Sub-efficient vertebral labeling on MP2RAGE

[jcohenadad]

https://github.com/spinal-cord-7t/coil-qc-code/blob/f9fcb557bc52a6afb02e99a15da0444d90376ee0/data_processing.ipynb#L172

Eg:

Moreover, C2-C3 detection failed in 3 out of 9 cases. So overall, labeling worked in 2 out of 9.

I would try:

• Running it on MP2RAGE T1w (if also doing that for seg)
• Running it on the B1map (if also doing that for seg)
• Forget about automatic labeling and provide manual labels
• Utilize prior information from scanning SOP (eg: some scans were prescribed such that the FOV is centered at a specific vertebral level)

Related to #22

[jcohenadad]

I think we should just go with manual disc labels.