search

MICA-MNI / z-brains

z-brains: open software for brain lesion lateralization and localization
https://z-brains.readthedocs.io
BSD 3-Clause "New" or "Revised" License
10 stars 4 forks source link

Native space feature sampling #8

Closed royj23 closed 8 months ago

royj23 commented 1 year ago

Should we map the surfaces to sample features in each modality's native space?

royj23 commented 1 year ago

Or at least make it consistent across cortex and hippocampus

royj23 commented 1 year ago

Friday 5pm verdict: everything goes to nativepro!

jordandekraker commented 1 year ago

Finally just got to reading through some of the code here!

Maybe we should just wait on this issue. Micapipe will soon output a copy of each preprocessed image in space-nativepro. If that's the case then we can skip the registration steps performed here, but we have to wait for that change to be implemented in micapipe first. Thoughts?

I'm also happy to work with the registrations as they are now, it'll just take some time to get the transforms right

royj23 commented 1 year ago

Makes sense to me @jordandekraker! No need to double the workload. One note for FLAIR though, the nativepro transform could be done in z-brains, as micapipe doesn't do anything with this modality (would do this in the z-brains proc_flair script). @rcruces what's the timeline for this and anything we can help with?

rcruces commented 8 months ago

The final verdict:

-proc_flair was updated in micapipe, its outputs are in nativepro space and everything is mapped from there. The latest version of micapipe v0.2.3 solve the normalization issue in flair processing. There is no need to do intra-subject normalization.