Test the model on a single subject with different acquisitions

[abelsalm]

Getting to the phase where I would want to test the model, as suggested during dcm normalization meeting, it would be good to test its consistency by applying it to various subjects that would have different T2w acquisitions, to see if the segmented volumes are matching.

May be @jcohenadad @sandrinebedard you could help me to access relevant datasets to perform this.

[valosekj]

You could try the datasets I used for the rootlets segmentation:

• spine-generic single-subject - single healthy subject scanned across nineteen sites on 3T scanners from three vendors
• Courtois-NeuroMod - healthy subjects scanned at regular intervals (for example, sub-001 has 10 sessions) at the same 3T scanner
• OpenNeuro ds004507 - 10 healthy participants, each scanned three times (different neck positions)

[jcohenadad]

just to clarify: Jan pointed to you to the SAME sequence acquired in the SAME subjects across time-- what you ALSO want is to compare segmentation in the SAME subjects across DIFFERENT sequences.

pls update your manuscript with a clear list of experiemnts you are planning to do for validation, so we are all on the same page--

[abelsalm]

pls update your manuscript with a clear list of experiemnts you are planning to do for validation, so we are all on the same page--

ok sure !

just to clarify: Jan pointed to you to the SAME sequence acquired in the SAME subjects across time-- what you ALSO want is to compare segmentation in the SAME subjects across DIFFERENT sequences.

but which dataset I should look into for those subjects with different sequences ?

[jcohenadad]

but which dataset I should look into for those subjects with different sequences ?

all these datasets are good (ie: they have different contrasts per subject)-- but don't limit yourself to these datasets-- eg: spine-generic ALSO has multicontrast per subject