Question about autoreject that reject non epoched data

Support & Discussions
1

Hi everyone,

I have a little naive question, just to understand a point I do not understand (but it does not have an impact on my analyses, it is just for understanding):

I have this basic code:

Epochs = mne.Epochs(RawData_Notched, events = Events_from_annot, tmin = Tmin, tmax = Tmax, baseline = None, preload = True, event_id = {“Stimulus/S 5”: 5, “Stimulus/S 10”: 10, “Stimulus/S 14”: 14})

Reject = get_rejection_threshold(Epochs)
Epochs.drop_bad(reject = Reject)
DroppedEpochs = [N for N, DL in enumerate(Epochs.drop_log) if len(DL)]
print(DroppedEpochs)

I noticed something I never cared about before. Indeed, when I epoched my data, I selected only specific events (5, 10, 14). In my task, the first 12 events are exemples for the patients (stim = 15) but in my code I do not considered them when I epoched the data (Epochs). However, when I print the epochs deleted by autoreject, it returns bad epochs but also the 12 first events (stim = 15) I thought I did not epoched. I do not understand what is doing autoreject to reject these epochs that should not appear (and that not appear when doing plot(Epochs))?

Thank you for yoiur help!

Hugues

2

I assume these are included in the drop log because they’re present in Events_from_annot. Check the drop log right after creating the epochs (even before running autoreject) to find out the specific drop reason; this should help you track this down.

Richard

1 Like