less restrictions on BIDS-accepted data file formats

[nbeliy]

I think it could be beneficial to relax restrictions on data formats.

Laboratories might want to adhere to BIDS for structuring their data, but the format is usually dictated by used analysis tools. Restricting to only a selected list of formats could prevent the spread of BIDS. Also, it will avoid the request to add additional formats to standard.

For example, in my laboratory we mostly use SPM12 meeg file format, which is not supported. When we converted to more standard EDF+, we didn't managed to integrate the events in the same file, so we created an additional EDF+ file containing only events. But it is still not BIDS compliant.

The existing restrictions could turned into suggestions, and imposed only for public datasets, where the common popular format should be imposed anyway.

From technical point of view, it could be implemented by imposing the naming of data files as follows: <prefix>_<name>.<extension> where <name> must not contain spaces and underscore, and extension must not be either .tsv of .json. Thus these files will never be confused with BIDS files.

All files with the same prefix are considered to belong to the same recording, and only one of them should appear in scans.tsv (for example the header .vhdr file for Brain vision format).

From analysis (or validation), it will be easy to separate prefixes (and parse the modalities) from the name, by searching the last underscore.

This will greatly increase the flexibility of the BID standard, but still preserve its spirit.