EEGdash›NeMAR›NM000452

Iss. 452 · 1 subjects · 1 recordings · CC-BY-4.0

Dataset Brief · ECoG_HB_02

NM000452: ieeg dataset, 1 subjects#

ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative)

Access recordings and metadata through EEGDash.

Citation: Xupeng Chen (2024). ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative). 10.82901/nemar.nm000452

Modality: ieeg Subjects: 1 Recordings: 1 License: CC-BY-4.0 Source: nemar

Metadata: Complete (100%)

1-participant iEEG dataset — ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative).

Data & curation Xupeng Chen
Senior author Xupeng Chen
Year 2024 · Distributed via NeMAR
Funding U.S. National Science Foundation IIS-1912286 · U.S. National Science Foundation 2309057 (paper acknowledgements) · + 3 more

iEEG · 225 ch125 HzBIDS 1.10.0Task · speech

Layer 01Study

What was asked

Hypothesis, independent & dependent variables, paradigm, cohort, and the editorial caveats around what the recordings can and cannot answer.

Layer 02Signal · BIDS

What was recorded

Sidecars, channels & electrodes, coordinate system, event semantics, and quality stats from the NEMAR pipeline when available.

Layer 03Training · ML

What you can train on

Recommended access modes — MNE Raw, braindecode windows, PyTorch DataLoader — plus the targets the metadata makes addressable.

§ 01Access · Get started

Quickstart#

Get Started

Install

pip install eegdash

Access the data

from eegdash.dataset import NM000452

dataset = NM000452(cache_dir="./data")
# Get the raw object of the first recording
raw = dataset.datasets[0].raw
print(raw.info)

Query & Filter

Filter by subject

dataset = NM000452(cache_dir="./data", subject="01")

Advanced query

dataset = NM000452(
    cache_dir="./data",
    query={"subject": {"$in": ["01", "02"]}},
)

Iterate recordings

for rec in dataset:
    print(rec.subject, rec.raw.info['sfreq'])

Cite This Dataset

If you use this dataset in your research, please cite the original authors.

BibTeX

@dataset{nm000452,
  title = {ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative)},
  author = {Xupeng Chen},
  doi = {10.82901/nemar.nm000452},
  url = {https://doi.org/10.82901/nemar.nm000452},
}

§ 02Study · The README

About This Dataset#

This is a DERIVATIVE dataset. It repackages the single file of the Mendeley Data record “ECoG_HB_02”

(Xupeng Chen, doi:10.17632/fp4bv9gtwk.2, CC-BY-4.0), the one participant of Chen et al. 2024 (Nature Machine Intelligence) who consented to public release of neural and audio data. The raw ECoG voltage recordings were not released by the authors; what is shared is their processed 125 Hz feature matrix.

“hybrid-density ECoG data with synchronized speech and other features”.

DOI

ECoGHB02 (derivative): hybrid-density ECoG high-gamma features with synchronised speech

Processing, as stated in the paper Methods (not re-done here): ECoG sampled at 2,048 Hz and downsampled to 512 Hz; electrodes with artefacts (line noise, poor contact, high-amplitude shifts) or interictal/epileptiform activity rejected; common average reference subtracted; Hilbert envelope of the high gamma (70-150 Hz) component, downsampled to 125 Hz; each electrode normalised (z-scored) to the mean and variance of 250-ms silent periods before each trial’s stimulus in the training set.

View full README

DOI

ECoGHB02 (derivative): hybrid-density ECoG high-gamma features with synchronised speech

