Hi,
It seems that the two outputs have similar structure but the subject specific white matter and ventricle masks were used for DCAN but a standard mask was used for HCP YA. Why is that?
Cindy
We don’t use these ingression functions very often, but I think that the DCAN derivatives include tissue masks derived from the subject’s data, while HCP derivatives don’t. However, there’s always room for improvement, so if you find appropriate tissue masks in the HCP derivatives, we can update the ingression function to use those instead of generic ones.
Thanks for the reply. So both T1w from abcd-bids and hcp pipelines are in standard space right? In theory I could use the generic white matter mask for DCAN derivatives too?
I think DCAN got the eroded masks from freesurfer segmentation file using fslmaths. I can attach that snippets later but I wonder how much difference that will make. Probably it’s fine to just use the standard mask?
That’s right, the images should all be in standard space in both pipelines.
The mean white matter and CSF time series are extracted using those masks and may be used for nuisance regression, so I’d actually recommend eroding the standard masks in the HCP-YA ingression over switching from eroded subject-specific ones to non-eroded standard ones in the DCAN ingression.
Okay. Is the ingression function automatically called if we specify the input type? Can we make new input types?
The ingression functions are hardcoded based on the input type. If you want to add a new input type, I’d recommend opening a pull request to the GitHub repository with the new ingression function(s).