search

mne-tools / mne-python

MNE: Magnetoencephalography (MEG) and Electroencephalography (EEG) in Python
https://mne.tools
BSD 3-Clause "New" or "Revised" License
2.68k stars 1.31k forks source link

[Question] Adding ECG channel to existing EEG raw object? #3816

Closed ppasler closed 7 years ago

ppasler commented 7 years ago

Hi there,

I recorded 14 EEG and 1 ECG channel with different devices. The sample rate for the EEG is roughly 128 and for the ECG 1000Hz. I tried to use add_channels with the EEG raw object but get an error 'All dimensions except channels must match'. So I resampled the ECG data to 128Hz, but the data differ in 2 rows. So I wondered if there is an automatic way to merge two raw objects? Did I miss some helper method or a tutorial for doing this?

By the way, I don't get why add_list in add_channels must be a list, as I am only adding on object.

Thanks in advance, paul

jaeilepp commented 7 years ago

So I resampled the ECG data to 128Hz, but the data differ in 2 rows.

I'm not sure what you mean.

So I wondered if there is an automatic way to merge two raw objects? Did I miss some helper method or a tutorial for doing this? By the way, I don't get why add_list in add_channels must be a list, as I am only adding on object.

Passing a list simply allows adding multiple objects at once. You can always do raw.add_channels([raw2]).

ppasler commented 7 years ago

@jaeilepp Even with resampling to the same sfreq the _data array has not the same length (they differ in length of 2), which leads to an error. My "workaround" was the following:

  1. Create the EEG raw object
  2. Create the ECG raw object
  3. Resample rawECG to the sFreq of rawEEG
  4. Crop rawECG to t_max of rawEEG where t_max = (eegRaw.n_times - 1) / eegRaw.info['sfreq']

Assuming EEG and ECG start at the same timestamp.

This works, but seems a bit awkward. If I didn't miss a better option, I'll close the issue for now.

jaeilepp commented 7 years ago

To me this sounds more like a synchronising issue than an issue with MNE.

ppasler commented 7 years ago

I'm not sure about that. I have two signals with the same time range but different sample rates, I assumed this was not a big deal - but anyhow I found a solution. I don't know how others record signals with different devices? Do you have any tips how to perform better?

jaeilepp commented 7 years ago

Yes. When you resample it is not always obvious which samples to include (especially when the sampling frequencies don't align well). For example, see discussions at https://github.com/mne-tools/mne-python/issues/2787 and https://github.com/mne-tools/mne-python/pull/2657

ppasler commented 7 years ago

I see, to me as noob it was not clear how this works. Resampling and cropping solved it for me, I am not sure how often people do this my way. Maybe it's worth a tutorial - I don't know.