Since spinal cord cross‑sectional area (CSA) estimation is affected by head and neck position in the scanner, I am wondering how this is handled in SCT.
For example, CSA values can differ for the same subject when the spinal cord centerline is perpendicular to the axial slice plane versus when the neck is angled, such that the centerline is oblique relative to the image plane. In the latter case, the axial slices are not orthogonal to the cord, which can bias CSA measurements.
Does SCT take this effect into account when computing CSA (e.g., MEAN(area))? Specifically, does it perform any form of registration, reformatting, or reslicing to straighten the spinal cord and ensure that CSA is measured in planes orthogonal to the local centerline?
This becomes especially important in follow‑up scans, where session‑to‑session variations in spinal cord tilt may introduce systematic bias and compromise the reliability of longitudinal CSA measurements.
Any clarification on how SCT addresses this would be much appreciated.
Hi @marcoganz,
Does SCT take this effect into account when computing CSA (e.g., MEAN(area))? Specifically, does it perform any form of registration, reformatting, or reslicing to straighten the spinal cord and ensure that CSA is measured in planes orthogonal to the local centerline?
SCT’s CSA tool (sct_process_segmentation) processes cord segmentation masks as-is, without any registration/straightening (though we do offer registration/straightening via other tools). Because of this, to account for non-orthogonal cord masks, SCT performs a simple angle correction calculation (controlled by the -angle-corr argument) to adjust the measures based on the angle of the cord relative to the SI axis. More details on this tutorial page: Verify the correctness of the metrics - Spinal Cord Toolbox documentation.
Additionally, because aggregated measures based on vertebral levels can vary due to neck flexion/extension, we offer a PMJ-based CSA tutorial, which attempts to control for neck position by measuring CSA at a fixed distance from a reference point (the pontomedullary junction). More details can be found here: Cross-sectional area (CSA) - Spinal Cord Toolbox documentation
Thank you for all of your useful questions, and please feel free to ask if anything needs clarification.
Kind regards,
Joshua
1 Like