Experiment Details Electroencephalography recordings from 16 subjects to fast streams of gabor-like stimuli. Images were presented in rapid serial visual presentation streams at 6.67Hz and 20Hz rates. Participants performed an orthogonal fixation colour change detection task.

Experiment length: 1 hour Raw and preprocessed data are available online through openneuro: https://openneuro.org/datasets/ds004357. Supplementary Material and analysis scripts are available on github: https://github.com/Tijl/features-eeg

Supporting materials for the GX Dataset.

The GX Dataset is a dataset of combined tES, EEG, physiological, and behavioral signals from human subjects.

Publication

A full data descriptor is published in Nature Scientific Data. Please cite this work as:

Gebodh, N., Esmaeilpour, Z., Datta, A. et al. Dataset of concurrent EEG, ECG, and behavior with multiple doses of transcranial electrical stimulation. Sci Data 8, 274 (2021). https://doi.org/10.1038/s41597-021-01046-y

Descriptions

A dataset combining high-density electroencephalography (EEG) with physiological and continuous behavioral metrics during transcranial electrical stimulation (tES; including tDCS and tACS). Data includes within subject application of nine High-Definition tES (HD-tES) types targeted three brain regions (frontal, motor, parietal) with three waveforms (DC, 5Hz, 30Hz), with more than 783 total stimulation trials over 62 sessions with EEG, physiological (ECG or EKG, EOG), and continuous behavioral vigilance/alertness metrics (CTT task).

Acknowledgments

Portions of this study were funded by X (formerly Google X), the Moonshot Factory. The funding source had no influence on study conduction or result evaluation. MB is further supported by grants from the National Institutes of Health: R01NS101362, R01NS095123, R01NS112996, R01MH111896, R01MH109289, and (to NG) NIH-G-RISE T32GM136499.

We would like to thank Yuxin Xu and Michaela Chum for all their technical assistance.

Extras

For downsampled data (1 kHz ) please see (in .mat format):

• Experiment 1 : https://doi.org/10.5281/zenodo.8401160
• Experiment 2 : https://doi.org/10.5281/zenodo.3840617

Code used to import, process, and plot this dataset can be found here:

• GitHub : https://github.com/ngebodh/GX_tES_EEG_Physio_Behavior

Additional figures for this project have been shared on Figshare. Trial-wise figures can be found here:

• PSD: https://doi.org/10.6084/m9.figshare.14810517.v1
• Topoplots During Stimulation: https://doi.org/10.6084/m9.figshare.14810478
• Voltage timeseries, spectrogram and behavior: https://doi.org/10.6084/m9.figshare.14810442.v1

The full dataset is also provided in BIDS format here:

• Data in BIDS format: https://doi.org/10.18112/openneuro.ds003670.v1.1.0

Data License
Creative Common 4.0 with attribution (CC BY 4.0)

NOTE

Please email ngebodh01@citymail.cuny.edu with any questions.

Follow @NigelGebodh for latest updates.

Dataset setting out to investigate neural responses to continuous musical pieces with bipolar EEG. Analysis code (and usage instructions) to derive neural responses to the temporal fine structure of the stimuli is on Github. The EEG data processed to this end is provided here, as well as the raw data to enable different analyses (e.g. slower cortical responses).

# Introduction

This dataset contains bipolar scalp EEG responses of 17 subjects listening to continuous musical pieces (Bach's Two-Part Inventions), and performing a vibrato detection task.

Naming conventions:

- The subject IDs are EBIP01, EBIP02 ... EBIP17.
- The different conditions are labelled to indicate the instrument that was being attended: fG and fP for the Guitar and Piano in quiet (Single Instrument (SI) conditions), respectively; and fGc and fPc for Competing conditions where both the instruments are playing together, but where the subjects should be selectively attending to the Guitar or Piano, respectively (Competing Instrument (CI) conditions).
- An appended index from 2 to 7 designates the invention that was played (index 1 corresponds to the training block for which no EEG data was recorded). Note that this index does not necessarily corresponds to the order in which the stimuli were played (order was pseudo-randomised).

For example, the EEG file named EBIP08_fGc_4 contains EEG data from subject EBIP08 performing the competing instrument task (CI condition), attending to the guitar (ignoring the piano), and the stimulus that was played was the invention #4.

