Development of 1D-CNN Methods for Classifying Pediatric Epilepsy Through EEG Signals

Flores-Palermo O.; Espiritu-Cueva C.; Ugarte W. · Communications in Computer and Information Science (2025) · DOI: 10.1007/978-3-031-91428-7_28

Resumen

Our work focuses on the development of a method that allows high-precision classification of ictal and preictal seizure activity in pediatric patients by analyzing EEG signals. In this study, different methods are analyzed using the CHB-MIT dataset, applying various preprocessing techniques and 1D-CNN model architectures. This paper compares two data acquisition methods identified during the experimentation process, namely training a 1D-CNN + LSTM model with single channel data (one channel per second) and multi-channel data (23 channels per second). The results showed that the multi-channel methodology outperformed its counterpart, achieving sensitivity, specificity, precision, accuracy and F1 score of 94.05%, 85.90%, 87.73%, 90.12% and 90.79%, respectively. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Instituciones: Universidad Peruana de Ciencias Aplicadas