EEGs from healthy motor control during neurofeedback training

He S
Everest-Phillips C
Brown P
Tan H
Description

The EEG data were recorded from 20 human volunteers (10 females) while they were performing a sequential neurofeedback-behaviour task, with the neurofeedback reflecting the occurrence of beta bursts over sensorimotor cortex (C3 or C4) quantified in real time.

Each participant was recorded three times over three different days. On each recording day, the participant performed the neurofeedback training task with each hemisphere using the EEG signals recorded from C3 or C4 (in a random order), and the contralateral hand for the motor task. Participants completed four experimental runs for each hemisphere on each training day. Each experimental run consisted of a 30 s of rest recording for calibration, 10 continuous trials in the training condition, and another 10 continuous trials in the no training condition. The order of training and no training blocks in each experimental run was randomized. In total, we recorded data from 20 hemispheres with 120 trials in each of the 'training' and 'no training' conditions for each hemisphere for each of the real feedback and sham feedback groups.

The details of the experimental design and behavioural task are described in He et al (2020).

The data files are in MATLAB format. The dataset consists of 120 raw data files (20 subject × 3 days × 2 hemispheres). Due to the size of this dataset (38 GB), it has been split into segments. However, you may still trouble downloading it, in which case please contact ben.micklem@bndu.ox.ac.uk.

Task

The participants were pseudo-randomly assigned to a sham feedback group or a real feedback group, with ten participants in each group. The neurofeedback training composed of multiple short trials. Each trial consisted of a 2 s period where the participants were instructed to get ready and a 4 s neurofeedback phase, which was followed by black screen presented for a time randomly drawn between 2 and 3 s and then a movement go-cue. The participants were instructed to perform a thumb of finger pinch movement as fast as possible in response to the go-cue to generate a force overshooting a predefined force level (50% of the maximum voluntary force measured before starting the task).

Instructions given to the participants were the same for both groups. In the training trials, the participants were instructed to keep the basketball floating at the top of the screen, which would require them to suppress the beta bursts. In the no training trials, the participants were instructed to simply pay attention to the movement of the ball displayed on the screen and get ready for the subsequent movement go-cue.

Each participant was recorded three times over three different days. On each recording day, the participant performed the neurofeedback training task with each hemisphere using the EEG signals recorded from C3 or C4 (in a random order), and the contralateral hand for the motor task. Participants completed four experimental runs for each hemisphere on each training day. Each experimental run consisted of a 30 s of rest recording to calibrate the threshold for triggering the vertical movement of the basketball, 10 continuous trials in the training condition, and another 10 continuous trials in the no training condition. The order of training and no training blocks in each experimental run was randomized. In total, we recorded data from 20 hemispheres with 120 trials in each of the training and no training conditions for each hemisphere for each of the real feedback and sham feedback groups.

Group information

Real feedback group: Subject - 1 4 6 7 10 11 13 15 17 20
Sham feedback group: Subject - 2 3 5 8 9 12 14 16 18 19

Data description

For each subject (e.g., Sub1), there are three subfolders (i.e., Day1, Day2, Day3) with two files (i.e., Raw_C3.mat and Raw_C4.mat) in each subfolder. The data are in Matlab format. In each .mat file, there are five fields indicate five data streams recorded using OpenVibe. The name of each data stream can be found in rawData.info.name.

The data stream 'openvibeSignal' contains the 32 channels raw data, including:
*** 24 monopolar EEG, channel 1-24: FP1, FP2, Fz, FCz, Cz, CPz, Pz, Oz, FC1, C1, CP1, FC3, C3, CP3, FC2, C2, CP2, FC4, C4, CP4, P3, P4, O1, O2.
*** 2 bipolar EMG from the flexor carpi radialis of both arms, channel 25-26: EMGL, EMGR.
*** 2 accelerometer measurements for both hands recorded from z-axis: Aclz, Acrz.
*** 2 pinch force: FrcL, FrcR.
    Note that these channel information were not included in the structure, but they were the same as indicated above for all recordings.

The data stream 'TriggerStream' contains the trigger information during the experiment. Here below are some keys triggers used to segment the trials:
*** 100 or 101: The start of a block, with 100 and 101 indicating no-training and training conditions, respectively.
*** 1-10: The onset of the basketball movement in trial 1-10. Note that there were 10 trials in each block.
*** -2: Trigger for the participants to get ready before the basketball movement.
*** 14: Tigger for the pinch task.

The data stream 'BallMove' contains the recorded positions of the basketball for each individual trial, which indicated the neurofeedback training performance.

The other two streams including 'openvibeMarkers' and 'Matlab' could be ignored.
In each data stream, the variable 'time_series' indicated the raw data and 'time_stamps' indicated the time stamps for each sample (column) in 'time_series'.
 
The following script was used to match the time stamps between 'openvibeSignal' and other two streams:

rawData{1,1}.time_stamps = rawData{1,1}.time_stamps+str2double(rawData{1,1}.info.created_at)
Here we assume rawData{1,1} was the 'openvibeSignal' data stream.

The sampling frequency was 2048 Hz.

The peak frequencies with maximum movement-related power reduction was:
BetaC3 = [15 19 15 20 18 17 22 19 18 16 20 22 20 18 24 19 20 22 23 19];
BetaC4 = [15 19 15 18 16 21 18 19 17 18 19 20 22 24 19 18 21 20 23 18];
For each hemisphere, a 5-Hz frequency band around the peak frequency was used for the neurofeedback training.

A sequential neurofeedback-behaviour task, with the neurofeedback reflecting the occurrence of beta bursts quantified in real-time based on EEG measurements over sensorimotor cortex, was used to evaluate the relationship between cortical beta bursts and movement initialisation.
Year Published
2020
DOI
10.5287/bodleian:9gM209oXo
Funders & Grant Numbers
Medical Research Council UK (MR/P012272/1)
Medical Research Council UK (MC_UU_12024/1)
National Institute for Health Research Oxford Biomedical Research Centre
Rosetrees Trust
Publisher
University of Oxford
First Published In
He S, Everest-Phillips C, Clouter A, Brown P, Tan H
2020. J. Neurosci., 40(20):4021-4032.
Group
Tan Group
Terms

Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) This is a human-readable summary of (and not a substitute for) the licence. You are free to: Share — copy and redistribute the material in any medium or format Adapt — remix, transform, and build upon the material for any purpose, even commercially. This licence is acceptable for Free Cultural Works. The licensor cannot revoke these freedoms as long as you follow the license terms. Under the following terms: Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use. ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same licence as the original. No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the licence permits.