Brain-Computer Gaming Control through Imagined Speech Commands Dataset
Description
Twenty-six healthy right-handed volunteers (19 males and 7 females, aged 20.1 ± 1.09, range 18–24) participated in the experiment. All participants were right-handed, native Spanish speakers from Mexico, and had no challenges in speech or language production. None had clinically diagnosed attention deficit disorders or physical impairments. The study was approved by Tecnológico de Monterrey and performed under the Helsinki Declaration. All volunteers signed an informed consent before the acquisitions, and their privacy was guaranteed through personal data anonymization.
Files
Steps to reproduce
The data are available in "Raw" form, without preprocessing, and in "Cleaned" with a filter from 0.5 to 40 Hz and manual epoch rejection. For an epoch to be visually rejected, it would have to contain (1) muscle artifacts, (2) eye artifacts like movement or blinks, or (3) too much noise, in comparison to other epochs in the recording. Each .mat file contains the tridimensional array in the form of Channels x Times x Trials along with the labels of the trials. The same information is extended for Matlab in one .mat for the data array and another for the labels in a folder called “Matlab Extended Version”. A Matlab script is provided to load the extended version into EEGLAB. Additionally, the repository includes two spreadsheets: one named "EEG Quality", which reports the number of trials each subject and game mode has, along with an informative channel quality opinion, and the other named "Forms Answers", containing all the questions in their original Spanish and the subjects' answers. The numerical order of the electrodes coincides with the order presented in the file "channel_locations_braincommand.txt". A PDF file containing the customized initial and wellness check-in questionnaires in English was provided. The remaining questionnaires were sourced from The Internal Representations Questionnaire, Edinburgh Handedness Inventory Questionnaire, Flow State Questionnaire, and Sense of Agency Scale. The code for the classification is available at https://github.com/AlmaCuevas/EEG-Classifiers-Ensemble, and the evaluated BCI game is at https://github.com/AlmaCuevas/BrainCommand. This work was supported by SECIHTI, Tecnológico de Monterrey and the NeuroTechs research group.
Institutions
Categories
Funders
- Secretaría de Ciencia, Humanidades, Tecnología e Innovación