Enh Add Callosum ROIs support

[bloomdt-uw]

#494

• Depends on merge with [WIP] [temp fix] ignore bundles for plot_tract_profiles when anat position unknown #537 Save non anat positioned bundles in their own plots #539

• Remove DEFAULT_BUNDLES

• Add aal atlas segmentation endpoints for callosum ROIs

• Review MNI_AAL.txt. Plot resulting callosal streamlines with aal endpoints

• Explore additional data sets

• Stanford Data Repo
  (See https://github.com/nrdg/cloudknot/blob/master/examples/02_process_mri_data.ipynb)

[pep8speaks]

Hello @bloomdt-uw! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-12-14 21:48:15 UTC

[36000]

@bloomdt-uw #539 should fix the anatomical plot position problems by saving the plots individually.

[arokem]

@jyeatman : what are the correct endpoints for each one of these segments of the CC?

[36000]

I added issue #614 to remind us to figure out where the callosum endpoints are, but I don't think that is high priority.

[36000]

@arokem this is ready for review.

@bloomdt-uw for your checklist, I moved the endpoint atlas to an issue, so we can merge this now and find out what the correct endpoints are later. I am not sure what the last part of your checklist is referring to, however this may merit its own issue.

I got rid of the test and instead have a callosum example. I also modified some of the examples to use less streamlines in tractography to save space, which I think was causing problems. I also modified the 'organize stanford data' function to optionally delete any afq derivatives folder in the stanfords bids directory. This is because the examples use the same BIDS directory and were using each others derivatives.

[bloomdt-uw]

The task to explore other datasets was meant to be a quality control check

[arokem]

Looks great. My main comment here is that we probably want to prevent users from asking for these new callosal bundles and also for FP/FA in the same AFQ object. The reason is that these are co-located and our current code makes sure that each streamline is classified into only one bundle. This would mean that (for example) the occipital callosum and FP would be competing for the same streamlines, which doesn't make sense.

[36000]

@arokem I integrated your advice. This is what the example produces now.

[arokem]

Looks great! One more small suggestion to follow up on the recent change.

[36000]

I added #616 for exploring different datasets.