Processing, as stated in the paper Methods (not re-done here): ECoG sampled at 2,048 Hz and downsampled to 512 Hz; electrodes with artefacts (line noise, poor contact, high-amplitude shifts) or interictal/epileptiform activity rejected; common average reference subtracted; Hilbert envelope of the high gamma (70-150 Hz) component, downsampled to 125 Hz; each electrode normalised (z-scored) to the mean and variance of 250-ms silent periods before each trial’s stimulus in the training set.
Measured here: values -2.388 to 4.466, unitless (z-score). 128 columns carry signal and 97 columns are exactly 0 for every frame; those are marked status=bad in channels.tsv (1-based columns: 2, 4, 6, 8, 16, 17, 18, 19, 20, 21, 22, 23, 32, 34, 36, 38, 46, 47, 48, 49, 50, 51, 52, 53, 62, 64, 66, 68, 76, 77, 78, 79, 80, 81, 82, 83, 92, 94, 98, 104, 106, 107, 108, 113, 118, 119, 120, 122, 124, 128, 134, 136, 137, 138, 139, 143, 147, 148, 149, 150, 152, 154, 158, 162, 164, 166, 167, 168, 169, 173, 177, 178, 179, 180, 182, 184, 188, 192, 194, 196, 197, 198, 199, 203, 204, 206, 207, 208, 209, 210, 212, 214, 216, 218, 220, 222, 224). The paper describes the hybrid-density grid as 128 contacts (64 macro contacts in an 8 x 8 layout at 10-mm spacing plus 64 interspersed 1-mm contacts). The mapping from the 225 columns to grid contacts, and contact coordinates, were not released; electrodes.tsv therefore lists the columns with n/a coordinates only.
Implant hemisphere left and sex female, per the authors’ configs/AllSubjectInfo.json in the code repository.

Time axis: not a continuous raw recording

The file is a single run, one BIDS sample per frame of ecog_alldataset (125 Hz, 1352.6 s, 0.376 h).

The release does not document how the 169075 frames were assembled (the authors’ notebook calls the arrays *_alldataset and segments trials by onset/offset indices). The trial order in the file is not chronological along the array (start_ind_re_valid_alldataset jumps backward after trial index 321), so the array is probably assembled from separate pieces. Do not treat the time axis as one continuous session and do not filter across the whole run without checking. RecordingType is therefore discontinuous.

Events

events.tsv has one row per entry of label_alldataset (372 trials, 51 distinct words), sorted by onset, with onset/duration from start_ind_re_valid_alldataset / end_ind_re_valid_alldataset (frames / 125 Hz), the word, and source_index (position in the deposit arrays). TestNum_cum in the file is 50; per the paper, 50 trials per participant were held out for testing (which ones is not encoded in the file).

Data format and fidelity

BrainVision, IEEE float32, multiplexed, 125 Hz. The float64 values were cast to float32: max absolute deviation 2.38e-07 (max relative 5.96e-08). The .eeg file reads back bit-exactly equal to float32 of the deposit, also through MNE-BIDS. The BrainVision header carries no unit (readers default to microvolts); the values are z-scores, and channels.tsv says units n/a.

sourcedata/

sourcedata/HB02.h5 is the deposit file, byte-identical (SHA-256 018294464a291521a6f6c14878ec04dd95ac3e953123dfe3c54a75d1f9c8aa5d). Besides the ECoG features it holds the synchronised speech audio (wave_re_alldataset, 16 kHz, 128 audio samples per ECoG frame), spectrograms (wave_re_spec_alldataset, wave_re_spec_amp_alldataset), formants, pitch, intensity, the noise profile (noisesample_re_alldataset) and the trial labels/indices. The speech audio was published by the authors under CC-BY-4.0 with the participant’s consent (paper, Data availability).

Source and licence

Data: Chen, Xupeng (2024), “ECoG_HB_02”, Mendeley Data, V2, doi:10.17632/fp4bv9gtwk.2. Licence CC-BY-4.0 (kept).
Paper: Chen, X., Wang, R., Khalilian-Gourtani, A., Yu, L., Dugan, P., Friedman, D., Doyle, W., Devinsky, O., Wang, Y. & Flinker, A. A neural speech decoding framework leveraging deep learning and speech synthesis. Nature Machine Intelligence 6, 467-480 (2024). doi:10.1038/s42256-024-00824-8
Code: flinkerlab/neural_speech_decoding
Ethics (paper): Institutional Review Board of NYU Grossman School of Medicine; written and oral consent.
Funding: NSF IIS-1912286 and 2309057; NIH R01NS109367, R01NS115929, R01DC018805.
Changes made here (CC-BY “indicate if changes were made”): format conversion only (float64 -> float32 BrainVision), sidecars and events added; the original file is included unchanged.

