[bids_import] Participants should have a ref to their own coordinates metadata file

aces / Loris-MRI

The set of scripts that preprocess and insert MRI data into the database.

10 stars 50 forks source link

[bids_import] Participants should have a ref to their own coordinates metadata file #1028

Open jeffersoncasimir opened 9 months ago

jeffersoncasimir commented 9 months ago

Instead of grep_or_inserting physiological_coord_systems (feature added by #885), every participant should have a ref to their own json file.

Nothing prevents a participant from referring to the file of a (typically the first) participant that belongs to another dataset. If the other dataset gets deleted, the FilePath becomes invalid and this breaks functionality.

This will also enable letting users download this file in the Electrophysiology Browser. It is currently possible, but the file might not have the correct BIDS-prefixed name.

christinerogers commented 6 months ago

this could be fixed or replaced with generic montages -- @jeffersoncasimir could you please:

add this to the Loris-EEG agenda for Feb.23.
reply here with a better 3-line high-level summary than this --Standard coordinate system mappings are not currently stored in LORIS since they are assumed to match a standard montage, and aren't provided in the pipeline. However, anytime a user downloads / exports data it would be better to include a (generated) coordinate file, properly named for the datasets. The time cost of re-packaging large downloads to include this generated JSON on demand is not ideal, so an elegant solution would be nice.

jeffersoncasimir commented 6 months ago

Coordinate systems are stored in LORIS as the combination of `NameID`, `TypeID`, 
`UnitID`, `ModalityID` and `FilePath`.

If a coordinate file gets ingested and it has a combination of (the first four) fields that already 
exists, a new entry (with that file's `FilePath`) is not created.

All subjects sharing the same fields will therefore reference the first participants' file, 
which contains their subject name in the path.

We may also want to consider storing generic montages that we can add to datasets that describe the cap but don't include electrode positions (Related issue: #1027)