PEARL-Neuro Database

Description

The PEARL-Neuro Database (Polish Electroencephalography, Alzheimer’s Risk-genes, Lifestyle and Neuroimaging) is a multimodal dataset of healthy middle-aged adults at genetic risk of late-onset Alzheimer’s disease, collected at the Nencki Institute of Experimental Biology, Warsaw, and described by (Dzianok & Kublik, 2024) in Scientific Data. The study recruited a cohort of 192 cognitively healthy, middle-aged (50—63 years) Polish individuals and characterized them genetically for two Alzheimer’s risk loci — apolipoprotein E (APOE) and phosphatidylinositol-binding clathrin assembly protein (PICALM) — alongside psychometric (memory, intelligence, mood, personality, stress-coping), demographic, and health/lifestyle measures. A subset of ~79 participants additionally underwent blood testing (blood counts, lipid profile, HSV serology) and functional neuroimaging: high-density EEG and fMRI recorded during a resting-state protocol (eyes open and eyes closed) plus two cognitive paradigms (the Multi-Source Interference Task and a Sternberg working-memory task). The stated aim is to enable study of how Alzheimer’s risk genotypes manifest in brain function, cognition, and hematology long before any clinical impairment.

Citation

• Paper DOI: 10.1038/s41597-024-03106-5
• Dataset DOI: 10.18112/openneuro.ds004796.v1.1.0
• Reference: Dzianok, P., & Kublik, E. (2024). PEARL-Neuro Database: EEG, fMRI, health and lifestyle data of middle-aged people at risk of dementia. Scientific Data 11, 276.

Catalog citation block

Dzianok, P., & Kublik, E. (2024). PEARL-Neuro Database: EEG, fMRI, health and lifestyle data of middle-aged people at risk of dementia. Scientific Data, 11, 276. DOI: 10.1038/s41597-024-03106-5. PMID 38453963; PMCID PMC10920678.

Dataset: Dzianok, P., & Kublik, E. (2024). PEARL-Neuro Database (ds004796). OpenNeuro. DOI: 10.18112/openneuro.ds004796.v1.1.0.

Download and access

Field	Value
Official source	https://openneuro.org/datasets/ds004796
Access class	open
Size	~240.3 GB across ~2,056 files (OpenNeuro, including fMRI)
Format	brainvision
BIDS	true
Mirrors	NEMAR ds004796; https://github.com/PTDZ/PEARL-Neuro (errata and updates)

Source & access

The dataset is hosted on OpenNeuro under accession ds004796 (version 1.1.0; OpenNeuro DOI 10.18112/openneuro.ds004796.v1.1.0). It is also indexed on NEMAR. OpenNeuro reports the deposit under a CC0 public-domain dedication, whereas the Scientific Data descriptor states CC-BY 4.0 — the discrepancy is noted below. Data can be retrieved via the OpenNeuro web UI, DataLad, Amazon S3, or the OpenNeuro Node.js CLI. The full OpenNeuro deposit is ~240.3 GB across ~2,056 files. A companion repository at github.com/PTDZ/PEARL-Neuro tracks errata, updates, and downstream research outputs. The work was funded by Polish National Science Centre grant 2018/31/N/HS6/03551.

Data structure

Data are organized in BIDS (Brain Imaging Data Structure), with EEG as BrainVision and (f)MRI as NIfTI (.nii.gz), plus _events.tsv, _channels.tsv, and JSON sidecars. Because the deposit lives on OpenNeuro it is DataLad/git-annex-backed — large binaries are content-addressed and fetched on demand (datalad get) rather than stored in the git tree. The GitHub companion repo notes additional, non-BIDS supplementary materials (genetic, psychometric, lifestyle tables) accompanying the neuroimaging release.