NEMAR Metadata#

[![DOI](https://img.shields.io/badge/DOI-10.82901%2Fnemar.nm000452-blue)](https://doi.org/10.82901/nemar.nm000452) # ECoG_HB_02 (derivative): hybrid-density ECoG high-gamma features with synchronised speech This is a DERIVATIVE dataset. It repackages the single file of the Mendeley Data record “ECoG_HB_02” (Xupeng Chen, doi:10.17632/fp4bv9gtwk.2, CC-BY-4.0), the one participant of Chen et al. 2024 (Nature Machine Intelligence) who consented to public release of neural and audio data. The raw ECoG voltage recordings were not released by the authors; what is shared is their processed 125 Hz feature matrix. ## What the signal is - Source: ecog_alldataset in HB02.h5 (169075 frames x 225 columns, float64), described by the record as

“hybrid-density ECoG data with synchronized speech and other features”.

Processing, as stated in the paper Methods (not re-done here): ECoG sampled at 2,048 Hz and downsampled to 512 Hz; electrodes with artefacts (line noise, poor contact, high-amplitude shifts) or interictal/epileptiform activity rejected; common average reference subtracted; Hilbert envelope of the high gamma (70-150 Hz) component, downsampled to 125 Hz; each electrode normalised (z-scored) to the mean and variance of 250-ms silent periods before each trial’s stimulus in the training set.
Measured here: values -2.388 to 4.466, unitless (z-score). 128 columns carry signal and 97 columns are exactly 0 for every frame; those are marked status=bad in channels.tsv (1-based columns: 2, 4, 6, 8, 16, 17, 18, 19, 20, 21, 22, 23, 32, 34, 36, 38, 46, 47, 48, 49, 50, 51, 52, 53, 62, 64, 66, 68, 76, 77, 78, 79, 80, 81, 82, 83, 92, 94, 98, 104, 106, 107, 108, 113, 118, 119, 120, 122, 124, 128, 134, 136, 137, 138, 139, 143, 147, 148, 149, 150, 152, 154, 158, 162, 164, 166, 167, 168, 169, 173, 177, 178, 179, 180, 182, 184, 188, 192, 194, 196, 197, 198, 199, 203, 204, 206, 207, 208, 209, 210, 212, 214, 216, 218, 220, 222, 224). The paper describes the hybrid-density grid as 128 contacts (64 macro contacts in an 8 x 8 layout at 10-mm spacing plus 64 interspersed 1-mm contacts). The mapping from the 225 columns to grid contacts, and contact coordinates, were not released; electrodes.tsv therefore lists the columns with n/a coordinates only.
Implant hemisphere left and sex female, per the authors’ configs/AllSubjectInfo.json in the code repository.

## Time axis: not a continuous raw recording The file is a single run, one BIDS sample per frame of ecog_alldataset (125 Hz, 1352.6 s, 0.376 h). The release does not document how the 169075 frames were assembled (the authors’ notebook calls the arrays *_alldataset and segments trials by onset/offset indices). The trial order in the file is not chronological along the array (start_ind_re_valid_alldataset jumps backward after trial index 321), so the array is probably assembled from separate pieces. Do not treat the time axis as one continuous session and do not filter across the whole run without checking. RecordingType is therefore discontinuous. ## Events events.tsv has one row per entry of label_alldataset (372 trials, 51 distinct words), sorted by onset, with onset/duration from start_ind_re_valid_alldataset / end_ind_re_valid_alldataset (frames / 125 Hz), the word, and source_index (position in the deposit arrays). TestNum_cum in the file is 50; per the paper, 50 trials per participant were held out for testing (which ones is not encoded in the file). ## Data format and fidelity BrainVision, IEEE float32, multiplexed, 125 Hz. The float64 values were cast to float32: max absolute deviation 2.38e-07 (max relative 5.96e-08). The .eeg file reads back bit-exactly equal to float32 of the deposit, also through MNE-BIDS. The BrainVision header carries no unit (readers default to microvolts); the values are z-scores, and channels.tsv says units n/a. ## sourcedata/ sourcedata/HB02.h5 is the deposit file, byte-identical (SHA-256 018294464a291521a6f6c14878ec04dd95ac3e953123dfe3c54a75d1f9c8aa5d). Besides the ECoG features it holds the synchronised speech audio (wave_re_alldataset, 16 kHz, 128 audio samples per ECoG frame), spectrograms (wave_re_spec_alldataset, wave_re_spec_amp_alldataset), formants, pitch, intensity, the noise profile (noisesample_re_alldataset) and the trial labels/indices. The speech audio was published by the authors under CC-BY-4.0 with the participant’s consent (paper, Data availability). ## Source and licence - Data: Chen, Xupeng (2024), “ECoG_HB_02”, Mendeley Data, V2, doi:10.17632/fp4bv9gtwk.2. Licence CC-BY-4.0 (kept). - Paper: Chen, X., Wang, R., Khalilian-Gourtani, A., Yu, L., Dugan, P., Friedman, D., Doyle, W., Devinsky, O., Wang, Y. & Flinker, A. A neural speech decoding framework leveraging deep learning and speech synthesis. Nature Machine Intelligence 6, 467-480 (2024). doi:10.1038/s42256-024-00824-8 - Code: https://github.com/flinkerlab/neural_speech_decoding - Ethics (paper): Institutional Review Board of NYU Grossman School of Medicine; written and oral consent. - Funding: NSF IIS-1912286 and 2309057; NIH R01NS109367, R01NS115929, R01DC018805. - Changes made here (CC-BY “indicate if changes were made”): format conversion only (float64 -> float32 BrainVision),

