Is mne.make_fixed_length_events working properly?

đź’¬ Support & Discussions
1
  • MNE version: e.g. 0.24.0
  • operating system: Windows 10

Dear all,

I was testing “mne.make_fixed_length_events” on the database “sample_audvis_filt-0-40_raw.fif”.

“eventsTest_raw” is the object that I’ve reduced to only EEG data, and cropped from second 4 to second 14.


new_events = mne.make_fixed_length_events(eventsTest_raw, start=1, stop=9, duration=1., overlap=0.5)  
mne.viz.plot_events(new_events, sfreq=eventsTest_raw.info['sfreq']) # plot events

image
image642Ă—546 10.5 KB

Looking at the plot represented above in figure something weird happened.
First because, as anticipated and showed by the below figure, data have been cropped, and should have a total duration of 10 seconds.

image
image1425Ă—683 196 KB

I do not understand why the events do not start from second 1, but around second 48.
In addition, it is not clear because now I have only 10 seconds of recording, it does not even reach such a long duration.
I understand that the issue is in “new_events”, as shown below (sample rate was around 150Hz)

image

What do you think about it? Thanks in advance

2

does https://mne.tools/stable/glossary.html#term-first_samp help?

Alex

1 Like
3

Thank you, setting “first_samp=False”, the positions of the events is now fine, as you can see in the upper part of the screenshot.
Neverthelss, while plotting the epochs I get a blank figure (bottom part of the screenshot) and the following error:

"Not setting metadata
Not setting metadata
10 matching events found
Applying baseline correction (mode: mean)
0 projection items activated
Loading data for 10 events and 151 original time points ...
10 bad epochs dropped"
....
....
    xlim = epoch_times[-1] + len(orig_epoch_times)
IndexError: index -1 is out of bounds for axis 0 with size 0

Here the code:

eventsTest_raw = raw_filtered.copy()
event_id = {'My Event': 1}
new_events = mne.make_fixed_length_events(eventsTest_raw, start=0, duration=1., overlap=0., first_samp=False) # overlap=0.5
mne.viz.plot_events(new_events, sfreq=eventsTest_raw.info['sfreq']) # plot events
epochs = mne.Epochs(eventsTest_raw, new_events, event_id, tmin=0, tmax=1,baseline=(0, 0), preload=True)  
mne.viz.plot_epochs(epochs) # plot epochs

image
image1517Ă—962 22.3 KB

So, at the moment, I’m not able to set epochs.
I’ve also read this post where you @agramfort answered on a similar topic because I was searching for a way to reset the “first samp” that maybe would have fixed the issue on the epochs creation, but I did not understand how to use " RawArray" for this…
What should I do? Thanks in advance

4

Just to summarize, I’d simply need to:

  • Work on 10 seconds interval of the sample EEG recording (I did it using “crop”)
  • Create the following epochs: (0-1 s | 1-2 s | 2-3 s … 9-10 s)
    thanks
5

After running this line, could you please run:

epochs.plot_drop_log()

This should indicate why the epochs where actually dropped.

Thanks!

Edit

I believe you shouldn’t pass first_samp=False when creating the events. Instead, you should only pass this parameter during plotting.

For creating and visualizing the epochs, you would do:

new_events = mne.make_fixed_length_events(eventsTest_raw, start=0, duration=1., overlap=0.) # overlap=0.5
epochs = mne.Epochs(eventsTest_raw, new_events, event_id, tmin=0, tmax=1,baseline=(0, 0), preload=True)
epochs.plot()
1 Like
6

Thanks @richard and @agramfort .

I’m agree that “first_samp=False” generates the error. I was a bit confused while plotting the events using

mne.viz.plot_events(new_events, sfreq=eventsTest_raw.info['sfreq'])

because the event plot showed a timeline outside the cropped data.
However, when plotting the epochs everything is fine. Thanks