License

• Name: CC0
• Note (verbatim from the directory): CC0 on OpenNeuro; the Scientific Data descriptor states CC BY 4.0 — the repository’s data license governs; structural MRI requires a Data Use Agreement
• 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	Brain Products actiCHamp with 128-electrode actiCAP
Device class	research-cap
Channels	128
Sampling rate (Hz)	1000
Online filters	280 Hz online low-pass; no online notch or high-pass
Reference	FCz
Mains frequency (Hz)	50
Paradigms	rest-eo, rest-ec, msit, sternberg
Participants	79 (79 in the EEG/fMRI/blood subset; 192 in the full psychometric/genetic cohort)
Population	79 cognitively healthy adults 50–63 at genetic Alzheimer’s risk (APOE/PICALM), of a 192-person cohort
Clinical groups	none
Sessions	1
Durations	rest 4 min EO + 6 min EC; MSIT ~8.5 min (2 runs); Sternberg ~13 min

Acquisition

EEG was recorded with a Brain Products actiCHamp amplifier and a high-density actiCAP cap carrying 128 electrodes (Dzianok & Kublik, 2024). The online reference was FCz (re-referencing is left to the user), the sampling rate was 1000 Hz, and the only online filter was a 280 Hz low-pass (no online notch or high-pass applied). Electrode impedances were kept to roughly 5—10 kΩ. The fMRI/MRI portion was acquired separately; structural MRI is access-restricted (see caveats).

Participants

The full Phase-I cohort is N=192 healthy middle-aged adults (89 male / 103 female), aged 50—63 (mean ≈55.1 ± 3.1 years); 77.6% held higher education. The neuroimaging/blood subset (Phase II) is N=79 (39 male / 40 female). Inclusion required cognitive health with no diagnosed neurological/psychiatric disease; exclusions included epilepsy, metal implants (MRI safety), sleep disorders, and recent infection (Dzianok & Kublik, 2024). Genotype distributions for the cohort: APOE haplotypes ε3/ε3 ≈62%, ε3/ε4 ≈24%, ε2/ε3 ≈10.9%, with rarer ε2/ε2, ε2/ε4, ε4/ε4 each ≤1.6%; PICALM rs3851179 A/A ≈8.3%, A/G ≈50.5%, G/G ≈41.1%. Note: 8 participants were withheld from public release relative to the originally screened sample, yielding the N=192 final.

Tasks / conditions

Per participant the EEG/fMRI session comprised four paradigms (Dzianok & Kublik, 2024):

• Resting state — 4 min eyes-open + 6 min eyes-closed for EEG (the fMRI resting run was ~12 min, eyes closed).
• Multi-Source Interference Task (MSIT) — a Stroop-like cognitive-interference task, ~8.5 min total (≈166 trials across 2 runs).
• Sternberg memory task — a working-memory item-recognition paradigm, ~13 min.

In BIDS these appear as task labels rest, msit, and sternberg.

Used on this site

L7.7, C6

Loader

Fetches only the stated subset. Recorded loader: DataLad (OpenNeuro ds004796).

# OpenNeuro serves ds004796 as a DataLad/git-annex repository; fetch one subject only.
datalad install https://github.com/OpenNeuroDatasets/ds004796.git
cd ds004796
datalad get sub-001   # replace with the subject id you need

Caveats worth knowing

• No online notch at 1000 Hz.

Notable caveats

• License ambiguity: OpenNeuro/NEMAR list CC0, but the Scientific Data descriptor states CC-BY 4.0. Treat attribution to (Dzianok & Kublik, 2024) as required regardless.
• Restricted structural MRI: EEG and functional/resting data are public, but structural MRI is protected and requires a Data Use Agreement through the authors (e.kublik@nencki.edu.pl / p.dzianok@nencki.edu.pl).
• Line noise: recorded in Poland (50 Hz mains), with no online notch — expect a 50 Hz line component (and harmonics) requiring offline notch/clean-line handling.
• Reference choice: online reference is FCz; any topographic or connectivity analysis should re-reference explicitly.
• No fluid biomarkers: the database has no plasma amyloid-β or phosphorylated-tau measures; “dementia risk” is genetic (APOE/PICALM), not biomarker- or diagnosis-confirmed. The cohort is cognitively healthy, so this is a risk-enriched, not patient, sample.
• Subset mismatch: psychometric/genetic data cover all 192 participants but EEG/fMRI/blood cover only ~79.

Related datasets

• ERP CORE (Compendium of Open Resources and Experiments)
• EEG Motor Movement/Imagery Dataset (EEGMMIDB)
• LEMON (MPI-Leipzig Mind-Brain-Body)
• Dortmund Vital Study resting EEG
• SRM Resting-state EEG

# Loader recorded in the catalog: DataLad (OpenNeuro ds004796)
# TODO(confirm) the exact call and the subset to fetch. Official source: https://openneuro.org/datasets/ds004796