Datasets
Standard Dataset
Synthetic Epileptic Spike EEG Database (SESED-WUT)
- Citation Author(s):
-
MarcinKolodziej
Warsaw University of TechnologyAndrzejMajkowskiWarsaw University of TechnologyMarcinJurczakWarsaw University of Technology - Submitted by:
- Marcin Kolodziej
- Last updated:
- Tue, 01/14/2025 - 06:51
- DOI:
- 10.21227/y5ge-nj12
- Data Format:
- License:
50 ViewsAbstract
Synthetic Epileptic Spike EEG Database (SESED-WUT)
The database contains EEG, EMG, and EOG signals with artificially generated epileptic spikes. The recordings were performed using the g.USBamp 2.0 amplifier. Data were collected from 5 EEG channels (C3, Cz, C4, Fz, Fp1), 1 EOG channel (VEOG), and 3 EMG channels (Nape, Cheek, Jaw). The signals were sampled at 256 Hz and processed with a bandpass filter (0.1–100 Hz) and a notch filter (48–52 Hz).
Participants were instructed voluntarily and as accurately as possible to imitate various tasks resembling artifacts observed during seizures. These tasks included clenching the jaw, squeezing the eyes shut, moving the cheeks, raising the eyebrows, furrowing the forehead, performing abrupt head movements, and contracting neck muscles. Each task involved brief activation lasting 0.5–1 second, followed by relaxation for 2–4 seconds, while participants sat comfortably with their eyes open. Each recording session lasted 3 minutes, capturing signals with prominent high-amplitude artifacts and a broad frequency spectrum. These artifacts interfered with EEG signals and mimicked pathological changes observed during epileptic seizures.
Artificially generated epileptic spikes, based on triangular signals, were added to the Cz channel in the presence of artifacts. The database contains signals from 20 participants (5 women, 15 men) aged 19–42 years. This resulted in 586 recordings with artificial spikes.
The database includes 686 .mat files (Matlab format).
File Interpretation Example
File S01-1.mat:
- S01: User number 01,
- 1: Example number (out of 686 total examples).
The file naming convention indicates the user (e.g., S01) and the unique example number within the entire database, ranging sequentially from 1 to 686. This structure ensures clarity and consistency across all files.
Each file contains recorded EEG, EMG, and EOG signals, along with samples featuring artificially added epileptic spikes. Each sample has a duration of 5 seconds, corresponding to 1280 samples at a 256 Hz sampling rate.
Variables in the File
- signals:
- A matrix of dimensions 7 x 1280, containing the recorded signals from various electrodes.
- Signal order in the signals matrix:
- Cz – EEG signal with an added artificial spike,
- Fp1 – EEG signal,
- HEOG – EOG signal (horizontal eye movement),
- Nape – EMG signal from the nape,
- Cheek – EMG signal from the cheek,
- Jaw – EMG signal from the jaw,
- Spike – Artificially generated epileptic spike.
- Sampling frequency: 256 Hz (data type: double).
- A list of electrode names corresponding to the signals in the signals matrix.
- Data type: cell, containing the names: {'Cz', 'Fp1', 'HEOG', 'Nape', 'Cheek', 'Jaw', 'Spike'}.
Data Characteristics
- All signals are stored in the signals array in units corresponding to their type (EEG, EMG, or EOG).
- Spike represents artificially added epileptic spikes to simulate pathological disturbances in the EEG channel.
- Each sample contains 1280 data points, representing 5 seconds of recorded signals.
The file is structured to facilitate the analysis of both raw signals and their modifications with artificial components, making it a valuable resource for studies on artifact detection and epileptic spike analysis.