sidecars and events added; the original file is included unchanged.

License: CC-BY-4.0

Authors:

Xupeng Chen

Versions:

Version	DOI	Released
`current`	10.82901/nemar.nm000452

§ 03Cohort · Participants

Cohort#

Dataset Statistics#

Sex composition

Female
1

Channel counts: 225 ch (n=1 recordings)

Sampling frequencies: 125.0 Hz (n=1 recordings)

Total recording duration: 22 min

§ 04Signal · Electrodes & trace

Signal · Electrodes & live trace#

Fig. 01 Signal & montage 225 ch · iEEG · 125 Hz · 1 subjects, 1 recordings

Live trace viewer — sub-HB02 · task-speech

Showing one representative recording out of 1 subjects and 1 recordings in this dataset. Browse the full set on OpenNeuro; drop any other _ieeg.{set,edf,bdf,vhdr} file onto the viewer (or pass ?ieeg=<url>) to inspect it.

No scalp electrode layout is currently indexed for this dataset. Once the eegdash montage registry ingests it, the interactive viewer will appear here automatically.

NEMAR Processing Statistics#

The plots below are generated by NEMAR’s automated EEG pipeline. The histogram shows pipeline success for data cleaning and ICA decomposition, the percentage of data frames and EEG channels retained after artefact removal, line noise per channel (RMS, dB), and the age/gender distribution of participants.

HED event descriptors word cloud

§ 05Manifest · BIDS tree

Manifest#

File Explorer#

Browse the BIDS file structure of this dataset. Records are fetched on demand from the EEGDash catalog the first time you open the explorer.

Recordings—

Files—

Subjects—

Modalities—

Click to load file structure…

§ 06API · Programmatic access

API Reference#

Signature

eegdash.dataset

class

eegdash.dataset.NM000452(cache_dir, query=None, s3_bucket=None, **kwargs)

Bases: EEGDashDataset

Author (year)—

Canonical—

Importable asNM000452

Sourceeegdash/dataset/registry.py · [source ↗]

class eegdash.dataset.NM000452(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative)

Study:: nm000452 (NeMAR)
Author (year):: —
Canonical:: —

Also importable as: NM000452.

Modality: ieeg; Subject type: Unknown. Subjects: 1; recordings: 1; tasks: 1.

Parameters:

cache_dir (str | Path) – Directory where data are cached locally.
query (dict | None) – Additional MongoDB-style filters to AND with the dataset selection. Must not contain the key dataset.
s3_bucket (str | None) – Base S3 bucket used to locate the data.
**kwargs (dict) – Additional keyword arguments forwarded to EEGDashDataset.

data_dir#

Local dataset cache directory (cache_dir / dataset_id).

Type:: Path

query#

Merged query with the dataset filter applied.

Type:: dict

records#

Metadata records used to build the dataset, if pre-fetched.

Type:: list[dict] | None

Notes

Each item is a recording; recording-level metadata are available via dataset.description. query supports MongoDB-style filters on fields in ALLOWED_QUERY_FIELDS and is combined with the dataset filter. Dataset-specific caveats are not provided in the summary metadata.

