Hey!
I am analysing the OpenNeuro datasets to create interoperability for machine learning and Deep Learning pipelines https://eegdash.org/, and I have one question about the Epoch dataset
I noticed some datasets are saved in epoch format.
In the EEG BIDS section, it mentions that the data can be saved as either epoch or raw, with the field indicated. This description leads me to believe that we can save this in the folder
sub-001/
— data with epoch format
I understand that epoch data may not be a derivative, but rather just a manipulation of raw data to stack based on the event. However, this isn’t super clear, even semantically, to the point that the maintainers of mne understand it should be a derivative.
I understand that this may have been an extensive discussion when BIDS-EEG was proposed, but I wasn’t there. Currently, in OpenNeuro, we have the following datasets saved as epochs:
| OpenNeuro ID |
|---|
| ds003805 |
| ds004785 |
| ds004368 |
| ds004771 |
| ds004554 |
| ds004520 |
| ds004521 |
| ds007081 |
| ds004519 |
| ds005946 |
| ds006367 |
| ds006370 |
| ds004166 |
My question is more about the official position on where this type of file should be saved, and if it’s in derivatives, can we consider a plan to move or remove these non-compliant datasets from OpenNeuro and update the validator?