- MNE version:1.7,1
- operating system: Windows 11
Hello, I am trying to build a BCI based on event-related potentials. My stimulation creates very close events, approx 150ms apart, and I would like to do real-time classification. However, when I epoch each segment of data to predict based on those events, one epoch gets dropped because “TOO_SHORT”.
Now I currently setup my epoching in this way
epoch_start = -0.2
epoch_end = 0.5
baseline=(epoch_start, 0)
eps = extract_epochs(data=filtered, events=events, ev_ids=ev_ids, tmin=epoch_start, tmax=epoch_end, baseline=baseline)
eps.plot_drop_log()
What I would like to have are fixed length epochs of 0.7 seconds, that consequently overlap, so that I can then pick the one with the highest probability of being the correct ERP.
If I increase the length of my EEG segment, I am able to create all the epochs but it occasionally includes one epoch from the next prediction.
Is there a way so that the epoch won’t be dropped automatically? I tried to set reject=None but it does not seem to prevent this issue from happening.
Thanks in advance to whoever can provide advice.