HCP HippUnfold and Post HippUnfold Implementations

[glasserm]

This is an implementation of the current version of HippUnfold 1.5.2-pre.2: https://hippunfold.readthedocs.io/en/latest/. This requires the current development version of Connectome Workbench to use. The file organization is non-final and needs evaluation and comment. There are a variety of unfinished things marked with TODO. Some of these are pending comments and some pending a new planned HippUnfold 2.0 release.

For those who have local access, the data are in here: myelin/brainmappers/Connectome_Project/YA_HCP_Final/100307/T1w/HippUnfold and MNINonLinear/HippUnfold

[demsarjure]

I reviewed this and added some minor comments/discussion points.

[coalsont]

I undid the "single loop body moved to function" and used a new variable for the space bids tag instead of clobbering Modality. I also made the physical output folder variable name consistent. Please test.

[mharms]

Just getting around to taking a look at this. So, is the next version of HippUnfold going to support direct specification of input files, rather than having to hack an intermediate "BIDS-style" input directory?

In terms of organization, I was wondering if it wouldn't be helpful to segregate all the *.surf.gii output files for the different mesh resolutions into their own surf directory, and same for all the *.shape.gii files, so that the *.dscalar.nii files more readily stand out on their own. (Yes, I know that would be different from the organization of the brain data, but this is a fresh chance to give a little more structure to the output organization).

Can you remind me, how does one actually view the tessellation of the surfaces? I tried zooming in a lot, but I still wasn't getting to see the actual triangles.

[glasserm]

1. That is the hope.
2. I prefer to maintain parallelism with existing HCP organizations and not generate a bunch of subfolders.
3. That is under the surface menu in Connectome Workbench.

[mharms]

3. That is under the surface menu in Connectome Workbench.

Ahh, found it. Surface -> Properties, and then change the "Drawing Type" to "Links (Edges)"

[glasserm]

Now supports and depends upon HippUnfold Version 2.0.0.

[demsarjure]

Hi, I already have this edited on my end to support containers and hippunfold directly. Will commit once it is tested. Jure

…

On Saturday, 10 May 2025, Tim Coalson ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In HippUnfoldHCP/HippUnfoldHCP.sh <#320 (comment)> : > +log_Msg "Running T1w HippUnfold for subject: $Subject" #Seriously: don't put a $ here and don't capitalize S in subject... #Seriously: don't put a $ here and don't capitalize S in subject... +apptainer run --bind $StudyFolder -e $HIPPUNFOLDPATH $HippUnfoldT1wFolder $HippUnfoldT1wFolder participant --modality T1w --path-T1w $HippUnfoldT1wFolder/s_{subject}_T1w_acpc_dc_restore.nii.gz --path-T2w $HippUnfoldT1wFolder/s_{subject}_T2w_acpc_dc_restore.nii.gz --cores all --force-output --generate_myelin_map --output-density native 512 2k 8k 18k +log_Msg "T1w HippUnfold completed." +log_Msg "Running T2w HippUnfold for subject: $Subject" #Seriously: don't put a $ here and don't capitalize S in subject... #Seriously: don't put a $ here and don't capitalize S in subject... +apptainer run --bind $StudyFolder -e $HIPPUNFOLDPATH $HippUnfoldT2wFolder $HippUnfoldT2wFolder participant --modality T2w --path-T1w $HippUnfoldT2wFolder/s_{subject}_T1w_acpc_dc_restore.nii.gz --path-T2w $HippUnfoldT2wFolder/s_{subject}_T2w_acpc_dc_restore.nii.gz --cores all --force-output --generate_myelin_map --output-density native 512 2k 8k 18k +log_Msg "T2w HippUnfold completed." +log_Msg "Running T1wT2w HippUnfold for subject: $Subject" #Seriously: don't put a $ here and don't capitalize S in subject... #Seriously: don't put a $ here and don't capitalize S in subject... +apptainer run --bind $StudyFolder -e $HIPPUNFOLDPATH $HippUnfoldT1wT2wFolder $HippUnfoldT1wT2wFolder participant --modality T2w --path-T1w $HippUnfoldT1wT2wFolder/s_{subject}_T1w_acpc_dc_restore.nii.gz --path-T2w $HippUnfoldT1wT2wFolder/s_{subject}_T2w_acpc_dc_restore.nii.gz --cores all --force-output --generate_myelin_map --output-density native 512 2k 8k 18k --force-nnunet-model T1T2w +log_Msg "T1wT2w HippUnfold completed." Is someone going to edit this to have the possibility of running with --use-conda instead of their container? The warnings probably deserve a real newline instead spaces after the log_Msg, and we may not need quite that many repetitions. — Reply to this email directly, view it on GitHub <#320 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIQ6TOR5BKN2SUSD46J2SLD25U25HAVCNFSM6AAAAABUZATW7KVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDQMRZHA4TQNJWGQ> . You are receiving this because your review was requested.Message ID: ***@***.***>

[coalsont]

A broader question...why is "HCP" contained in the folder name that the scripts are in, and in the script filenames? The fact that they are in this repository is good enough for all the other folders and scripts (SetUp... and related notwithstanding, because we expect some users to copy it outside the pipelines folder before editing it).