Discover datasets

HEFMI-ICH is a hybrid EEG-fNIRS motor imagery dataset designed for brain-computer interface applications in intracerebral hemorrhage rehabilitation. The dataset comprises 37 participants (17 healthy controls and 20 ICH patients) performing 2-class hand motor imagery tasks (left and right hand grasping) across multiple sessions. With 32-channel EEG recordings at 256 Hz and 3,330 trials, this dataset supports the development and evaluation of BCI systems for clinical stroke rehabilitation.

Participants

37

Channels

32 (10-10)

Citations

5

HED

v8.4.0

Size

6.42 GB

Version

v1.0.3

Updated

Aug 18, 2026

A 4-target, 2-class code-VEP brain-computer interface dataset from 12 healthy participants comparing burst c-VEP and m-sequence stimulation paradigms at two amplitude depths (100% and 40%). The study evaluates classification performance and user experience using 32-channel EEG recorded at 500 Hz. This derivative dataset is derived from the original Castillos et al. (2023) study and optimizes stimulus design for reactive BCI applications while maintaining visual comfort. CNN-based decoding achieved up to 95.6% accuracy (best-case scenario with 52.8 s of calibration data).

Participants

12

Channels

32 (10-10)

Citations

21

HED

v8.4.0

Size

352 MB

Version

v1.0.2

Updated

Aug 18, 2026

This dataset comprises electroencephalographic recordings from 12 healthy participants performing a reactive brain-computer interface task based on code-modulated visual evoked potentials (c-VEP). The study introduces an innovative 'Burst c-VEP' paradigm using short bursts of aperiodic visual flashes at 2-4 Hz, systematically varying stimulus amplitude (100% vs. 40%) to optimize classification performance while improving visual comfort. Data were collected using a 32-channel BrainProducts LiveAmp system at 500 Hz sampling rate and analyzed with convolutional neural networks for bitwise decoding.

Participants

12

Channels

32 (10-10)

Citations

21

HED

v8.4.0

Size

337 MB

Version

v1.0.2

Updated

Aug 18, 2026

This dataset comprises EEG recordings from 12 healthy participants performing a code-modulated visual evoked potential (c-VEP) brain-computer interface task. The study introduces Burst c-VEP, a novel variant using short bursts of aperiodic visual flashes, and compares its performance to traditional m-sequence stimuli across two amplitude depths (100% and 40%). Data were acquired at 500 Hz using a 32-channel BrainProducts system and include 60 trials per subject with convolutional neural network-based classification achieving 95.6% accuracy.

Participants

12

Channels

32 (10-10)

Citations

21

HED

v8.4.0

Size

352 MB

Version

v1.0.2

Updated

Aug 18, 2026

This open-access EEG dataset comprises multi-session, multi-task recordings from 15 healthy participants performing resting state and graded difficulty levels of the MATB-II task. Acquired at 500 Hz using 62 active electrodes, the dataset includes 90 trials per participant across two sessions and is designed to support passive brain-computer interface research and mental workload estimation in neuroergonomic applications.

Participants

15

Channels

61 (10-10)

Citations

64

HED

v8.4.0

Size

4.04 GB

Version

v1.0.3

Updated

Aug 18, 2026

A code-modulated Visual Evoked Potential (c-VEP) electroencephalography dataset from 12 healthy participants investigating burst c-VEP as a reactive brain-computer interface paradigm. This dataset contains m-sequence c-VEP data at 40% amplitude modulation depth. The original study employed a factorial design manipulating stimulus pattern (burst vs. m-sequence) and amplitude depth (100% vs. 40% modulation) to optimize classification accuracy while improving user visual comfort. Data were acquired at 500 Hz using 32-channel EEG with convolutional neural network classification achieving up to 95.6% accuracy on the full amplitude burst condition.

Participants

12

Channels

32 (10-10)

Citations

22

HED

v8.4.0

Size

340 MB

Version

v1.0.2

Updated

Aug 18, 2026

This dataset comprises electroencephalographic recordings from 12 healthy subjects during music listening with and without a passive head-mounted display (PHMD). The study employed a 16-channel EEG system sampling at 512 Hz. A passive head-mounted display is a head-mounted device without integrated electronics, consisting of a smartphone holder. The dataset provides raw, unfiltered EEG data in mat and csv formats, designed to support research on the effects of head-mounted displays and auditory stimuli on brain activity. This is a BIDS-converted derivative of the original Zenodo dataset (10.5281/zenodo.2617084).

Participants

12

Channels

16 (10-10)

Citations

19

HED

v8.4.0

Size

516 MB

Version

v1.0.5

Updated

Aug 18, 2026

BigP3BCI Study J is a P300-based brain-computer interface dataset comprising EEG recordings from 20 healthy subjects performing a 9x8 character grid speller task. This derivative dataset is part of the larger BigP3BCI collection, the largest public P300 BCI dataset with ~267 subjects across 20 studies. The data were acquired at 256 Hz using 16-channel EEG with a g.USBamp amplifier and include target and non-target event classifications suitable for machine learning applications.

Participants

20

Channels

16 (10-10)

HED

v8.4.0

Size

1.49 GB

Version

v1.0.3

Updated

Aug 18, 2026

A large-scale longitudinal dataset of sensorimotor rhythm-based brain-computer interface (BCI) training in 62 healthy adults. The dataset comprises over 600 hours of EEG recordings across 598 sessions with more than 250,000 trials of motor imagery tasks (left hand, right hand, both hands, and rest). This resource enables investigation of BCI learning dynamics and algorithm development for non-invasive neural control applications.

Participants

62

Channels

60 (10-10)

Citations

59

HED

v8.4.0

Size

722 GB

Version

v1.0.4

Updated

Aug 18, 2026

A comprehensive EEG dataset comprising 54 healthy subjects performing three major brain-computer interface (BCI) paradigms: motor imagery (MI), event-related potential (ERP), and steady-state visually evoked potential (SSVEP) across two sessions. The dataset investigates BCI illiteracy rates and performance variations, revealing that while MI showed the highest illiteracy rate (53.7%), all participants could control at least one BCI paradigm. Data were acquired at 1000 Hz using 62 EEG channels with concurrent electromyography recordings.

Participants

54

Channels

62 (10-05)

Citations

99

HED

v8.4.0

Size

121 GB

Version

v1.0.5

Updated

Aug 18, 2026