A Deep Learning Approach for Epilepsy Seizure Identification Using Electroencephalogram Signals: A Preliminary Study

Comparison of deep learning architectures for seizure identification from raw EEG signals.

Journal article · 2023

Graphical abstract

Publication record

Journal: IEEE Latin America Transactions, 21(3), 419–426
DOI: 10.1109/TLA.2023.10068845

Overview

This preliminary study compares five deep learning networks for seizure identification using raw electroencephalogram signals from the TUH EEG Seizure Corpus. The workflow addresses the computational cost of long time-series recordings through signal selection, seizure-event aggregation, and separate patient–control and patient-specific evaluation strategies.

The results reported in the article show that comparatively simple recurrent and hybrid architectures can provide competitive seizure-detection performance while keeping the computational workflow manageable.

Research perspective

This was one of my first English-language journal articles. It shaped how I approach reproducible signal-processing experiments, technical writing, and the communication of preliminary biomedical results.

Citation

@article{jacobozavaleta2023epilepsy,
  author  = {Jácobo-Zavaleta, Sergio and Zavaleta, Jorge},
  title   = {A Deep Learning Approach for Epilepsy Seizure Identification Using Electroencephalogram Signals: A Preliminary Study},
  journal = {IEEE Latin America Transactions},
  volume  = {21},
  number  = {3},
  pages   = {419--426},
  year    = {2023},
  doi     = {10.1109/TLA.2023.10068845}
}