/C3, according to this convention.

Thank you so much. I got your point. finding CSA at disc level could be better. but in the end, I want to have a spreadsheet including CSA (from a T1 image) and fa (from a DWI image). should I compute fa at disc level as well? Thank you.

It depends what the underlying hypotheses of your study are. One thing to keep in mind is that computing FA on a single slice only is prone to less accuracy and precision (less voxels used to generate statistics).

1 Like

In this situation is there any way to find fa with more slices around disc level (to reach more precision)? or we can only find it each vertebral level? (because if I want to find CSA at disc level, I think the fa should also be computed around the disc). Thank you very much.

sure, you could e.g. dilate the disc labels (in the example below by 3 slices above and 3 slices below):

sct_maths -i t2_seg_labeled_discs.nii.gz -dilate 3 -o t2_seg_labeled_discs_dil.nii.gz

And then use that mask as input:

sct_process_segmentation -i t2_seg.nii.gz -vertfile t2_seg_labeled_discs_dil.nii.gz -vert 2:5 -perlevel 1 -o csa_disc.csv