# Content

The general organisation of the dataset is as follow:

data
 ├─── behav              folder containing the behavioural data
 ├─── EEG            folder containing the EEG data
 │   ├─── processed
 │   └─── raw
 ├─── linearModelResults       folder containing the results from the analysis code
 └─── stimuli folder containing the stimuli
     ├─── features
     ├─── processedInventions
     └─── rawInventions

This general organisation is the one expected by the code. The location of the data folder and/or these main folders can be personalised in the functions/+EEGmusic2020/getPath.m function in the Github repository. The architecture of the sub-folders in each of these folders is specified by the functions makePathEEGFolder, makePathFeatureFiles and makePathSaveResults. The naming of the files within them is implemented by makeNameEEGDataFile and makeNameEEGDataFile (all these functions being in functions/+EEGmusic2020).

- The behav folder is structured as follow:

behav
├─── EBIP02
│ ├─── EBIP02_keyboardInputs_fGc_2.mat file containing variables:
│ │ ├─── timePressed key press time (in seconds, relative to stimulus onset)
│ │ └─── keyCode ID of the keys that were pressed
│ └─── ...
├─── ...
├─── vibTime
│ ├─── vibTime_2.mat file containing variables:
│ │ ├─── idxNoteVib index (in the MIDI files) of the notes in which vibratos were inserted
│ │ ├─── instrumentOrder order of the instruments in idxNoteVib and vibTiming variables
│ │ └─── vibTiming timing of vibrato onsets in the track (in s)
│ ├─── ...
└─── clickPerformance_RT_2.0.mat file containing behavioural results for all subjects (FPR, TPR, etc.):

instrumentOrder indicates to what instrument each column of idxNoteVib and vibTiming refers to. The data for EBIP01 missing due to a technical error.

- The EEG/raw folder contains unprocessed EEG data for all subjects, and files indicating the order in which the inventions were played. It is structured as follow:

EEG
├─── raw
│ ├─── EBIP01
│ │ ├─── EBIP01_EEGExpParam.mat file containing variables:
│ │ │ ├─── conditionOrder whether this subject started by listening to the guitar or piano
│ │ │ └─── partsOrder order in which the inventions were presented to this subject
│ │ ├─── EBIP01_fGc_2.[eeg/vhdr/vmrbk] raw EEG data files
│ │ ├─── ...
│ ├─── ...

The conditionOrder variable can assume two values: either {'fG','fP'} indicating the subject started by listening to the guitar or {'fP','fG'} indicating the subject started by listening to the piano. The partsOrder variable is a 2 x 6 matrix containing the indices (2 to 7) of the inventions that were played, ordered in the presentation order. During the first block, the instrument conditionOrder{1} was attended, and the invention # partsOrder(1,1) was played. During the second block, the instrument conditionOrder{2} was attended, and the invention #partsOrder(2,1) was played, etc.

Each EEG files contains 3 channels: 2 are the bipolar electrophysiological channels, and one (labelled Sound) contains a recording of the stimuli that were played and that was simultaneously recorded at the same sampling rate as the EEG data (5 kHz) by the amplifier through an acoustic adapter. The files also contain triggers that indicate the beginning and end of the stimuli (labelled S 1 and S 2 respectively). The sound channel and triggers can be used to temporally align the EEG data and stimuli.

The EEG/processed folder contains processed EEG data for all subjects, as required for the analyses carried out in the code. It is organised as follow:

EEG
├─── processed
│ └─── Fs-5000 sampling rate
│ └─── HP-130 processing that was applied
│ │ ├─── EBIP01
│ │ │ ├─── ... processed EEG data files
│ │ ├─── ...
│ └─── noProc
│ ├─── ...

This structure is specified by the makePathEEGFolder function, and the file names by makeNameEEGDataFile. In the files in the noProc folder, the EEG data was simply aligned with the stimuli, but is otherwise unprocessed. Events were added to mark stimulus onset and offset (labelled stimBegin and stimEnd). In the other folders, the EEG data was furthermore high-pass filtered at 130 Hz (HP-130).

- The linearModelResults folder contains the results from the linear model analyses:

linearModelResults
└─── Fs-5000 sampling rate
│ └─── HP-130 processing of the EEG data
│ │ └─── LP-2000 processing of the stimulus feature
│ │ ├─── ... result files
│ ├─── ...

This structure and file names are specified by the makePathSaveResults function.

- The rawInventions folder contains the orignal data that was used to construct the stimuli:

rawInventions
├─── invent1 invention index
│ ├─── invent1_60bpm.mid MIDI file
│ ├─── invent1_60bpm_guitar.wav guitar track
│ └─── invent1_60bpm_piano.wav piano track
│
├─── ...

