Minor Priority: Issue with converting MNE Raw into BIDS with mne_bids - Age/Birthday Info of Subject

Support & Discussions
1

Dear All,

I found a minor issue in the Patient/Subject Information raw.info[‘subject_info’] which is potentially new after some recent changes (adding weight and height).
I used some of my old code parts to convert eeg data to bids using mne_bids (which worked perfectly with the previous versions).

As before I added the age of the participants using the birthday key for the subject_info dict before converting/writing the object into bids:
raw.info[‘subject_info’][‘birthday’] = (year, month, day) in integer (int)

Unfortunately, it was not recognized after the conversion (no info (n/a) in the participants.tsv file).

I looked into some documentation on the subject_info and found this interesting format specification in the dictionary documentation :sweat_smile: - could it be a copy-paste mistake which causes the issue?

subj_birth_day 404 julian - “Birthday of the subject”

Additional Info

  • MNE version: ‘1.1.1’
  • mne_bids version: ‘0.11.1’
  • operating system: Windows 10

Many thanks for looking into it :hugs:
Best,
Katharina

2

Hello @kathlin42, I believe the date must be a datetime.datetime object, could you try with that? (I cannot check myself right now!)

Best wishes,
Richard

3

Hi Richard,

Thanks for the first idea.
I tested it with datetime.datetime object but it is still not working properly. It is correctly added in the subject_info dict but gets lost after the conversion to bids (tsv file and raw object after read_raw_bids). Of course, the issue could also be in mne_bids since they also changed some things in the last months…

Best
Katharina

4

I cannot reproduce this…

# %%
from pathlib import Path
import pandas as pd
import mne
import mne_bids

sample_dir = mne.datasets.sample.data_path()
sample_fname = sample_dir / 'MEG' / 'sample' / 'sample_audvis_raw.fif'

raw = mne.io.read_raw_fif(sample_fname)
subject_info = {
    'birthday': (1980, 12, 30)  # YYYY, MM, DD
}
raw.info['subject_info'] = subject_info

bids_root = Path('/tmp/bids-test')
bp = mne_bids.BIDSPath(
    subject='sample', task='av', suffix='meg', extension='.fif',
    datatype='meg', root=bids_root
)
mne_bids.write_raw_bids(raw, bids_path=bp, overwrite=True)

# %%
participants_tsv_path = bids_root / 'participants.tsv'
participants_tsv = pd.read_csv(participants_tsv_path, sep='\t')
participants_tsv

produces:

Screenshot 2022-11-14 at 19.40.27
Screenshot 2022-11-14 at 19.40.27828×142 11.3 KB

The only problem I noticed is that we don’t support the birthday to be a datetime object, so I passed a tuple instead …

Can you share the code you’re using?

Richard

cc @sappelhoff

5

Hi Richard,

Of course:
Maybe it is an EEG specific issue? BrainVision format issue? I also get a warning for the Coordinate frame but I considered this independent from the subject_info…

CODE

# Import necessary Packages
import os
import mne_bids
import mne
import pickle
# Project Specifications
task = 'errormonitoring'
datatype = 'eeg'
suffix = 'eeg'
extension = '.vhdr'
data_format= 'BrainVision'

mne_events_id = {'Fixation': 0,
                 'Good': 100,
                 'Error': 200,
                 'Response_Cue': 2,
                 'Key_Press': 3,
                 'ITI': 4}

# Data Paths
bids_root = 'raw_data' # Path where to save BIDS files
source_root = 'source_data' # Path where to copy 'example_subj_raw.fif' and 'mne_events.pickle'

# Read Data
raw = mne.io.read_raw(os.path.join(source_root, 'example_subj_raw.fif'))

with open(os.path.join(source_root, 'mne_events.pickle'), 'rb') as file:
    # A new file will be created
    mne_events = pickle.load(file)

# Add Information for BIDS Conversion
raw.info['line_freq'] = 50
raw.info['subject_info'] = {}
raw.info['subject_info']['first_name'] = 'sub-001'
raw.info['subject_info']['his_id'] = 'sub-001'
raw.info['subject_info']['birthday'] = (1996, 10, 1)
raw.info['subject_info']['sex'] = 1
raw.info['subject_info']['hand'] = 1
montage = mne.channels.make_standard_montage('standard_1020')
print(montage)
raw.set_montage(montage)
# Write Run Task Data to BIDS
bids_path = mne_bids.BIDSPath(subject='001', task='errormonitoring', run = 1,root=bids_root, datatype='eeg', suffix='eeg',
                     extension='.vhdr')
mne_bids.write_raw_bids(raw = raw, bids_path=bids_path, events=mne_events, event_id=mne_events_id,
                        allow_preload=True, overwrite=True, format='BrainVision', montage = montage)

# Test Result
raw_bids = mne_bids.read_raw_bids(bids_path, extra_params=None, verbose=None)
print(raw_bids.info['subject_info'])

If it helps, I can also share the raw fif file and mne_events in pickle format…

Thanks a lot for your help!

6

If you could share the input file(s), that would be helpful!

Btw you shouldn’t use pickle for long-term storage, it will cause problems sooner or later.

7

You should be able to download the files via the owncloud Link: Fraunhofer ownCloud
I didn’t find an option to upload the file types directly in discourse :wink:
If there are any problems I can try another way to share them.

8

The problem was that the measurement date was not set, so even though the birthday was present, the participant’s age could not be calculated.

MNE-BIDS should probably warn or even raise an exception in a case like this!

The following code works for me:

import datetime
import mne_bids
import mne

bids_root = '/tmp/bids-mwe'

raw = mne.io.read_raw('example_subj_raw.fif')
raw.info['subject_info'] = {}
raw.info['subject_info']['birthday'] = (1996, 10, 1)

# meas_date must be set so participant age can be calculated
raw.set_meas_date(
    datetime.datetime(2017, 1, 30, tzinfo=datetime.timezone.utc)
)

bids_path = mne_bids.BIDSPath(
    subject='001',
    task='errormonitoring',
    root=bids_root,
    datatype='eeg',
    suffix='eeg',
    extension='.vhdr'
)
mne_bids.write_raw_bids(raw=raw, bids_path=bids_path, overwrite=True)
2 Likes
9

Thanks a lot!!! Now it worked - I’m sorry that it had such a trivial reason!