search

sct-pipeline / dcm-metric-normalization

MIT License
0 stars 0 forks source link

SC straightening changes spinal cord anatomy #3

Open valosekj opened 1 year ago

valosekj commented 1 year ago

Description

While testing different settings of sct_register_to_template to register T2*w axial image to the PAM50 template, I discovered that -ref subject preserves spinal cord anatomy, but -ref template does not.

-ref subject

sct_register_to_template -i t2s.nii.gz -s t2s_seg.nii.gz -ldisc labels.nii.gz -ref subject -c t2s -param step=0,type=label,dof=Tx_Ty_Tz_Sz:step=1,type=seg,algo=centermass,dof=Tx_Ty_Tz -ofolder ref_subject

-ref template

sct_register_to_template -i t2s.nii.gz -s t2s_seg.nii.gz -ldisc labels.nii.gz -ref template -c t2s -param step=0,type=label,dof=Tx_Ty_Tz_Sz:step=1,type=seg,algo=centermass,dof=Tx_Ty_Tz -ofolder ref_template

Comparison (notice the changed SC shape/anatomy for the -ref template):

image

The main difference between -ref subject and -ref template is that within the -ref template, the straightening of the spinal cord is done (see the help of sct_register_to_template for details).

To verify if the straightening causes the anatomy alteration, I run sct_straighten_spinalcord:

sct_straighten_spinalcord -i t2s.nii.gz -s t2s_seg.nii.gz

And indeed, sct_straighten_spinalcord modifies the spinal cord anatomy:

image

Steps to Reproduce

Example T2*w data: t2s.zip

Thanks @sandrinebedard and @jcohenadad for the cooperation!

jcohenadad commented 1 year ago

Thank you for the nice summary of the problem @valosekj ! I suggest you incorporate this in the manuscript (maybe it could be an annex?)