In this folder (and only in this folder), the numbering of the inventions differs from the one otherwise used throughout. The correspondence is as shown below:
Raw invention # |

This EEG Brain Computer Interface (BCI) dataset was collected as part of the study titled: “Transcranial Focused Ultrasound to V5 Enhances Human Visual Motion Brain-Computer Interface by Modulating Feature-Based Attention”. If you use a part of this dataset in your work, please cite the following publication:J. Kosnoff, K. Yu, C. Liu, and B. He, “Transcranial Focused Ultrasound to V5 Enhances Human Visual Motion Brain-Computer Interface by Modulating Feature-Based Attention.” bioRxiv, p. 2023.09.04.556252, Sep. 5, 2023. doi: 10.1101/2023.09.04.556252.In the shared dataset, the README file includes the essential data structure and event labels for reading the EEG data collected from 25 healthy human subjects in our study. The file "BCI Outcomes" includes the behavior readout associated with specific experiment conditions in this visual motion based BCI task implemented by those human subjects. The associated code and software to read the EEG data can be accessed through GitHub at https://github.com/bfinl/tFUS-mVEPBCI-Analysis. This work was supported by NIH grants R01NS124564 (PI: B.H.), R01AT009263 (PI: B.H.), U18EB029354 (PI: B.H.), T32EB029365 (J.K.), RF1NS131069 (PI: B.H.), R01NS096761 (PI: B.H.), and NS127849 (PI: B.H.), as well as National Science Foundation Graduate Research Fellowship Program grant DGE2140739 (J.K.). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Institutes of Health or the National Science Foundation.Please direct correspondence about the dataset to: Dr. Bin He, Carnegie Mellon University, Department of Biomedical Engineering, Pittsburgh, PA 15213. E-mail: bhe1@andrew.cmu.edu

If using this dataset, please cite the following paper above and the current Zenodo repository:A. Mundanad Narayanan, R. Zink, and A. Bertrand, "EEG miniaturization limits for stimulus decoding with EEG sensor networks", Journal of Neural Engineering, vol. 18, 2021, doi: 10.1088/1741-2552/ac2629

Experiment*************

This dataset contains 255-channel electroencephalography (EEG) data collected during an auditory attention decoding experiment (AAD). The EEG was recorded using a SynAmps RT device (Compumedics, Australia) at a sampling rate of 1 kHz and using active Ag/Cl electrodes. The electrodes were placed on the head according to the international 10-5 (5%) system. 30 normal hearing male subjects between 22 and 35 years old participated in the experiment. All of them signed an informed consent form approved by the KU Leuven ethical committee.

Two Dutch stories narrated by different male speakers divided into two parts of 6 minutes each were used as the stimuli in the experiment [1]. A single trial of the experiment involved the presentation of these two parts (one of both stories) to the subject through insert phones (Etymotic ER3A) at 60dBA. These speech stimuli were filtered using a head-related transfer function (HRTF) such that the stories seemed to arrive from two distinct spatial locations, namely left and right with respect to the subject with 180 degrees separation. In each trial, the subjects were asked to attend to only one ear while ignoring the other. Four trials of 6 minutes each were carried out, in which each story part is used twice. The order of presentations was randomized and balanced over different subjects. Thus approximately 24 minutes of EEG data was recorded per subject.

File organization and details********************************

The EEG data of each of the 30 subjects are uploaded as a ZIP file with the name Sx.tar.gzip here x=0,1,2,..,29. When a zip file is extracted, the EEG data are in their original raw format as recorded by the CURRY software [2]. The data files of each recording consist of four files with the same name but different extensions, namely, .dat, ,dap, .rs3 and .ceo. The name of each file follows the following convention: Sx_AAD_P. With P taking one of the following values for each file:1. 1L2. 1R3. 2L4. 2R

The letter 'L' or 'R' in P indicates the attended direction of each subject in a recording: left and right respectively. A MATLAB function to read the software is provided in the directory called scripts. A python function to read the file is available in this Github repository [3].The original version of stimuli presented to subjects, i.e. without the HRTF filtering, can be found after extracting the stimuli.zip file in WAV format. There are 4 WAV files corresponding to the two parts of each of the two stories. These files have been sampled at 44.1 kHz. The order of presentation of these WAV files is given in the table below: Stimuli presentation and attention information of files

Trial (P) Stimuli: Left-ear Stimuli: Right-ear Attention

1L part1_track1_dry part1_track2_dry Left

1R part1_track1_dry part1_track2_dry Right

2L part2_track2_dry part2_track1_dry Left

2R part2_track2_dry part2_track1_dry Right

Additional files (after extracting scripts.zip and misc.zip):

scripts/sample_script.m: Demonstrates reading an EEG-AAD recording and extracting the start and end of the experiment.

misc/channel-layout.jpeg: The 255-channel EEG cap layout

misc/eeg255ch_locs.csv: The channel names, numbers and their spherical (theta and phi) scalp coordinates.

