Reduction of ERP measurement time by weighted averaging for responses using EEGNet

DOI
  • Yoshida Aoi
    Graduate School of Science and Engineering, Yamagata University, Yamagata, Japan
  • Sato Hikaru
    Graduate School of Science and Engineering, Yamagata University, Yamagata, Japan
  • Ishikawa Bunnoshin
    Hotokukai Utsunomiya Hospital, Tochigi, Japan
  • Kaga Kimitaka
    Tokyo Medical Center, Tokyo, Japan
  • Fukami Tadanori
    Graduate School of Science and Engineering, Yamagata University, Yamagata, Japan

Bibliographic Information

Other Title
  • EEGNetを用いた応答に対する加重平均処理による事象関連電位計測時間の短縮

Abstract

<p>The ERP can be estimated by averaging the responses to multiple target stimuli and suppressing background EEG and artifacts. As the number of responses used for averaging increases, the effect of suppressing components other than EPR becomes higher, but the longer the measurement time, the greater the mental and physical burden on the subject. Therefore, we aim to shorten the measurement time by using EEGNet, which is one of the deep learning networks. Here, by learning so that the network outputs 1 for the input of the response of the target stimulus and 0 for the response of the non-target stimulus, the ERP is estimated by the weighted averaging using the output value as the weight for the response. As a result, compared to the conventional averaging, the P300 showed a 14% larger amplitude with a 13% smaller number of responses while maintaining the waveform shape and the peak latency.</p>

Journal

Details 詳細情報について

  • CRID
    1390289765040834560
  • NII Article ID
    130008105335
  • DOI
    10.11239/jsmbe.annual59.486
  • ISSN
    18814379
    1347443X
  • Text Lang
    ja
  • Data Source
    • JaLC
    • CiNii Articles
  • Abstract License Flag
    Disallowed

Report a problem

Back to top