Use `fwd["info"]["dev_head_t"] when raw.info["dev_head_t"] is None?

[OleBialas]

Describe the new feature or enhancement

I tried to simulate EEG data using a forward model:

raw_fname = mne.datasets.eegbci.load_data(subjects=1, runs=[6])[0]
raw = mne.io.read_raw_edf(raw_fname, preload=True)
montage = mne.channels.make_standard_montage("standard_1005")
mne.datasets.eegbci.standardize(raw)
raw.set_montage(montage);
fs_dir = mne.datasets.fetch_fsaverage(verbose=True)
src = fs_dir / "bem" / "fsaverage-ico-5-src.fif"
bem = fs_dir / "bem" / "fsaverage-5120-5120-5120-bem-sol.fif"
fwd = mne.make_forward_solution(
    raw.info, trans="fsaverage", src=src, bem=bem, eeg=True, mindist=5.0, n_jobs=None
)

labels = mne.read_labels_from_annot('fsaverage', parc='aparc', subjects_dir=fs_dir.parent)
label_names = ["transversetemporal-lh", "transversetemporal-rh"]
labels = [l for l in labels if l.name in label_names]

times = np.arange(0, 1, 1/raw.info["sfreq"])
stc = mne.simulation.simulate_sparse_stc(
    fwd["src"],
    n_dipoles=2,
    times=times,
    labels=labels
)
n_epochs = 10
raw_sim = mne.simulation.simulate_raw(raw.info, [stc] * n_epochs, forward=fwd, verbose=False)

However, this gave me the error:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[32], line 1
----> 1 raw_sim = mne.simulation.simulate_raw(raw.info, [stc] * n_epochs, forward=fwd, verbose=False)

File <decorator-gen-429>:10, in simulate_raw(info, stc, trans, src, bem, head_pos, mindist, interp, n_jobs, use_cps, forward, first_samp, max_iter, verbose)

File ~/projects/iBOTS-Intro-to-EEG-MEG-Analysis-with-MNE-Python/.pixi/envs/default/lib/python3.13/site-packages/mne/simulation/raw.py:295, in simulate_raw(info, stc, trans, src, bem, head_pos, mindist, interp, n_jobs, use_cps, forward, first_samp, max_iter, verbose)
    288 if any(x is not None for x in (trans, src, bem, head_pos)):
    289     raise ValueError(
    290         "If forward is not None then trans, src, bem, "
    291         "and head_pos must all be None"
    292     )
    293 if not np.allclose(
    294     forward["info"]["dev_head_t"]["trans"],
--> 295     info["dev_head_t"]["trans"],
    296     atol=1e-6,
    297 ):
    298     raise ValueError(
    299         "The forward meg<->head transform "
    300         'forward["info"]["dev_head_t"] does not match '
    301         'the one in raw.info["dev_head_t"]'
    302     )
    303 src = forward["src"]

TypeError: 'NoneType' object is not subscriptable

I could fix it by copying the "dev_head_t" from fwd:

raw.info["dev_head_t"] = fwd["info"]["dev_head_t"]

However, this feels kind of hacky.
Is there a reason why copying the transform from fwd is not the default behavior when raw.info["dev_head_t"] is None?

Describe your proposed implementation

In simulate raw, copy fwd["info"]["dev_head_t"] when raw.info["dev_head_t"] is None.

Describe possible alternatives

If there are reasons why copying the transform is a bad idea, at least print a more explicit error message that mentions the missing transform in raw.info

Additional context

No response

[larsoner]

In general, assuming fwd["info"]["dev_head_t"] is correct when raw.info["dev_head_t"] is None would be dangerous I think. Better for a user to explicitly set it. But for your particular case, if this fails:

raw_fname = mne.datasets.eegbci.load_data(subjects=1, runs=[6])[0]
raw = mne.io.read_raw_edf(raw_fname, preload=True)
assert raw.info["dev_head_t"] is None

then there is something bad happening with read_raw_edf (it only ever reads EEG data so dev_head_t should be None, and hopefully was done in #13112). And if that's the case then you should also have

assert fwd["info"]["dev_head_t"] is None

Which of these fails?

[OleBialas]

Thanks for the quick response!
I ran this code

import mne

raw_fname = mne.datasets.eegbci.load_data(subjects=1, runs=[6])[0]
raw = mne.io.read_raw_edf(raw_fname, preload=True)
montage = mne.channels.make_standard_montage("standard_1005")
mne.datasets.eegbci.standardize(raw)
raw.set_montage(montage)
fs_dir = mne.datasets.fetch_fsaverage(verbose=True)
src = fs_dir / "bem" / "fsaverage-ico-5-src.fif"
bem = fs_dir / "bem" / "fsaverage-5120-5120-5120-bem-sol.fif"
fwd = mne.make_forward_solution(
    raw.info,
    trans="fsaverage",
    src=src,
    bem=bem,
    meg=False,
    eeg=True,
    mindist=5.0,
    n_jobs=None,
)
assert raw.info["dev_head_t"] is None
assert fwd["info"]["dev_head_t"] is None

and got an error on the second assertion

    assert fwd["info"]["dev_head_t"] is None
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError

Using MNE 1.11.0

[larsoner]

Okay, good to know... I suspect this is a bug with our forward code that it adds a non-None dev_head_t

[drammock]

trans="fsaverage"

isn't this the culprit?

[larsoner]

That should set the head<->MRI (head_mri_t) trans, not the MEG<->head trans (dev_head_t)