search

BioMedIA / dhcp-structural-pipeline

structural analysis of neonatal brain MRI scans
Other
52 stars 16 forks source link

code version for dHCP release 3? #26

Open maxpietsch opened 2 years ago

maxpietsch commented 2 years ago

What version of this code was used for dHCP release 3? @jcupitt do you know?

jcupitt commented 2 years ago

Hi @maxpietsch,

Sorry for the long silence. Jianliang ran the pipeline for rel3 using the dockerhub binary built from the Dockerfile in git master. If you look at that plus the associated scripts you'll see the versions of this package and the various sub-packages and dependencies which were put together to make that binary.

After the run, we had to do a range of emergency repairs to fix up problems people reported, eg. the defacing, fixing surface generation, and others. Some of these fixes have been integrated back into the source (the surfaces, for example), but there was not time to reintegrate and retest them all. I did many evenings and weekends as it was (as we all did).

The scripts I wrote for the fixups are in my brain-atlas repo, which unfortunately I've had to make private. There's a clone in /vol/dhcp-derived-data which you should be able to find easily -- it has directories called things like "deface-kings-data2".

ecr05 commented 1 year ago

Sorry to pick this up again now @jcupitt @maxpietsch. But I don't understand all of the above, and I'm struggling to remember all that went on previously. I could do with some help - if you can remember.

We are trying to compare data from the dHCP to a clinical (CHD) dataset and I am fairly sure we have run different versions of the pipeline. If we are just interested in the surfaces and not the defacing, is the current released version of the docker the right one to use if we want to run surfaces that are compatible with release 3 from the dHCP?

Thanks in advance

Emma

jcupitt commented 1 year ago

Hi again Emma, nice to hear from you.

Sorry, this is all a long time ago now and subsequent projects have pushed the detail out of my brain. It would take a few months work to get back into this again.

I wrote some long-ish handover notes for Daniel and Amir, I'm not sure I could add anything useful to them now.