TABLE 4.
A short description of preprocessing steps undergone by the publicly available datasets and its application.
| Dataset | Description | Application |
|---|---|---|
| University of Bonn | Artifacts removal | Seizure detection |
| CHB‐MIT Scalp EEG | ‐ | Seizure detection/Prediction |
| Melbourne‐NeuroVista seizure trial (NeuroVista Ictal) | Selective segmentation (ictal) | Seizure detection/Prediction |
| Kaggle UPenn and Mayo Clinic's Seizure Detection Challenge | Selective segmentation:1 h from any seizure for interictal and 4 h from any seizure events for ictal | Seizure detection |
| Kaggle American Epilepsy Society Seizure Prediction Challenge | Selective segmentation: 5 min before seizure onset and 4 h from any seizure events for preictal, 1 wk minimum from seizure events for interictal | Seizure prediction |
| Neurology and Sleep Centre Hauz Khas | Frequency filtering between 0.5 and 70 Hz | Seizure detection/Prediction |
| Kaggle Melbourne‐University AES‐MathWorks‐NIH Seizure Prediction Challenge Data | Selective segmentation: 5 min before seizure onset and 4 h from any seizure events for preictal, 3 h before and 4 h after seizure for interictal. | Seizure prediction |
| TUH EEG Seizure Corpus (TUSZ) | 1 h pruned segment. | Seizure detection/Prediction |
| Helsinki University Hospital EEG | ‐ | Seizure detection/Prediction |
| Siena Scalp EEG | ‐ | Seizure detection/Prediction |