Create derivative and json sidecar for QC-ed automatic segmentations

[sandrinebedard]

Context

SC segmentations that are manually qc-ed and that don't need manual corrections need to be added to derivatives and a json sidecar needs to be created.

Idea

A way would be to adjust the script manual_correction.py and pass as an argument the list of files that need correction (as usual but the files would be exlcluded instead). The rest of the files would be added to derivatives folder and a json sidecar would be created.

I would only make this available for segmentations since we don't automatically create disc labeling for C1-2, C2-3 and C3-4 as we do manually.

Tasks

• Add flag to specify that the list from -config are the subjects to exclude
• Go through all images except exluded ones (ONLY T1w images)
• Add segmentation with suffix -manual to derivatives
• Create jsonsidecar
• Don't call function correct_segmentation where ITK-snap is opened

[sandrinebedard]

@jcohenadad do we want to also exclude from this images in exclude.yml ? or we still keep the segmentations from those images

[jcohenadad]

@jcohenadad do we want to also exclude from this images in exclude.yml ? or we still keep the segmentations from those images

hum... good question. I would keep the images because in the future we might want to come up with a deep learning algorithm that finds "bad looking" images. So having the images and the exclude.yml file would be perfect to train such model

[sandrinebedard]

Ok great!