multiple empty room recordings

[harrisonritz]

We're working with OPM. Due to some sensor non-stationarity, right now we are collecting emptyroom recordings before (and even after) each subject.

Is there a way for me to have multiple emptyroom recordings per day, eg associate a specific time with the empty room recordings?

Right now I'm just saving the empty room in the subject's meg directory.
But it would be nice 1) match the standard and 2) link the data and emptyroom recordings.

[welcome]

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

[hoechenberger]

Hello, do you want to use only one empty-room recording per subject or more than one?
One per subject is already possible.

[harrisonritz]

Thanks for the quick reply!

One per subject if fine (eg I can concatenate recordings if needed).

I guess that I could manually create the emptyroom path. Set session using YYYYMMDD, and set run using HHMM. Then pass that bids path to write_raw_bids to associate the raw and the emptyroom?

The issue is that YYYYMMDD is enforced pretty strictly (eg the emptyroom session has to match meas_date), but think using run for time would get around that.

This hasn't actually been an issue yet, I'm just trying to set up a system that will work once our data collection ramps up.

[hoechenberger]

@harrisonritz, load both the empty-room and experimental data, and when calling write_raw_bids() for the experimental data, pass the empty-room data via the empty_room parameter.

See: https://mne.tools/mne-bids/stable/auto_examples/convert_mne_sample.html#convert-mne-sample-data-to-bids-format

raw = mne.io.read_raw(raw_fname)
raw_er = mne.io.read_raw(er_fname)

# specify power line frequency as required by BIDS
raw.info["line_freq"] = 60
raw_er.info["line_freq"] = 60

task = "audiovisual"
bids_path = BIDSPath(
    subject="01", session="01", task=task, run="1", datatype="meg", root=output_path
)
write_raw_bids(
    raw=raw,
    bids_path=bids_path,
    events=events_fname,
    event_id=event_id,
    empty_room=raw_er,
    overwrite=True,
)

[harrisonritz]

Unfortunately that doesn't work. The er labeling is based on YYYYMMDD, so it can't handle different er recordings on the same date.

If I modify meas_date in the sample dataset to generate two participants on the same day, each with a different er recordings, it overwrites participant 1's er recordings

import mne
import mne_bids
import datetime



sample_dir = mne.datasets.sample.data_path()
bids_root = sample_dir / 'bids'

raw = mne.io.read_raw(sample_dir / "MEG" / "sample" / "sample_audvis_filt-0-40_raw.fif")
raw_er = mne.io.read_raw(sample_dir / "MEG" / "sample" / "ernoise_raw.fif")

# specify power line frequency as required by BIDS
raw.info["line_freq"] = 60
raw_er.info["line_freq"] = 60



# create morning dataset
raw_morning = raw.copy().set_meas_date(datetime.datetime(2025, 1, 1, 9, 40, 10, 720100, tzinfo=datetime.timezone.utc))
raw_er_morning = raw_er.copy().set_meas_date(datetime.datetime(2025, 1, 1, 9, 00, 10, 720100, tzinfo=datetime.timezone.utc))

# participant 1
bids_path = mne_bids.BIDSPath(
    subject="01", session="01", task="sample", run="1", datatype="meg", root=bids_root
)

mne_bids.write_raw_bids(
    raw=raw_morning,
    bids_path=bids_path,
    empty_room=raw_er_morning,
    overwrite=True,
)


# create afternoon dataset
raw_afternoon = raw.copy().set_meas_date(datetime.datetime(2025, 1, 1, 13, 40, 10, 720100, tzinfo=datetime.timezone.utc))
raw_er_afternoon = raw_er.copy().set_meas_date(datetime.datetime(2025, 1, 1, 13, 00, 10, 720100, tzinfo=datetime.timezone.utc))

# participant 1
bids_path = mne_bids.BIDSPath(
    subject="02", session="01", task="sample", run="1", datatype="meg", root=bids_root
)

mne_bids.write_raw_bids(
    raw=raw_afternoon,
    bids_path=bids_path,
    empty_room=raw_er_afternoon,
    overwrite=True,
)

mne_bids.print_dir_tree(bids_root, max_depth=4)

note that there is only one er recording:

|bids/
|--- README
|--- dataset_description.json
|--- participants.json
|--- participants.tsv
|--- sub-01/
|------ ses-01/
|--------- sub-01_ses-01_scans.tsv
|--------- meg/
|------------ sub-01_ses-01_coordsystem.json
|------------ sub-01_ses-01_task-sample_run-1_channels.tsv
|------------ sub-01_ses-01_task-sample_run-1_meg.fif
|------------ sub-01_ses-01_task-sample_run-1_meg.json
|--- sub-02/
|------ ses-01/
|--------- sub-02_ses-01_scans.tsv
|--------- meg/
|------------ sub-02_ses-01_coordsystem.json
|------------ sub-02_ses-01_task-sample_run-1_channels.tsv
|------------ sub-02_ses-01_task-sample_run-1_meg.fif
|------------ sub-02_ses-01_task-sample_run-1_meg.json
|--- sub-emptyroom/
|------ ses-20250101/
|--------- sub-emptyroom_ses-20250101_scans.tsv
|--------- meg/
|------------ sub-emptyroom_ses-20250101_task-noise_channels.tsv
|------------ sub-emptyroom_ses-20250101_task-noise_meg.fif
|------------ sub-emptyroom_ses-20250101_task-noise_meg.json

Do you think my best bet is manually creating er BIDS files with run reflecting the time, and then passing that BIDSPath to write_raw_bids?

[drammock]

Does it work to save the empty room separately as a task-noise file in the subject's meg directory and naming it in the JSON sidecar (as AssociatedEmptyRoom)? IIRC that will avoid putting the ER file under sub-emptyroom so the date-based filename clash / overwrite shouldn't happen. (that assumes you're talking about different ERs for different subjects recorded on the same day --- i.e., not different ERs for different runs of the same subject). cf #1364.

[harrisonritz]

Saving in the same folder is what I have now, and I was concerned that I was out of spec.
I'll manually add to AssociatedEmptyRoom (eg passing ER path in write_raw_bids).
This resolves my question, thank you @drammock!

[drammock]

great! Heads up though that #1364 is only on main, not in a released version yet.

[harrisonritz]

gotcha, thanks. I'll be sure to use AssociatedEmptyRoom rather than relying on auto-matching

[hoechenberger]

Interesting, I actually thought passing empty_room would store it in the subject's directory. We should definitely support this. @sappelhoff WDYT?
In fact, I'd suggest this should become our default behavior…

[sappelhoff]

I think we should support both strategies supported in the BIDS spec and make it easy for users to adopt either of them. And given that I personally don't work with this type of data, it'd be great to get the input from someone "in the field".