[1] Radioboeken voor kinderen, http://radioboeken.eu/kinderradioboeken.php?lang=NL, 2007 (Accessed: 8 Feb 2021)

[2] CURRY 8 X – Data Acquisition and Online Processing, https://compumedicsneuroscan.com/product/curry-data-acquisition-online-processing-x/ (Accessed: 8, Feb, 2021)

[3] Abhijith Mundanad Narayanan, "EEG analysis in python", 2021. https://github.com/mabhijithn/eeg-analyse , (Accessed: 8 Feb, 2021)

EEG signals with 256 channels captured from 11 subjects executing a SSVEP-based experimental protocol. Five different frequencies (6.66, 7.50, 8.57, 10.00 and 12.00 Hz) have been used for the visual stimulation, and the EGI 300 Geodesic EEG System (GES 300), using a 256-channel HydroCel Geodesic Sensor Net (HCGSN) and a sampling rate of 250 Hz has been used for capturing the signals. Check https://www.youtube.com/watch?v=8lGBVvCX5d8&feature=youtu.be for a video demonstrating one trial.Check https://github.com/MAMEM/ssvep-eeg-processing-toolbox for the processing toolbox.Check http://arxiv.org/abs/1602.00904 for the technical report.

Emotion classification using electroencephalography (EEG) data and machine learning techniques have been on the rise in the recent past. However, past studies use data from medical-grade EEG setups with long set-up times and environment constraints. The images from the OASIS image dataset were used to elicit valence and arousal emotions, and the EEG data was recorded using the Emotiv Epoc X mobile EEG headset. We propose a novel feature ranking technique and incremental learning approach to analyze performance dependence on the number of participants. The analysis is carried out on publicly available datasets: DEAP and DREAMER for benchmarking. Leave-one-subject-out cross-validation was carried out to identify subject bias in emotion elicitation patterns. The collected dataset and pipeline are made open source.

Code: https://github.com/rohitgarg025/Decoding_EEG

Summary:

This dataset contains electroencephalographic recordings of 12 subjects listening to music with and without a passive head-mounted display, that is, a head-mounted display which does not include any electronics at the exception of a smartphone. The electroencephalographic headset consisted of 16 electrodes. A full description of the experiment is available at https://hal.archives-ouvertes.fr/hal-02085118. Data were recorded during a pilot experiment taking place in the GIPSA-lab, Grenoble, France, in 2017 (Cattan and al, 2018). Python code for manipulating the data is downloadable at https://github.com/plcrodrigues/py.PHMDML.EEG.2017-GIPSA. The ID of this dataset is PHMDML.EEG.2017-GIPSA.

Full description of the experiment and dataset: https://hal.archives-ouvertes.fr/hal-02085118

Principal Investigator: Eng. Grégoire Cattan

Technical Supervisors: Eng. Pedro L. C. Rodrigues

Scientific Supervisor: Dr. Marco Congedo

ID of the dataset: PHMDML.EEG.2017-GIPSA

Summary:

This dataset contains electroencephalographic (EEG) recordings of 25 subjects testing the Brain Invaders (Congedo, 2011), a visual P300 Brain-Computer Interface inspired by the famous vintage video game Space Invaders (Taito, Tokyo, Japan). The visual P300 is an event-related potential elicited by a visual stimulation, peaking 240-600 ms after stimulus onset. EEG data were recorded by 16 electrodes in an experiment that took place in the GIPSA-lab, Grenoble, France, in 2012 (Van Veen, 2013 and Congedo, 2013). A full description of the experiment is available https://hal.archives-ouvertes.fr/hal-02126068. Python code for manipulating the data is available at https://github.com/plcrodrigues/py.BI.EEG.2012-GIPSA. The ID of this dataset is BI.EEG.2012-GIPSA.

