EEG-Based Dataset Explicitly Targets the Transitions between Sitting and Standing for Exploring Neural Activation Patterns in Motor Imagery and Execution

EEG-Based Dataset Explicitly Targets the Transitions between Sitting and Standing for Exploring Neural Activation Patterns in Motor Imagery and Execution Benjakarn Uengsawapak, Supavit Kongwudhikunakorn, Suktipol Kiatthaveephong, Wipamas Polpakdee, Rattanaphon Chaisaen, Poramate Manoonpong, Chanitsada Chuenchit, Gun Bhakdisongkhram and Theerawit Wilaiprasitporn Abstract This study presents the first publicly accessible electroencephalography (EEG) dataset explicitly targeting sit-to-stand and stand-to-sit transitions during both motor execution (ME) and motor imagery (MI) tasks. Twenty-two healthy participants performed sitting and standing transitions under well-controlled experimental conditions while 60-channel EEG, electrooculography (EOG), and electromyography (EMG) signals were synchronously recorded. The dataset enables the exploration of neural activation patterns associated with lower-limb movements and supports the development of EEG-based brain–computer interface (BCI) algorithms for mobility assistance and rehabilitation. To validate the dataset, a benchmark classification was conducted using EEGNet, a compact convolutional neural network. Results demonstrated consistent decoding performance with mean accuracies of approximately 80\% for ME and 70\% for MI, indicating the reliability and usability of the dataset. Additionally, analyses of movement-related cortical potentials (MRCPs) and event-related desynchronization/synchronization (ERD/ERS) patterns revealed distinct neural signatures across the transition phases. This dataset provides a comprehensive foundation for studying lower-limb motor control, neural dynamics, and the advancement of MI-based BCIs for rehabilitation and assistive technologies. Dataset Description Participants Twenty-three healthy participants (aged 22–28 years; fifteen males) with no known neurophysiological abnormalities were recruited for this study. One participant (S05) was excluded due to poor signal quality, resulting in a final cohort of twenty-two participants. Prior to the experiment, research staff provided both verbal and written explanations of the study objectives, protocol, questionnaire, and experimental setup to ensure participants' comprehension. 1 The demographics of the subjects are in the table below. Subject Code Age Gender Dominant side EEG Experience Remark S01 26 M right yes S02 23 F right yes S03 26 M right yes S04 24 M right no S05 26 M right no Excluded S06 25 M right no S07 25 M right no S08 26 F right no S09 23 F right no S10 23 F right yes S11 18 F right no S12 30 M right no S13 22 F right no S14 23 M right no S15 26 M both yes S16 25 M right no S17 23 M right no S18 26 M right no S19 25 M left no S20 27 M right no S21 28 F right no S22 28 M right yes S23 24 M left no 2 Data Collection Physiological signals, including 60 EEG, 2 EOG, and 6 EMG, were obtained from the participants. EEG and EOG electrodes are recorded with 1200 Hz sampling rate and their placements are illustrated in the figures below. The hEOG is placed at the right temple, while the vEOG is placed at the right infra orbital. EMG data were recorded with a sampling rate of 2000 Hz. Each of 3 sensors was attached to the left and right legs, targeting the following muscles: Soleus (SL), Tibialis Anterior (TA), and Rectus Femoris (RF). The EMG sensors’ placement and sequence are informed in the figures and tables below. Electrode placement positions for recording EEG and EOG signals. Channel Index Channel Name Channel Index Channel Name Channel Index Channel Name Channel Index Channel Name 1 fp1 17 f5 33 cpz 49 p7 2 fp2 18 f6 34 pz 50 p8 3 af7 19 fcz 35 cp1 51 poz 4 af8 20 cz 36 cp2 52 oz 5 f7 21 fc1 37 cp3 53 po3 6 f8 22 fc2 38 cp4 54 po4 7 ft7 23 fc3 39 cp5 55 po7 8 ft8 24 fc4 40 cp6 56 po8 9 af3 25 fc5 41 tp7 57 po9 10 af4 26 fc6 42 tp8 58 po10 11 afz 27 c1 43 p1 59 o1 12 fz 28 c2 44 p2 60 o2 13 f1 29 c3 45 p3 61 hEOG 14 f2 30 c4 46 p4 62 vEOG 15 f3 31 c5 47 p5 63 trigger 16 f4 32 c6 48 p6 EEG and EOG channel indices and channel names description. 3 Surface EMG sensor placement positions for recording the EMG signal. Channel Index Channel Name Location EMG-1 sl_l left Soleus muscle EMG-2 sl_r right Soleus muscle EMG-3 ta_l left Tibialis Anterior muscle EMG-4 ta_r right Tibialis Anterior muscle EMG-5 rf_l left Rectus Femoris muscle EMG-6 rf_r right Rectus Femoris muscle EMG channel indices, channel names, and placement locations. Data/Files Format Two versions of the dataset are provided. Raw data, in .mat format, allows flexible modification and further research exploration; while, preprocessed data, in .fif format, undergoes data preprocessing steps, including bandpass filtering and downsampling for enhancing reproducibility, improving data quality, and reducing computational load. Details are as follows. Raw Dataset The raw EEG and EOG data for each participant are structured as a 2-D matrix of dimension n_channels ╳ n_timepoints. An extra channel (the 63rd channel) is allocated for event triggers, which annotate task-related events, in which the details are presented below. 