Reducing False Alarms in Seizure Prediction: A Specialized CNN Architecture and a Novel EEG Sampling Technique

• Published in: International Journal of Computing and Digital System (Jāmiʻat al-Baḥrayn. Markaz al-Nashr al-ʻIlmī) Vol. 17; no. 1; pp. 1 - 14
• Main Authors: Hayder, Ameer, Ismael, Haider, Hameed, Arwa
• Format: Journal Article
• Language: English
• Published: University of Bahrain, Deanship of Graduate Studies and Scientific Research 01.01.2025
• Subjects: Architectural design; Electroencephalography; Epilepsy; Methods; Neural networks; Seizures (Medicine)
• ISSN: 2210-142X; 2535-9886; 2210-142X
• DOI: 10.12785/ijcds/1571046226

Deep learning played a vital role in the seizure prediction challenge. However, most studies used generic architectures that fail to consider the distinct characteristics of multivariate time-series Electroencephalography signals. Additionally, many suggested methods depend on inadequate EEG segmentation techniques, resulting in unreliable results. This study presents an in-depth architectural design of a Convolutional Neural Network specifically tailored to extract features from wavelet-transformed EEG signals using Wavelet packet decomposition (WPD). In addition, the chosen testing strategy and data segmentation methodology ensures accurate, trustworthy, and reproducible performance results. This study introduces a data segmentation method to generate distinct intervals effectively capturing more temporal dynamics of the time-series data. The proposed model evaluation utilized 12 subjects' EEG data from the CHB-MIT dataset in a subject-specific manner, employing a Leave-One-Out cross-validation technique. The proposed architecture outperformed five reproduced state-of-the-art CNN models in the segment-based evaluation metrics. The proposed model achieved 78.00% accuracy, 65.17% sensitivity, and a high 90.83% specificity rate. Evaluation using the more straightforward KFold cross-validation technique demonstrated robust performance, achieving 96.68% accuracy, 97.41% sensitivity, and 95.95% specificity. The significant improvement in the model's specificity rates indicates a substantial reduction in false alarms, making the proposed model a reliable tool for seizure prediction. Keywords: Epilepsy, Seizure Prediction, Electroencephalography, Convolutional Neural Network, Wavelet Packet Decomposition