use bids-matlab to create derivatives filenames

[Remi-Gau]

Quick FYI that I have added some functions in the dev branch of BIDS-matlab to help with derivatives filenames and JSON creation. And also to help "initializing" derivatives datasets.

See the Jupyter notebook demo: https://github.com/bids-standard/bids-matlab/blob/dev/examples/BIDS-Matlab_filenames-and-JSON.ipynb

One thing that would be nice to do from there is create a function (set of functions that can loop through the BIDS hierarchy of a dataset preprocessed with SPM and rename the files into something BIDSy).

fMRwhy does that. CPP_SPM will need to do that. I suspect there is a bazillion of other toolboxes that could benefit from having a repo with those functions.

Because it is SPM specific though, I don't want to put that in BIDS matlab, though BIDS matlab could have a more abstract overarching function that can help with this.

Let me know what you think

[jsheunis]

This is really cool Remi, thanks for posting here!

I think helpers to create/update SPM-processed derivative data would be useful. There are some questions that jump to mind:

• Like you mention, perhaps bids-matlab is not the best place for these to live. In the absence of anything else, I will (and have) implemented some basic functionality like this in fMRwhy. But do you think a separate package would be better?
• There would have to be some consensus on how to name these derivatives, if it will be a single set of functions that are used widely. The SPM processed filenames are mostly easy to parse, and the bids structure is easy-ish to adhere to. We would just have to suggest a "standard" (e.g. using desc-sra_bold.nii for smoothed, realigned, slice-time corrected data [in reverse order]).

[Remi-Gau]

* Like you mention, perhaps bids-matlab is not the best place for these to live. In the absence of anything else, I will (and have) implemented some basic functionality like this in fMRwhy. But do you think a separate package would be better?

yes a separate package or repo would be easier: easier for people to grab (fewer files and don't need to grab a whole pipeline and then ditch 99% of it) and add as library, but also to maintain and have people contribute back.

* There would have to be some consensus on how to name these derivatives, if it will be a single set of functions that are used widely. The SPM processed filenames are mostly easy to parse, and the bids structure is easy-ish to adhere to. We would just have to suggest a "standard" (e.g. using `desc-sra_bold.nii` for smoothed, realigned, slice-time corrected data [in reverse order]).

I would suggest we give some sensible defaults but make sure there is an easy way to map a certain prefixes into a series of sensible BIDS entity-label pairs.