Write HOWTO read an fMRIPrep report

[oesteban]

A step-by-step guide for users to do QC.

[effigies]

While looking through a report with an eye to teaching it, I'm noticing there are a lot of terms that are undefined, such as DVARS and aCompCor. I think it would be good to link out to definitions, possibly in our docs, possibly external.

We could start a glossary, so that we can put persistent URLs in the docs, and then have up-to-date references on the website.

This may be better to write as a separate issue...

[effigies]

Also, I think the demos we've given (https://oesteban.github.io/fmriprep-demo, https://effigies.github.io/fmriprep-demo) should provide a pretty good skeleton to a howto. There'll be a lot more space to flesh it out.

And really, pretty much anything that should go in this HOWTO should probably go into the reports themselves. Only advantage of the HOWTO is that we can show examples of errors alongside good reportlets.

Sorry, I'm in a slight stream-of-consciousness mode as I write slides, so this is not well-edited.

[oesteban]

Just found out this is a duplicate of #1524.

[sarenseeley]

A related issue: the images in the sample report linked from the readthedocs page are broken, and the report is from an early version (0.5.5-dev) and no longer reflects what users will actually see in their own reports: https://fmriprep.readthedocs.io/en/stable/_static/sample_report.html

[sarenseeley]

Hey all, I've been sitting on this issue WAY too long - sorry! Applying for clinical residency and finishing my dissertation had to take priority these past months. I would love to get some other eyes on my draft and maybe people can fill in some of the gaps?

Also would be helpful to have visual examples of desirable/undesirable outcomes.

https://docs.google.com/document/d/1TE6ZWzNg8cDpvL4Vu0VGOZQLXkQ88Fa59AORzN01Avk/edit

[oesteban]

I would love to get some other eyes on my draft and maybe people can fill in some of the gaps?

I'm halfway through it, will try to finish before OHBM ends - but that's going to be hard. Probably this extends to the rest of the members of the community.

Also would be helpful to have visual examples of desirable/undesirable outcomes.

Yes, I think we have a good bunch of problematic examples on neurostars and the github's issue tracker. For good outcomes it's probably easier.