Full description of the experiment and dataset: https://hal.archives-ouvertes.fr/hal-02126068

Principal Investigator: B.Sc. Gijsbrecht Franciscus Petrus van Veen

Technical Supervisors: Ph.D. Alexandre Barachant, Eng. Anton Andreev, Eng. Grégoire Cattan, Eng. Pedro. L. C. Rodrigues

Scientific Supervisor: Ph.D. Marco Congedo

ID of the dataset: BI.EEG.2012-GIPSA

EEG: silent and perceive speech on 30 spanish sentences Large Spanish Speech EEG dataset

Authors

• Carlos Valle
• Carolina Mendez-Orellana
• María Rodríguez-Fernández

Resources:

• Code availaible at: https://github.com/CarlosValleA/Large_Spanish_EEG
• Publication:

Abstract: Decoding speech from brain activity can enable communication for individuals with speech disorders. Deep neural networks have shown great potential for speech decoding applications, but the large data sets required for these models are usually not available for neural recordings of speech impaired subjects. Harnessing data from other participants would thus be ideal to create speech neuroprostheses without the need of patient-specific training data. In this study, we recorded 60 sessions from 56 healthy participants using 64 EEG channels and developed a neural network capable of subject-independent classification of perceived sentences. We found that sentence identity can be decoded from subjects without prior training achieving higher accuracy than mixed-subject models. The development of subject-independent models eliminates the need to collect data from a target subject, reducing time and data collection costs during deployment. These results open new avenues for creating speech neuroprostheses when subjects cannot provide training data.

Please contact us at this e-mail address if you have any question: cgvalle@uc.cl

The TMS-EEG signal analyser (TESA) is an open source extension for EEGLAB that includes functions necessary for cleaning and analysing TMS-EEG data. Both EEGLAB and TESA run in Matlab (r2015b or later). The attached files are example data files which can be used with TESA.

To download TESA, visit here:

http://nigelrogasch.github.io/TESA/

To read the TESA user manual, visit here:

https://www.gitbook.com/book/nigelrogasch/tesa-user-manual/details

File info:

example_data.set

WARNING: file size = 1.1 GB. A raw data set for trialling TESA. Load the data file in to EEGLAB using the existing EEGLAB data set functions. Note that both the .fdt and .set files are required.

example_data_epoch_demean.set

File size = 340 MB. A partially processed data file of smaller size corresponding to step 8 of the analysis pipeline in the TESA user manual. Channel locations were loaded, unused electrodes removed, bad electrodes removed, epoched (-1000 to 1000 ms) and demeaned (baseline correct -1000 to 1000). Load the data file in to EEGLAB using the existing EEGLAB data set functions. Note that both the .fdt and .set files are required.

example_data_epoch_demean_cut_int_ds.set

File size = 69 MB. A further processed data file even smaller in size corresponding to step 11 of the analysis pipeline in the TESA user manual. In addition to the above steps, data around the TMS pulse artifact was removed (-2 to 10 ms), replaced using linear interpolation, and downsampled to 1,000 Hz. Load the data file in to EEGLAB using the existing EEGLAB data set functions. Note that both the .fdt and .set files are required.

Example data info:

Monophasic TMS pulses (current flow = posterior-anterior in brain) were given through a figure-of-eight coil (external diameter = 90 mm) connected to a Magstim 2002 unit (Magstim company, UK). 150 TMS pulses were delivered over the left superior parietal cortex (MNI coordinates: -20, -65, 65) at a rate of 0.2 Hz ± 25% jitter. TMS coil position was determined using frameless stereotaxic neuronavigation (Localite TMS Navigator, Localite, Germany) and intensity was set at resting motor threshold of the first dorsal interosseous muscle (68% maximum stimulator output). EEG was recorded from 62 TMS-specialised, c-ring slit electrodes (EASYCAP, Germany) using a TMS-compatible EEG amplifier (BrainAmp DC, BrainProducts GmbH, Germany). Data from all channels were referenced to the FCz electrode online with the AFz electrode serving as the common ground. EEG signals were digitised at 5 kHz (filtering: DC-1000 Hz) and EEG electrode impedance was kept below 5 kΩ.

EEG signals with 256 channels captured from 11 subjects executing a SSVEP-based experimental protocol. Five different frequencies (6.66, 7.50, 8.57, 10.00 and 12.00 Hz) have been used for the visual stimulation, and the EGI 300 Geodesic EEG System (GES 300), using a 256-channel HydroCel Geodesic Sensor Net (HCGSN) and a sampling rate of 250 Hz has been used for capturing the signals. Check https://github.com/MAMEM/ssvep-eeg-processing-toolbox for the processing toolbox.

