I want to divide the spinal cord into left and right sides to compare the degree of atrophy between the two sides. However, it seems that the file $SCT_DIR/data/PAM50/atlas/info_label.txt has been deleted in version 6.5. Is there still a way to measure the CSA and volume of the left and right sides of each segment of the spinal cord separately? Is this method reliable?
Or are there other ways to determine the segment of spinal cord atrophy in patients?
Dear @jiang,
My apologies for the late reply. I am just returning from the weekend.
v6.5 is associated with the PAM50 release r20240215 (download.py). This release does contain the atlas/info_label.txt file. So, this suggests that your copy of the PAM50 template is missing/broken/etc.
It is possible that this is related to the previous network errors that you encountered. I would recommend redownloading the PAM50 template, either via the command 'sct_download_data -d PAM50' or by manually downloading the link and extracting it to '$SCT_DIR/data/PAM50'.
Kind regards,
Joshua
Also, as you mentioned, you would use the info_label.txt file to define the left/right hemi-cords, as described in this page of the tutorials.
I’m not sure I can speak to the reliability of this method. I will talk with my colleagues, who have experience working with spinal cord compression analysis.
(Please note that SCT has a tutorial specifically for spinal cord compression that may be of use to you: Quantify spinal cord compression - Spinal Cord Toolbox documentation)
Kind regards,
Joshua
Dear @jiang,
Defining left and right hemicord masks is not straightforward, as it depends on several factors such as spinal cord rotation, segmentation quality, and the presence of pathologies like cord compression. You might find this previous thread helpful: How to create a hemicord mask.
Best, Jan
Our study subjects are patients without spinal cord compression. This group typically exhibits spinal cord atrophy on one side. Our goal is to determine the extent and stage of spinal cord atrophy by comparing the volume and cross-sectional area (CSA) of the spinal cord on both sides of the same patient. Are there reliable methods for segmentation and analysis?
@valosekj shared the following insight with me on our lab Slack:
Regarding the left and right hemicord CSA, it’s still very much a WIP.
We are currently exploring different approaches (defining the hemicords using the AP diameter versus a template-based method). At this point, I think we can’t recommend one over the other (especially in cord compression or when detecting a subtle atrophy). Personally, I’d note that it needs further research
In terms of available SCT tools, I do not believe that we have specific tools designed to detect asymmetrical LR cord atrophy. So, we would have to proceed with the available tools.
One thing that I may suggest (if you have T2* data available) would be to look at the “GM-informed template registration” tutorial (in other words, performing registration that accounts for the shape of the gray matter). By segmenting both the gray matter and the cord, the resulting white matter mask (SC minus GM) should hopefully properly represent any asymmetrical cord atrophy. Then, when using this mask to register the template, the PAM50 atlas files would conform to the WM mask, which would preserve the atrophy in the “CombinedLabel” hemicord masks. (Then, you would proceed with the template-based approach linked by @valosekj in previous replies.)
However, I should note that I have not actually tried this approach – I am brainstorming based on the available SCT tools.
Note: If you have sample data that you can provide, then perhaps we could take a look as well. Our lab is in the process of developing approaches for determining spinal cord orientation, symmetry, etc. which is closely related to the problem of sub-dividing the cord into hemicords. And so, any sample data would be greatly helpful for us as well.
Kind regards,
Joshua