Model fitting approach for noise handling in the power spectrum density of electromagnetic signals

DOI
  • Hoshi Hideyuki
    Precision Medicine Centre, Hokuto Hospital, Obihiro, Japan Medical Imaging Business Centre, Healthcare Business Group, RICOH Company, Ltd., Tokyo, Japan
  • Okumura Naohiro
    Medical Imaging Business Centre, Healthcare Business Group, RICOH Company, Ltd., Tokyo, Japan
  • Kobayashi Momoko
    Precision Medicine Centre, Kumagaya General Hospital, Kumagaya, Japan
  • Sakamoto Yuki
    Precision Medicine Centre, Kumagaya General Hospital, Kumagaya, Japan
  • Hirata Yoko
    Department of Neurosurgery, Kumagaya General Hospital, Kumagaya, Japan Department of Neurosurgery, Toho University Ohashi Medical Center, Tokyo, Japan
  • Ichikawa Sayuri
    Clinical Laboratory, Kumagaya General Hospital, Kumagaya, Japan
  • Fukasawa Keisuke
    Clinical Laboratory, Kumagaya General Hospital, Kumagaya, Japan
  • Shigihara Yoshihito
    Precision Medicine Centre, Hokuto Hospital, Obihiro, Japan Precision Medicine Centre, Kumagaya General Hospital, Kumagaya, Japan

Bibliographic Information

Other Title
  • 関数モデルを使ったパワースペクトル密度のノイズ低減手法の検討

Abstract

<p>Background: A power spectrum density (PSD) of electromagnetic signal is used for evaluating alertness and cognitive status of patients. The PSD is often distorted by artefacts containing periodical components. In this study, we applied modelling approach to the PSD data and its noise reduction efficiency was assessed. Method: The PSD (1-70 Hz) of MEG signal was estimated using Blackman Tukey method. The PSD was fitted using a combination of Lorentzian function and multi Gaussian functions. The estimated coefficients were used for calculating clean PSD as fitted functions, which were visually inspected against the original PSD. Results: The result showed that periodic noises were reduced by the fitting procedure with maintaining key features of PSD. Discussion: This procedure can be useful for cleaning PSD contaminated by periodical noises which are resistant to classical filtering and/or signal separation approach.</p>

Journal

Details 詳細情報について

  • CRID
    1390852714994579840
  • NII Article ID
    130008105421
  • DOI
    10.11239/jsmbe.annual59.417
  • ISSN
    18814379
    1347443X
  • Text Lang
    ja
  • Data Source
    • JaLC
    • CiNii Articles
  • Abstract License Flag
    Disallowed

Report a problem

Back to top