T2w Flow Void Artefacts

[valosekj]

The r20241126 model (included in SCT) struggles with T2w flow void artefacts.

Image details: T2w TSE images "SliceThickness": 3, "SpacingBetweenSlices": 3.6.

From Markus (email thread):

We often have extensive flow void artefacts in our T2w cervical spine MRI

CSF flow void is more common in narrowed conditions (as CSF velocity increases), like In DCM

---

sub-127_ses-M0_acq-axial_T2w.nii.gz:

Markus: I think the black spaces in 16 and 17 are CSF spaces with flow void, while the grey area is bone

---

sub-004_ses-M0_acq-axial_T2w.nii.gz:

Markus: I think underestimation in 13 and 14 (due to this flow void problem, which is common in t2w) – I think also the black spaces are part of the CSF space (you can cross-check if there is visible CSF signal in sagittal slices). It is often hard to distinguish manually between degenerative soft tissue changes and flow void, while we often have a look at neighbouring image slices.
Markus: Overestimation maybe in images 17-19 (not sure, with a second look it could be also appropriate. Probably it is more an underestimation in the other images and right canal segmentation in 17-19.

[valosekj]

It is often hard to distinguish manually between degenerative soft tissue changes and flow void, while we often have a look at neighbouring image slices.

I believe that the 3D architecture of the model could be useful in this case.