Development of Epileptic EEG Pre-Processing Software with Seizure Detection System by Using 1D-CNN Model

DOI

抄録

<p>Epilepsy is a neurological disorder characterized by abnormal brain activity and recurrent seizures. The diagnosis of epileptic seizures typically involves experts visually examining the EEG (Electroencephalogram) signals of patients. The noise present in EEG signals can significantly disrupt the diagnosis process for experts, requiring more time and effort and placing a heavy burden. To alleviate the burden, this study developed software for pre-processing EEG signals of epileptic patients and used a pre-trained 1D-CNN model to classify the EEG signals. Our epileptic EEG pre-processing software, which named EEGreader, was designed and developed by using Qt Designer. It includes time scale adjustment, amplitude adjustment, and filter-based noise removal to facilitate epileptic EEG signal display. We pre-trained a 1D-CNN (One Dimensional Convolutional Neural Network) model by using the CHB-MIT dataset and implemented real-time classification of EEG signals. The test accuracy of the model achieved 82% which means the model could successfully detecting epileptic seizures. Our developed software for pre-processing epileptic EEG signals and the epileptic seizure detection system can effectively help experts reduce the difficulty in reading EEG signals and the burden of diagnosing epileptic seizures, making it of great practical significance.</p>

収録刊行物

  • 生体医工学

    生体医工学 Annual61 (Abstract), 113_1-113_1, 2023

    公益社団法人 日本生体医工学会

詳細情報 詳細情報について

  • CRID
    1390017345590446848
  • DOI
    10.11239/jsmbe.annual61.113_1
  • ISSN
    18814379
    1347443X
  • 本文言語コード
    ja
  • データソース種別
    • JaLC
  • 抄録ライセンスフラグ
    使用不可

問題の指摘

ページトップへ