Electroencephalogram (EEG) and behavioural data (joystick) was collected from 15 healthy participants who completed a modified version of a Go/No-go task. This dataset consists of raw data, pre-processed EEG and behavioural data, along with impulsivity scores in a .csv file. The pre-processed data is in MATLAB .mat format.

Raw Data

The EEG and behavioural data (Joystick) along with trigger data was collected using a TMSi Porti amplifier with a sampling rate of 2,048Hz and is in .een format. The raw EEG files contain brain activity recorded in the first 16 channels and last 2 channels (channels 17 and 18) correspond to Joystick and Trigger information (used to identify the type of event – Go/Conflict/NoGo) respectively.

The Raw data is segregated into 2 folders- Active and Sham which is further divided into baseline and after stimulation conditions.

The main behavioural outcome is the change in NoGo errors (pre-processed folder- Figure 1C in from the article ‘Tuning the brakes – Modulatory role of transcranial random noise stimulation on inhibition,’ Brain Stimulation, 2024), comparing baseline and after-stimulation in sham and active conditions. Metadata corresponding to impulsivity scores and the change in NoGo behaviour are provided in ‘UPPS_nogo.csv’ (used for Figure 1D). The EEG data was recorded while the participants completed the task during baseline and after stimulation, and was used to calculate the spectral power (Figure 1E). The study also presents intermittent bursts from the EEG data, comparing the average burst durations at baseline and after-stimulation (Figure 1F) in sham and active stimulation conditions.

Code

All data were analysed in MATLAB (2018b) using a combination of EEGLAB, ERP LAB and FieldTrip packages.

Installation guides can be found on
https://sccn.ucsd.edu/eeglab/index.php
https://matlab.mathworks.com/ https://erpinfo.org/erplab  https://www.fieldtriptoolbox.org/download/

The behavioural data plots use the software IOSR toolbox : https://github.com/IoSR-Surrey/MatlabToolbox

Code_figure_IC.m: This script plots the NoGo error rates in baseline and after stimulation in sham and active conditions. This script uses the mat file ‘Nogo_behav_pre_post.mat’

Code_figure_1E.m: This script plots the spectral power and grand average of the after-stimulation EEG data with clusters obtained from a non-parametric analysis. This script uses the mat file ‘data_psd_trns_pre_post.mat’.

Code_figure_1F.m: This script plots the intermittent burst durations during sham and active conditions and uses the file ‘Nogo_bursts_pre_post.mat’

Description

We collected EEG signal data from 10 college students while they watched MOOC video clips. We extracted online education videos that are assumed not to be confusing for college students, such as videos of the introduction of basic algebra or geometry. We also prepare videos that are expected to confuse a typical college student if a student is not familiar with the video topics like Quantum Mechanics, and Stem Cell Research. We prepared 20 videos, 10 in each category. Each video was about 2 minutes long. We chopped the two-minute clip in the middle of a topic to make the videos more confusing. The students wore a single-channel wireless MindSet that measured activity over the frontal lobe. The MindSet measures the voltage between an electrode resting on the forehead and two electrodes (one ground and one reference) each in contact with an ear. After each session, the student rated his/her confusion level on a scale of 1-7, where one corresponded to the least confusing and seven corresponded to the most confusing. These labels if further normalized into labels of whether the students are confused or not. This label is offered as self-labelled confusion in addition to our predefined label of confusion.

Content

These data are collected from ten students, each watching ten videos. Therefore, it can be seen as only 100 data points for these 12000+ rows. If you look at this way, then each data point consists of 120+ rows, which is sampled every 0.5 seconds (so each data point is a one minute video). Signals with higher frequency are reported as the mean value during each 0.5 second.

EEG_data.csv: Contains the EEG data recorded from 10 students

demographic.csv: Contains demographic information for each student

video data : Each video lasts roughly two-minute long, we remove the first 30 seconds and last 30 seconds, only collect the EEG data during the middle 1 minute.

Acknowledgements

• Wang, H., Li, Y., Hu, X., Yang, Y., Meng, Z., & Chang, K. M. (2013, June). Using EEG to Improve Massive Open Online Courses Feedback Interaction. In AIED Workshops. [PDF]

The data is collected from a software that we implemented ourselves. Check HaohanWang/Bioimaging for the source code.

