Compute normative values of adapted spinal canal occupation ratio (aSCOR) and spinal canal

[sandrinebedard]

Description

Adapted spinal canal occupation ratio (aSCOR) is commonly computed as a metric to characterize spinal cord compressions. It is defined as the ratio of the spinal cord area divided by the spinal canal area multiplied by 100 at the level of compression.

However, this measure does not consider the inter-subject variability of aSCOR and its variability across spinal levels.

The idea would be to provide normative values of the aSCOR on the spine generic dataset in the PAM50 space.

Steps

• Get CSF segmentation for T2w images (run sct_propseg, perform quality control and manually correct when necessary)

• Compute slicewise the spinal canal area and spinal cord area to compute aSCOR

• Report the normative values and report inter-subject and inter-level variability

• Other shape metrics about the spinal canal could also be intergrated in the database (area, diameter, etc.)

References:

• https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10663304/
• https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8432264/
• https://neurolibre.org/papers/10.55458/neurolibre.00017
• https://www.nature.com/articles/s41393-024-00993-8

[valosekj]

Note: CSF segmentations can be used to extend the contrast-agnostic model to segment also CSF; context in https://github.com/sct-pipeline/contrast-agnostic-softseg-spinalcord/issues/89

[sandrinebedard]

To inlcude in a new function sct_compute_canal_to_cord_ratio

[valosekj]

We now have a first model for canal segmentation from T2-w images trained on spine-generic: https://github.com/ivadomed/model-csf-seg