concatenate_epochs and subset of them

✉️ Mailing List Archive (read-only)
1

Dear mne users,
I?m dealing with a 3 blocks experiment and for each block I have a raw
object.
Then I?m epoching every raw object using the same event_id = [A, B, C, D].

Later I concatenate the 3 epochs objects with:
epochs_merged = mne.concatenate_epochs([epochs1, epochs2, epochs3])

First question, when concatenating, in the console I have the message:
*Applying baseline correction (mode: mean)* , but in the documentation I
can?t find any reference of it.
Since I already used baseline correction when creating epochs1, epochs2,
epochs3, could this additional correction generate any problem?

Later I?m doing some processing on epochs_merged (AutoReject, ICA). Is it
possible to access a subset of epochs_merged, for example just the epochs
with the event A, and average them?
Otherwise I have to re-run the analysis on a newly created epochs object
with event_id = A, but I would prefer to avoid this, since from my
understanding ICA performs better with more data.

Thank you in advance,

LA

MSc student, University of Padova
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20161215/89fafb02/attachment.html

2

Hi,

the first logging message is actually a bit misleading because it is only
storing the baseline parameters. So the baselining actually only happens
when the data is loaded (probably on concatenation in your case).

If you construct the epochs by passing a dict as the event_id (see for
example
http://mne-tools.github.io/stable/auto_tutorials/plot_object_epochs.html),
you can get a copy of the 'A' category by doing ``epochs['A']``.

-Jaakko

3

Hi Jaakko,
that was very clear thank you!
I saw that you have already commited a fix on github, awesome!

Lorenzo

4

Hi all,

Then I?m epoching every raw object using the same event_id = [A, B, C, D].

Later I concatenate the 3 epochs objects with:
epochs_merged = mne.concatenate_epochs([epochs1, epochs2, epochs3])

[...]

Later I?m doing some processing on epochs_merged (AutoReject, ICA). Is it
possible to access a subset of epochs_merged, for example just the epochs
with the event A, and average them?
Otherwise I have to re-run the analysis on a newly created epochs object
with event_id = A, but I would prefer to avoid this, since from my
understanding ICA performs better with more data.

I would be interested in an easy solution for this issue, too. Of
course one could extract these subsets of data manually, but is there
maybe an easier way I have simply overlooked so far?

Thanks,

    Richard

5

for a given epochs instance you have access to epochs.events and
epochs.selection attributes.
that should allow you to track what epoch correspond to what.

HTH
Alex

6

for a given epochs instance you have access to epochs.events and
epochs.selection attributes.
that should allow you to track what epoch correspond to what.

Thanks Alex! I realized my question had already been answered by Jaakko: