Brainstorm version: 19-Sep-2020
OS: MacOS Catalina 10.15.6
Matlab version: R2020a
When I epoch my data using bst_process('CallProcess', 'process_import_data_event', ...), the events often fall under the wrong group header. For example, event 1 here is sorted correctly, but the event 3 grouping contains epochs from event 9. (They aren't shown in the screenshot, but event 11 is in the event 9 grouping, and 3 is in 11.)
Note that within the grouping, the labels are correct: 9 (#1) does correspond to the first instance of event 9 in the continuous file.
At first I thought the issue was that I was running two different epoching processes, in order to split the epochs into 2 different folders representing trial types that, for the purpose of my current analysis, are equivalent. The events in the first folder are fine, but the second folder is shown above. However, the same problem occurs if I do all of the epochs at the same time, except that it causes event types that were originally correct to go in the wrong group.
Thing's I've tried to fix this:
- import all events individually as separate calls, including separate calls to start & save bst_report
- importing all events at the same time
- importing events using the "Run" interface in the GUI (Import > Import recordings > Import MEG/EEG: Events) instead of scripting
- importing events by right clicking on the link to the raw file > Import in database
- toggling CTF compensation (though this is EEG data, so it shouldn't apply anyway)
It's also not an issue with the filenames in the database. Before the most recent update, the filenames had matched the assigned grouping rather than the actual event type, but that appears to have changed after I updated today. Now, the filenames are correct, but the grouping is still wrong.
The one way I've found to get around this is to toggle createcond ("Create a separate folder for each event type" in the GUI) in order to make a new folder for each event type. This turns out to work mostly fine for my purposes, though it makes working with the data somewhat more tedious. However, since creating folders for each event may not work for everyone, such as described in the epoching tutorial, I thought I would report the bug.