Inspiration

• This dataset is an extremely challenging data set to perform binary classification. Here are some recent classification results for reference:

  • SVM: 67.2%
  • Bi-LSTM (Ni et al 2017): 73.3%
  • CF-Bi-LSTM (Wang et al 2018): 75.0%

• It is an interesting data set to carry out the variable selection (causal inference) task that may help further research. Past research has indicated that Theta signal is correlated with confusion level.

• It is also an interesting data set for confounding factors correction model because we offer two labels (subject id and video id) that could profoundly confound the results.

Other Resources

Promotion Video

Source Code of Data Collection Software

Contact

Haohan Wang

@HaohanWang

The dataset contains EEG signals from 11 subjects with labels of alert and drowsy. It can be opened with Matlab. We extracted the data for our own research purpose from another public dataset:Cao, Z., et al., Multi-channel EEG recordings during a sustained-attention driving task. Scientific data, 2019. 6(1): p. 1-8.If you find the dataset useful, please give credits to their works. The details on how the data were extracted are described in our paper:"Jian Cui, Zirui Lan, Yisi Liu, Ruilin Li, Fan Li, Olga Sourina, Wolfgang Müller-Wittig, A Compact and Interpretable Convolutional Neural Network for Cross-Subject Driver Drowsiness Detection from Single-Channel EEG, Methods, 2021, ISSN 1046-2023, https://doi.org/10.1016/j.ymeth.2021.04.017."The codes of the paper above are accessible from:https://github.com/cuijiancorbin/A-Compact-and-Interpretable-Convolutional-Neural-Network-for-Single-Channel-EEGThe data file contains 3 variables and they are EEGsample, substate and subindex."EEGsample" contains 2022 EEG samples of size 20x384 from 11 subjects. Each sample is a 3s EEG data with 128Hz from 30 EEG channels."subindex" is an array of 2022x1. It contains the subject indexes from 1-11 corresponding to each EEG sample."substate" is an array of 2022x1. It contains the labels of the samples. 0 corresponds to the alert state and 1 correspond to the drowsy state.The unbalanced version of this dataset is accessible from:https://figshare.com/articles/dataset/EEG_driver_drowsiness_dataset_unbalanced_/16586957

Dataset descriptionThis dataset contains EEG signals from 73 subjects (42 healthy; 31 disabled) using an ERP-based speller to control different brain-computer interface (BCI) applications. The demographics of the dataset can be found in info.txt. Additionally, you will find the results of the original study broken down by subject, the code to build the deep-learning models used in 1 and a script to load the dataset.Original article:[1] Santamaría-Vázquez, E., Martínez-Cagigal, V., Vaquerizo-Villar, F., & Hornero, R. (2020). EEG-Inception: A Novel Deep Convolutional Neural Network for Assistive ERP-based Brain-Computer Interfaces. IEEE Transactions on Neural Systems and Rehabilitation Engineering. https://doi.org/10.1109/TNSRE.2020.3048106Some people report problems to register and use IEEE Dataport. Additional sources:Official code repository: https://github.com/esantamariavazquez/EEG-InceptionDataset copy in kaggle: https://www.kaggle.com/esantamaria/gibuva-erpbci-dataset

Summary:

This dataset contains electroencephalographic (EEG) recordings of 38 subjects playing in pair to the multi-user version of a visual P300-based Brain-Computer Interface (BCI) named Brain Invaders (Congedo et al., 2011). The interface uses the oddball paradigm on a grid of 36 symbols (1 Target, 35 Non-Target) that are flashed pseudo-randomly to elicit a P300 response, an evoked-potential appearing about 300ms after stimulation onset. EEG data were recorded using 32 active wet electrodes per subjects (total: 64 electrodes) during three randomized conditions (Solo1, Solo2, Collaboration). The experiment took place at GIPSA-lab, Grenoble, France, in 2014. A full description of the experiment is available at https://hal.archives-ouvertes.fr/hal-02173958. Python code for manipulating the data is available at https://github.com/plcrodrigues/py.BI.EEG.2014b-GIPSA. The ID of this dataset is bi2014b.

