SSP remove ECG/EOG artifacts: Unable to generate artificial ECG channel

[Serendipityzzz]

Description of the problem

Hi there,

I tried to run the tutorial Repairing artifacts with SSP with my own data.

I have problems with using function create_ecg_epochs, compute_proj_ecg. I tried both EEG and MEG data, and the bug goes like: "Produces a ValueError: Unable to generate artificial ECG channel"

I want to run ssp before ica, since ica is not enough for me to repair ecg/eog artifacts, but I fail to compute the projections with SSP. Any tips or hints will be appreciated - Thx!

Wentao

Steps to reproduce

The first is what I do for EEG

base = os.path.join(os.getcwd(), os.pardir, os.pardir)
path_data = os.path.join(base, "data", "eeg", "demo","005", "20221223")
path_eeg = os.path.join(path_data, "eeg", "image-audio-post.vhdr")
path_montage = os.path.join(path_data, "standard-1020-cap64.locs")
path_output = os.path.join(os.getcwd(), "output")

# Execute preprocess pipeline.
data = preprocess_session(path_eeg, path_montage, path_output)
data = mne.io.read_raw_brainvision(path_eeg, preload=True)
# Annotate channel types according to config.
data = init_channel_types(data, channel_types=config.channel_types)
# Update the locations of channels for future plot.
data = init_channel_montage(data, path_montage)
## Preprocess data.
# Remove bad channels from data.
data = remove_bad_channels(data, bad_channels=config.bad_channels)
# Set reference channels of data.
data = set_ref_channels(data, ref_channels=config.ref_channels)
# Filter parts of signals' spectrum.
data = filter_spectrum(data, band_frange=config.band_frange, notch_freqs=config.notch_freqs)
# SSP
rojs_ecg, events_ecg  = mne.preprocessing.compute_proj_ecg(data1,n_eeg = 2,reject=None)

The Second is what I do for MEG
base = os.path.join(os.getcwd(), os.pardir, os.pardir)
path_data = os.path.join(base, "data", "meg", "demo","001", "20230214")
path_meg = os.path.join(path_data, "meg", "FL2_TEST1.fif")
path_output = os.path.join(os.getcwd(), "output")

# Execute preprocess pipeline.
data = mne.io.read_raw_fif(path_meg, preload=True)
ecg_evoked = mne.preprocessing.create_ecg_epochs(data).average()

Link to data

No response

Expected results

Created an SSP operator (subspace dimension = 3) 3 projection items activated Using data from preloaded Raw for 58 events and 61 original time points ... 0 bad epochs dropped Adding projection: planar--0.200-0.400-PCA-01 Adding projection: axial--0.200-0.400-PCA-01 Adding projection: eeg--0.200-0.400-PCA-01

Actual results

Traceback (most recent call last): File "F:\PKU2\Research2\Yunzhe_Liu\Code\eeg2label\preprocess\meg\session.py", line 71, in data = preprocess_session(path_meg, path_output) File "F:\PKU2\Research2\Yunzhe_Liu\Code\eeg2label\preprocess\meg\session.py", line 56, in preprocess_session ecg_evoked = mne.preprocessing.create_ecg_epochs(data).average() File "", line 12, in create_ecg_epochs File "E:\Anaconda3\envs\tensorflow\lib\site-packages\mne\preprocessing\ecg.py", line 346, in create_ecgepochs events, , _, ecg = find_ecg_events( File "", line 12, in find_ecg_events File "E:\Anaconda3\envs\tensorflow\lib\site-packages\mne\preprocessing\ecg.py", line 194, in find_ecgevents ecg, = _make_ecg(raw, start=None, stop=None) File "", line 12, in _make_ecg File "E:\Anaconda3\envs\tensorflow\lib\site-packages\mne\preprocessing\ecg.py", line 383, in _make_ecg raise ValueError('Unable to generate artificial ECG channel') ValueError: Unable to generate artificial ECG channel

Additional information

Platform: Windows-10-10.0.19044-SP0 Python: 3.9.13 (main, Oct 13 2022, 21:23:06) [MSC v.1916 64 bit (AMD64)] Executable: E:\Anaconda3\envs\tensorflow\python.exe CPU: AMD64 Family 23 Model 96 Stepping 1, AuthenticAMD: 12 cores Memory: 15.4 GB

mne: 1.3.0 numpy: 1.24.1 {MKL 2022.1-Product with 6 threads} scipy: 1.10.0 matplotlib: 3.6.3 {backend=QtAgg}

sklearn: 1.2.0 numba: Not found nibabel: 5.0.0 nilearn: 0.10.0 dipy: 1.6.0 openmeeg: Not found cupy: Not found pandas: 1.5.2 pyvista: 0.37.0 {OpenGL could not be initialized} pyvistaqt: 0.9.0 ipyvtklink: 0.2.3 vtk: Not found qtpy: 2.3.0 {PyQt5=5.15.2} ipympl: Not found pyqtgraph: 0.13.1 pooch: v1.6.0

mne_bids: Not found mne_nirs: Not found mne_features: Not found mne_qt_browser: 0.4.0 mne_connectivity: Not found mne_icalabel: 0.4

[welcome[bot]]

Hello! 👋 Thanks for opening your first issue here! ❤️ We will try to get back to you soon. 🚴🏽‍♂️

[Serendipityzzz]

For the code that gives ValueError, it's

[larsoner]

Artificial ECG requires MEG channels. We should update this error message to be clearer that this is the case

[Serendipityzzz]

Thx! I've solved the problem. And I wonder that after we calculate the ecg/eog projection with compute_proj_ecg/eog, what should we do to clean the raw data with these peojections. Maybe I should use raw.apply_proj() to rroject raw data onto a hyperplane orthogonal to noise?

which gose like “”“ projs_eog, events_eog = mne.preprocessing.compute_proj_eog(raw,ch_name="EOG",reject=None) raw.add_proj(projs_eog) raw.apply_proj() ”“”

Is it right?

[larsoner]

This is venturing into usage territory rather than code bug fix territory, please open a mne.discourse.group forum post for usage questions!