[Mne_analysis] Error with load epochs after cropping after baseline correction

[Alexandre Gramfort]

        External Email - Use Caution        

hi Sebastian,

your hack should not cause any problems down the road but I could
consider this a bug in mne-python.

Can you open an issue on github so we can fix?

Alex


On Sat, Apr 18, 2020 at 5:08 PM S.P.H. Speer <speer at rsm.nl> wrote:

>         External Email - Use Caution
>
> Dear MNE team,
>
>
>
> I’ve recently started preprocessing some EEG data with the MNE toolbox and
> it has been a great experience so far.
>
>
>
> I did my baseline correction on a period of 1s which is 6s prior to the
> onset of the event of interest (due to unfortunate experimental design), so
> I initially epoched my data to -7 to 3s . After epoching the data I then
> cropped the data to -0.3 to 3s, as I no longer needed the 7 seconds prior
> to the data after the baseline correction was done. When I then save the
> epoched data and load it again somewhere else and error occurs because the
> baseline parameters are outside the epoched data.
>
> To my understanding the baseline correction is already applied during
> epoching so I the decided to just set the baseline parameters to (0,0)
> after baseline correction has been done and data has been cropped. If I do
> that I can load the epoched data without no problem. My question is now:
> Will I run into problem later in my analysis pipeline because I set the
> baseline parameters to (0,0) even though baseline correction was done on
> another time window?
>
> Is there a better way to solve this issue?
>
>
>
> See below the code:
>
>
>
>     decision_epochs = mne.Epochs(eeg_data_interp_ref_ica, events, tmin=-7,
> tmax=3,baseline=(-7,-6),
>
>                         #reject=reject_criteria,  flat=flat_criteria, #
> commented beause we're using autoreject here
>
>                         event_id = decision_events_of_interest,
>
>                         reject_by_annotation=False, preload=True)
>
>
>
>     decision_epochs_cropped = decision_epochs.copy().crop(tmin=-0.3,
> tmax=3, include_tmax=True)
>
>
>
>
>
>     # drop epochs after cropping
>
>
>
>     ar = AutoReject(verbose='tqdm_notebook', n_jobs=n_jobs,
> consensus=np.linspace(0, 1, 20), n_interpolate = np.array([1,4,8,16,32]) )
>
>     decision_epochs_cropped_clean_ar =
> ar.fit_transform(decision_epochs_cropped)
>
>
>
>     out_dir='/data/sebastian/EEG/neural_analysis/'
>
>
>
>     # This version can't be loaded after
>
>
> decision_epochs_cropped_clean_ar.save(opj(out_dir,task,sub+'-TEST_'+str(cutoff_l)+'_HP_'+str(cutoff_h)+'_'+reference+'.fif'),
> overwrite=True)
>
>
>
>     # This version can't be loaded after
>
>     decision_epochs_cropped_clean_ar.baseline=(0,0)
>
>
> decision_epochs_cropped_clean_ar.save(opj(out_dir,task,sub+'-TEST2_'+str(cutoff_l)+'_HP_'+str(cutoff_h)+'_'+reference+'.fif'),
> overwrite=True)
>
>
>
>
>
> Thank you very much for your help,
>
>
>
> Sebastian
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200418/0013c460/attachment-0001.html