Full description of the experiment and dataset: https://hal.archives-ouvertes.fr/hal-02173958

Investigators: Eng. Louis Korczowski, B. Sc. Ekaterina Ostaschenko

Technical Support: Eng. Anton Andreev, Eng. Grégoire Cattan, Eng. Pedro. L. C. Rodrigues, M. Sc. Violette Gautheret

Scientific Supervisor: Ph.D. Marco Congedo

ID of the dataset: bi2014b

This dataset combines multimodal biosignals and eye tracking information gathered under a human-computer interaction framework. The dataset was developed in the vein of the MAMEM project that aims to endow people with motor disabilities with the ability to edit and author multimedia content through mental commands and gaze activity. The dataset includes EEG, eye-tracking, and physiological (GSR and Heart rate) signals along with demographic, clinical and behavioral data collected from 36 individuals (18 able-bodied and 18 motor-impaired). Data were collected during the interaction with specifically designed interface for web browsing and multimedia content manipulation and during imaginary movement tasks. Alongside these data we also include evaluation reports both from the subjects and the experimenters as far as the experimental procedure and collected dataset are concerned. We believe that the presented dataset will contribute towards the development and evaluation of modern human-computer interaction systems that would foster the integration of people with severe motor impairments back into society.Please use the following citation: Nikolopoulos, Spiros, Georgiadis, Kostas, Kalaganis, Fotis, Liaros, Georgios, Lazarou, Ioulietta, Adam, Katerina, Papazoglou – Chalikias, Anastasios, Chatzilari, Elisavet , Oikonomou, Vangelis P., Petrantonakis, Panagiotis C., Kompatsiaris, Ioannis, Kumar, Chandan, Menges, Raphael, Staab, Steffen, Müller, Daniel, Sengupta, Korok, Bostantjopoulou, Sevasti, Zoe, Katsarou , Zeilig, Gabi, Plotnik, Meir, Gottlieb, Amihai, Fountoukidou, Sofia, Ham, Jaap, Athanasiou, Dimitrios, Mariakaki, Agnes, Comanducci, Dario, Sabatini, Edoardo, Nistico, Walter & Plank, Markus. (2017). The MAMEM Project - A dataset for multimodal human-computer interaction using biosignals and eye tracking information. Zenodo. http://doi.org/10.5281/zenodo.834154Read/analyze using the following software:https://github.com/MAMEM/eeg-processing-toolbox

The following author contributed equally to this dataset: Accou, Bernd; Bollens, Lies. Researchers investigating the neural mechanisms underlying speech perception often employ electroencephalography (EEG) to record brain activity while participants listen to spoken language. The high temporal resolution of EEG enables the study of neural responses to fast and dynamic speech signals. Previous studies have successfully extracted speech characteristics from EEG data and, conversely, predicted EEG activity from speech features. Machine learning techniques are generally employed to construct encoding and decoding models, which necessitate a substantial amount of data. We present SparrKULee: A Speech-evoked Auditory Repository of EEG, measured at KU Leuven, comprising 64-channel EEG recordings from 85 young individuals with normal hearing, each of whom listened to 90-150 minutes of natural speech. This dataset is more extensive than any currently available dataset in terms of both the number of participants and the amount of data per participant. It is suitable for training larger machine learning models. We evaluate the dataset using linear and state-of-the-art non-linear models in a speech encoding/decoding and match/mismatch paradigm, providing benchmark scores for future research. Our github repository contains the necessary code to perform preprocessing steps needed to obtain the files in the derivatives folder, as well as extra code to show the technical validation of our dataset and tools to download the dataset more easily. This link provides a download of the whole dataset in one big zip file ( > 100GB) . For a download of the dataset using already zipped files, split up into smaller chunks, click here. Due to privacy concerns, there are some restricted files in the dataset. Users requesting access should send a mail to sparrkulee@kuleuven.be , stating what they want to use the data for. Access will be granted to non-commercial users, complying to the CC-BY-NC-4.0 licence

This dataset contains the experimentally collected data used for the paper: Shin H, Suma D and He B (2022) Closed-loop motor imagery EEG simulation for brain-computer interfaces. Front. Hum. Neurosci. 16:951591. doi:10.3389/fnhum.2022.951591

The data was recorded from 10 unique healthy human subjects (assigned subject numbers S01 to S10). All data is organized into LIVE and SIMULATED directories, followed by BW, NT and CV subdirectories according to which experimental condition the data came from. For details on the live vs. simulated parallel experiment design and the implementation of BW, NT and CV parameters, please refer to the paper. All identifying information have been removed.

Please cite the above paper if you use any data included in this dataset. Code related to this study are also available from https://github.com/mcvain/bci-simulator.