Open multimodal iEEG-fMRI dataset, naturalistic film

Description

This is the “Open multimodal iEEG-fMRI dataset from naturalistic stimulation with a short audiovisual film” released on OpenNeuro as ds003688 (Berezutskaya et al., 2022). It was collected at the University Medical Center (UMC) Utrecht in the lab of Nick F. Ramsey, with data from epilepsy surgery patients undergoing clinical intracranial monitoring. Participants viewed a short audiovisual film while neural activity was recorded; the purpose of the deposit is to support reproducible, naturalistic-stimulation research into multimodal perception and language comprehension, and to compare signals acquired during the same task across two recording modalities — intracranial EEG (iEEG) and functional MRI (fMRI) (Berezutskaya et al., 2022). The cohort is mixed pediatric and adult.

The release is genuinely multimodal: the iEEG arm comprises 51 participants (ages 5—55) and the fMRI arm 30 participants (ages 7—47), with 18 participants performing both versions of the task non-simultaneously (typically weeks apart). The union spans the 63 subject folders (sub-01 … sub-63) listed in participants.tsv and on the OpenNeuro landing page.

Citation

• Paper DOI: 10.1038/s41597-022-01173-0
• Dataset DOI: 10.18112/openneuro.ds003688.v1.0.7
• Reference: Berezutskaya, J., Vansteensel, M. J., Aarnoutse, E. J., Freudenburg, Z. V., Piantoni, G., Branco, M. P., & Ramsey, N. F. (2022). Open multimodal iEEG-fMRI dataset from naturalistic stimulation with a short audiovisual film. Scientific Data 9, 91.

Catalog citation block

Berezutskaya, J., Vansteensel, M. J., Aarnoutse, E. J., Freudenburg, Z. V., Piantoni, G., Branco, M. P., & Ramsey, N. F. (2022). Open multimodal iEEG-fMRI dataset from naturalistic stimulation with a short audiovisual film. Scientific Data 9:91. doi: 10.1038/s41597-022-01173-0.

OpenNeuro deposit: ds003688, DOI 10.18112/openneuro.ds003688.v1.0.7 (CC0) — https://openneuro.org/datasets/ds003688. Institution: UMC Utrecht (Ramsey lab).

Download and access

Field	Value
Official source	https://openneuro.org/datasets/ds003688
Access class	open
Size	TODO(confirm)
Format	brainvision
BIDS	true
Mirrors	NEMAR ds003688

Source & access

• Repository: OpenNeuro, accession ds003688 — https://openneuro.org/datasets/ds003688.
• Dataset DOI: 10.18112/openneuro.ds003688.v1.0.7.
• License: CC0 (public-domain dedication), per the OpenNeuro dataset_description.json (note: the data-descriptor paper itself is CC-BY 4.0 — a distinct license that applies to the article, not the deposited data).
• Data-descriptor paper: Berezutskaya et al. (2022), Scientific Data 9:91, doi 10.1038/s41597-022-01173-0 — https://www.nature.com/articles/s41597-022-01173-0 (open-access full text at PMC8938409).
• Also mirrored on NeMAR (https://nemar.org/dataexplorer/detail?dataset_id=ds003688).

Data structure

The deposit is organized in BIDS v1.2.1 (with iEEG-BIDS / iBIDS extensions). iEEG signals are stored in BrainVision format (.eeg / .vhdr / .vmrk) with JSON/TSV sidecars; fMRI is NIfTI + JSON. Channel sets are per-subject (mixed ECoG/SEEG) with no single fixed montage — as a representative composition, sub-01 carries 103 SEEG + 1 EOG + 1 ECG + 3 EMG + 1 misc + 2 trigger channels. The release ships electrode coordinate systems, electrode-position TSVs, and rendered electrode-placement photos (*acq-render_photo_seeg_*.jpg). Auxiliary physiology (ECG, EMG, EOG, respiration) is present for some participants. OpenNeuro serves the data via DataLad/git-annex (S3-backed).

License

• Name: CC0
• Note (verbatim from the directory): CC0 per the OpenNeuro dataset_description.json; the data-descriptor article is CC BY 4.0, which applies to the article, not the data
• Snippets on this site: allowed — the site may ship short derived snippets and precomputed products from this dataset, labelled as derivatives (re-referenced, filtered, cropped) with citation, license and DOI.