4 Event Trigger Number Description 1 Eyes closed in the resting state. 2 Eyes opened in the resting state. 10 Start of trials in ME activity. 11 Start of ME_SIT_STD in ME activity. 12 Start of ME_STD_SIT in ME activity. 13 Start of resting condition (ME_R) in ME activity. 20 Start of trials in MI during sit. 21 Start of MI_SIT_STD in MI during sit. 22 Start of MI_SIT_SIT task in MI during sit. 23 Start of resting condition in MI during sit (MI_R_SIT, rest while sitting). 30 Start of trials in MI during stand. 31 Start of MI_STD_STD task in MI during stand. 32 Start of MI_STD_SIT in MI during stand. 33 Start of resting condition in MI during stand (MI_R_STD, rest while standing) Note: Trigger 0 indicates that no event occurred. The participant is performing the corresponding task. Event trigger number description table. The event trigger is stored at EEG channel number #63 in the .mat file and embedded into MNE Epochs from the .fif file. Dictionary of Attribute Definitions Attribute definitions are in the table below. Attribute Type Description Units eeg float array EEG data μV emg float array EMG data V eeg_ts float array EEG timestamps Seconds emg_ts float array EMG timestamps Seconds eeg_channels string array List of EEG channels, EOG channels, and trigger - emg_channels string array List of EMG channels - eeg_fs int array EEG sampling rate Hz emg_fs int array EMG sampling rate Hz 5 Mini-Tutorial Step-By-Step for the Raw Dataset import scipy.io as sio # Load .mat file data = sio.loadmat('raw_dataset.mat') # List variables in the loaded file print(data.keys()) # Example: check EEG data eeg = data['eeg'] print(eeg.shape, eeg.dtype) # Check timestamps eeg_ts = data['eeg_ts'] print(eeg_ts.shape, eeg_ts.dtype) # EEG first 5 samples print(eeg[:, :5]) # EEG timestamps first 5 samples print(eeg_ts[:5]) Preprocessed Dataset The preprocessed EEG data is organized as a single three-dimensional matrix with the dimensions n_trials × n_channels × n_timepoints. This matrix contains the complete dataset, encompassing all movement tasks from all recording sessions. Each trial represents an activity segment aligned with a trigger marker. The data for specific movement tasks can be accessed using their task names (i.e. me_sit_std, me_r_sit, me_std_sit, me_r_std, mi_sit_std, mi_r_sit, mi_std_sit, mi_r_std), and the trials from a desired session can be accessed using the index range specified in the table below. 6 Activity Session Round Index Range (start_idx:end_idx) ME 1 1 [0:20, :, :] 2 1 [20:40, :, :] MI 1 1 [0:10, :, :] 2 [10:20, :, :] 2 1 [20:30, :, :] 2 [30:40, :, :] Mini-Tutorial Step-By-Step for the Preprocessed Dataset import mne # Load epoched .fif file epochs = mne.read_epochs('MI_S01.fif', preload=True) #inspect epoch information print(epochs) # overview of epochs #access data data = epochs.get_data() # shape: (n_epochs, n_channels, n_times) print(data.shape) #select specific event ## Select only 'mi_sit_std' task epochs_mi_sit_std = epochs['mi_sit_std'] print(epochs_mi_sit_std) #select specific session ## select all trials from session 1 session1_mi_sit_std_epochs = epochs_mi_sit_std[0:20,:,:] Folder Structure Dataset/ │ ├── readme.pdf # Data description document │ └── v1_raw_S01.zip # Original/raw data for subject 1 (each subject and session) │ ├── S01_S1.mat # subject 1 session 1 │ └── S01_S2.mat # subject 1 session 2 │ └── v1_raw_S02.zip 7 │ ├── S02_S1.mat │ └── S02_S2.mat └── … │ └── v1_raw_S05.zip # S05 Excluded from dataset └── … │ └── v1_raw_S23.zip │ ├── S23_S1.mat │ └── S23_S2.mat │ └── v2_processed_all.zip # Preprocessed/cleaned data ├── ME/ # Motor Execution, S01–S23 (22 subjects, exclude S05) │ ├── S01.fif │ ├── S02.fif │ └── … │ └── S23.fif └── MI/ # Motor Imagery, S01–S23 (22 subjects, exclude S05) ├── S01.fif ├── S02.fif └── … └── S23.fif File Types: .mat (raw data) and .fif (preprocessed data) File Naming Convention: 1. Raw dataset: S<ID>_S<session num>.mat 2. Preprocessed dataset: S<ID>.fif Variable Definitions: ID = Subject ID session num = Session number 8 File size: The dataset is distributed as compressed .zip files in two categories: 1. raw data: Provided as separate .zip files for each subject, including two sessions. Example: v1_raw_S01.zip contains raw data for subject 1 (~ 2 GB per subject). For all 22 subjects (S01–S23, excluding S05): approximately 44.12 GB 2. preprocessed data: Provided as a single compressed .zip file, v2_processed_all.zip (~3.77 GB), including ME and MI data from all 22 subjects (S01–S23, excluding S05). Data Availability The raw and preprocessed data are available via the following DOIs in the open-access online repository, Zenodo (https://zenodo.org). ● raw data: DOI:10.5281/zenodo.17561969 ● preprocessed data: DOI:10.5281/zenodo.17629950 9