search

bids-standard / bids-specification

Brain Imaging Data Structure (BIDS) Specification
https://bids-specification.readthedocs.io/
Creative Commons Attribution 4.0 International
274 stars 157 forks source link

Information duplication should be avoided: example "Landmark photos" #361

Open yarikoptic opened 4 years ago

yarikoptic commented 4 years ago

Currently two similar subsections:

(git)lena:~/proj/bids/bids-specification[master]git
$> git grep -i "Landmark photos"
src/04-modality-specific-files/02-magnetoencephalography.md:## Landmark photos (`*_photo.jpg`)
src/04-modality-specific-files/03-electroencephalography.md:## Landmark photos (`*_photo.jpg`)

I think it is worth moving that subsection into appendix and referencing from the text, or making it some other way having its own single presence (e.g. referring to MEG subsection from EEG section).

What do you think?

effigies commented 4 years ago

@bids-standard/raw-electrophys

sappelhoff commented 4 years ago

It is highly likely that there is lots of duplication between the MEG, EEG, and iEEG modality specific sections.

This is the result of EEG and iEEG trying to model after what has been proposed in MEG ... and then trying to come as close to each other as possible.

All of these developments happened in separate Google Docs, which were then separately introduced into the spec.

This historical tangent aside, I do agree that duplication within the specification is not great and it would be less error-prone and maintainable to store information in one place, and refer to it from wherever the information is important to mention.

teonbrooks commented 4 years ago

are we proposing having a top-level electrophys grouping that has common parts and the specific underneath it? are there other sections that are identical. i think i recall there are a few sections that are nearly the same but minor tweaks between the MEG and EEG (like coordsystem or channels, i think). it would be nice to note what is truly shared that could be perhaps refactored and what cannot.

one thing to just note is that the three ephys will seemingly converge onto a similar path with derivatives with greatly deduped workflows.

effigies commented 4 years ago

are we proposing having a top-level electrophys grouping that has common parts and the specific underneath it?

This makes sense to me as an outsider.

robertoostenveld commented 4 years ago

The part which is obviously in common is the _meg.json, _eeg.json and _ieeg.json content. There are also significant differences, especially in the sidecar files that contain the information which cannot be directly obtained from the raw sourcedata (electrode type and material, placement, implicit reference schemes, etc), which might become larger once we start adding other electrophysiological data (multi electrode array, or EMG) that are now in a preparatory (partially BEP) state. Hence I propose not to change things.

It is something to also keep in mind when adding CT and PET, which presumably will have a lot in common with MR as all being imaging modalities... But then, I don't think that a PET or CT researcher wants to be squeezed in the MR straitjacket, just as we have seen for EEG and IEEG, which don't fit the MEG constraints very well.

The duplication is now (a bit) hard for us as maintainers, but if we merge modalities that researchers don't see as each others' natural extensions, we will make it hard on the readers of the spec. My impression is that EEG, iEEG and MEG folks won't be reading each others specs, there are only few that work with two of them, or all three.