Acquisition

Field	Value
Device	128-channel Micromed clinical iEEG system (ECoG grids, HD-ECoG, sEEG); Blackrock for HD-ECoG channels in 2 patients
Device class	intracranial
Channels	TODO(confirm) (per-subject channel sets (mixed ECoG/SEEG); no single fixed montage; value from the catalog prose (registry silent))
Sampling rate (Hz)	512 (registry value: 512-2048)
Online filters	hardware band-pass 0.15–134.4 Hz (512 Hz patients) or 0.3–500 Hz (2048 Hz patients); no notch
Reference	external mastoid reference electrode
Mains frequency (Hz)	50
Paradigms	film, rest
Participants	63 (63 in participants.tsv is the union of iEEG (51) and fMRI (30) cohorts; 12 are fMRI-only)
Population	63 epilepsy-surgery patients 5–55: iEEG 51, fMRI 30, both 18; ECoG grids 46, HD-ECoG 6, sEEG 16 (overlapping)
Clinical groups	epilepsy
Sessions	1
Durations	6.5-min film excerpt in 13 × 30-s speech/music blocks; task-rest ~3 min (26 patients) plus natural rest from clinical monitoring (24 patients)

Acquisition

iEEG was recorded with a 128-channel Micromed system (Treviso, Italy). The sampling rate is not uniform: most patients (38) were sampled at 512 Hz (online band-pass 0.15—134.4 Hz), while 13 patients were sampled at 2048 Hz (online band-pass 0.3—500 Hz) (Berezutskaya et al., 2022). In two HD-ECoG participants, the HD channels were additionally recorded at 2000 Hz (0.3—500 Hz) on a Blackrock system simultaneously with the clinical Micromed channels. Thus a hardware band-pass was applied at recording; what is absent is a power-line notch (the Netherlands line frequency is 50 Hz). The original recording reference was “an external reference electrode typically placed on the mastoid part of the temporal bone.”

The fMRI arm used a Philips Achieva 3 T scanner with a 3D-PRESTO sequence (TR/TE = 22.5/33.2 ms, 608 ms/volume, FA 10°, 40 slices, 4 mm isotropic). Per the BIDS layout there is also evidence of a 7 T session label for some subjects in addition to the 3 T fMRI.

Participants

The participants.tsv lists 63 subjects (sub-01…sub-63), the union of the two modality cohorts (Berezutskaya et al., 2022):

• iEEG: 51 participants, age range 5—55 years (mean 25 ± 15 SD), 32 female. All are epilepsy surgery candidates monitored intracranially for clinical reasons, so coverage is dictated by clinical need rather than experimental design.
• fMRI: 30 participants, age range 7—47 years (mean 22 ± 11 SD), 14 female.
• 18 participants did both modalities (non-simultaneously).

Implant types among the iEEG cohort (per the paper): 46 patients with standard subdural ECoG grids (2.3 mm exposed diameter, 10 mm spacing, 48—128 contacts); 6 patients additionally received high-density (HD) ECoG grids (1.3 mm diameter, 3—4 mm spacing); and 16 patients had stereo-EEG (sEEG) depth electrodes (4—173 contacts) — categories overlap because some patients carried both grids and depths. Hemispheric coverage is approximately 45 left, 9 right, 3 bilateral.

Tasks / conditions

The naturalistic stimulus is a 6.5-minute excerpt of the film “Pippi on the Run” (the Swedish original, Dutch-dubbed), structured as 13 blocks of 30 s each — 6 speech blocks and 7 music blocks (Berezutskaya et al., 2022). Detailed speech and video annotations of the stimulus are provided with the release. In addition to the film task there is a rest condition: a short task-rest (~3 min, 26 patients) plus natural rest extracted from the continuous 24/7 clinical monitoring (24 patients). On disk the clinical acquisition carries the tag acq-clinical; a separate acq-render provides rendered electrode-placement photos.

Used on this site

L0.1, L7.7

Loader

Fetches only the stated subset. Recorded loader: DataLad (OpenNeuro ds003688).

# OpenNeuro serves ds003688 as a DataLad/git-annex repository; fetch one subject only.
datalad install https://github.com/OpenNeuroDatasets/ds003688.git
cd ds003688
datalad get sub-001   # replace with the subject id you need

Caveats worth knowing

• 38 patients at 512 Hz and 13 at 2048 Hz.
• A hardware band-pass was applied; a notch was not.

Notable caveats

• Sampling rate is mixed, not 2048 Hz uniformly. 38 patients are at 512 Hz, only 13 at 2048 Hz, and 2 HD-ECoG patients at 2000 Hz (Blackrock).
• A hardware band-pass was applied online (0.15—134.4 Hz or 0.3—500 Hz).
• Re-reference required. Recording reference was an external mastoid electrode; re-reference (common-average) separately for HD-ECoG and sEEG.
• 63 vs 51. The 63 entries in participants.tsv are the union of iEEG (51) and fMRI (30) cohorts.
• Coverage is clinically determined and predominantly left-hemisphere; montage and contact count differ subject-to-subject.

Related datasets

• HUP iEEG Epilepsy Dataset

# Loader recorded in the catalog: DataLad (OpenNeuro ds003688)
# TODO(confirm) the exact call and the subset to fetch. Official source: https://openneuro.org/datasets/ds003688