Closed Remi-Gau closed 2 months ago
I've changed the title because it may be the case that data cannot be shared without anonymization and yet you want to share them after anonymization.
It's more about warning that sharing data under sourcedata/ must be done extra-carefully before any damage is done.
Thanks and my bad for the lack of clarity.
Thanks for this folks!
That's something we discussed a lot in BIDSonym where we copy the OG, ie non-pseudonymized/anonymized data from the BIDS
root into /sourcedata/
, apply deidentification there and save the respectively pseudonymize/anonymize in BIDS
root.
This might be another use case to bring up/discuss as there's not only "classic" sourcedata
, ie dicoms
or so, but also BIDS
dataset(s) before pseudonymization/anonymization. I'm of course happy to adapt things given the outcomes of this discussion/new guidelines proposed here.
Thanks again.
we want to include here some language that the specification is completely naive about whether those data can be shared. Along these lines, it would be a good place to mention that source files requiring anonymization can be stored here and still be considered "source data" (although there's been one processing step in between).
The point is, many students and young researchers may be unaware of the circumstances under which the data may or may not shared and not adding some sort of warning or language here to say: hey, discuss with your PI or your university or whoever this needs be discussed before dropping anything and using this as a means of releasing data public.
_Originally posted by @oesteban in https://github.com/bids-standard/bids-specification/pull/1737#discussion_r1534513241_