Seizure detection using ultra‐long‐term subcutaneous electroencephalography: A deep learning <scp>CNN</scp> ‐ <scp>BiLSTM</scp> approach
High sensitivity and specificity are possible in automated seizure detection in two-channel subscalp EEG data using a CNN-BiLSTM framework. Performance of the detector is superior using subcutaneous EEG data for training rather than intracranial EEG, but addition of scalp EEG seizures for training was helpful.
Authors
- Václav Křemen (ORCID: https://orcid.org/0000-0001-9844-7617)
- Sihyeong Park (ORCID: https://orcid.org/0000-0002-6371-265X)
- Pedro F. Viana (ORCID: https://orcid.org/0000-0003-0861-8705)
- Nicholas M. Gregg (ORCID: https://orcid.org/0000-0002-6151-043X)
- Mark P. Richardson (ORCID: https://orcid.org/0000-0001-8925-3140)
- Benjamin H. Brinkmann (ORCID: https://orcid.org/0000-0002-2392-8608)
- Gregory A. Worrell (ORCID: https://orcid.org/0000-0003-2916-0553)
- Jonas Duun‐Henriksen (ORCID: https://orcid.org/0000-0003-1558-8225)
- Jie Cui (ORCID: https://orcid.org/0000-0003-1000-8869)
- James S. Clark (ORCID: https://orcid.org/0000-0002-5677-9733)
- Jayawant N. Mandrekar (ORCID: https://orcid.org/0000-0002-9718-4082)
Institutions
- Mayo Clinic (US)
- King's College London (GB)
- Hospital de Egas Moniz (PT)
- Vivostat (Denmark) (DK)
- King's College Hospital NHS Foundation Trust (GB)
- Mayo Clinic in Florida (US)
Publication Details
- Journal
- Epilepsia
- Published
- 2025-10-07
- DOI
- https://doi.org/10.1111/epi.18652
- Primary Topic
- EEG and Brain-Computer Interfaces
- Type
- article
- Field-Weighted Citation Impact
- 0.00