References

OpenNeuro dataset: https://openneuro.org/datasets/nm000452 NeMAR dataset: https://nemar.org/dataexplorer/detail?dataset_id=nm000452 DOI: https://doi.org/10.82901/nemar.nm000452

Examples

>>> from eegdash.dataset import NM000452
>>> dataset = NM000452(cache_dir="./data")
>>> recording = dataset[0]
>>> raw = recording.load()

__init__(cache_dir: str, query: dict | None = None, s3_bucket: str | None = None, **kwargs)[source]#

save(path: str, overwrite: bool = False, offset: int = 0)[source]#

Save datasets to files by creating one subdirectory for each dataset:

path/
    0/
        0-raw.fif | 0-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)
    1/
        1-raw.fif | 1-epo.fif
        description.json
        raw_preproc_kwargs.json (if raws were preprocessed)
        window_kwargs.json (if this is a windowed dataset)
        window_preproc_kwargs.json  (if windows were preprocessed)
        target_name.json (if target_name is not None and dataset is raw)

Parameters:

path (str) –

Directory in which subdirectories are created to store
-raw.fif | -epo.fif and .json files to.
overwrite (bool) – Whether to delete old subdirectories that will be saved to in this call.
offset (int) – If provided, the integer is added to the id of the dataset in the concat. This is useful in the setting of very large datasets, where one dataset has to be processed and saved at a time to account for its original position.

Access modesMNE → braindecode → PyTorch → ML

.rawMNE Raw object — standard tools (filter, epoch, ICA, plot_psd).mne

BaseConcatDatasetEach record is a lazy BaseDataset from braindecode — windowed via create_windows_from_events.braindecode

DataLoaderWraps the windowed dataset into a PyTorch DataLoader; supports parallel workers and on-the-fly augmentations.pytorch

Zarr cacheOptional braindecode Zarr mirror for fast resume; persisted to cache_dir.zarr

Hugging FaceNo per-dataset mirror published yet — browse the EEGDash org listing for sibling datasets. See the datasets loader API.huggingface

Croissant 1.0Machine-readable JSON-LD descriptor — NM000452.croissant.json (MLCommons schema, ingestible by PyTorch / TensorFlow / JAX).mlcommons

Examples using EEGDashcurated · start here

Find datasets with the EEGDash APIQuery the catalogue, filter by task or modality, list candidates.

Load one EEG recordingResolve a single record to an MNE Raw with channels and events.

EEG recording to PyTorch DataLoaderWrap braindecode windows in a DataLoader for model training.

Preprocess EEG and create windowsFilter, resample, epoch — and persist the windowed dataset.

Save and reload prepared dataCache a windowed dataset to disk and reattach it without recompute.

Download a dataset locallyPrefetch BIDS files to a local cache and validate the layout.

Swap any load_dataset(...) call for nm000452 to reproduce the tutorial on this dataset.

Citation

Xupeng Chen (2024). ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative). 10.82901/nemar.nm000452

Provenance

¹Contributed to nemar in BIDS format.

²Curated & ingested by the EEGDash catalog; see CITATION.cff for canonical reference.

³Persistent identifier: 10.82901/nemar.nm000452.

Related & sibling datasets

NM000469IEEG · 1 subj NM000437IEEG · 5 subj NM000475IEEG · 3 subj NM000472IEEG · 1 subj NM000462IEEG · 3 subj

+ 1 more — see See Also below →

BIDS

BIDS 1.10.0

Sidecars

events · events.json · channels · eeg.json

Provenance

CC-BY-4.0 · 10.82901/nemar.nm000452

Machine-readable

schema.org/Dataset · Croissant

Mirrors

OpenNeuro · NEMAR · HF org

Dataset ID	`NM000452`
Title	ECoG_HB_02: hybrid-density ECoG high-gamma features with synchronised speech, one participant (derivative)
Author (year)	—
Canonical	—
Importable as	`NM000452`
Year	2024
Authors	Xupeng Chen
License	CC-BY-4.0
Citation / DOI	10.82901/nemar.nm000452
Source links	OpenNeuro \| NeMAR \| Source URL

NM000452: ieeg dataset, 1 subjects#

Quickstart#

About This Dataset#

NEMAR Metadata#

Cohort#

Dataset Statistics#

Signal · Electrodes & live trace#

NEMAR Processing Statistics#

Manifest#

File Explorer#

API Reference#

Citation

Provenance

Related & sibling datasets

See Also#