• DocumentCode
    122970
  • Title

    Non-invasive extraction of fetal electrocardiogram using fast independent component analysis technique

  • Author

    Shehada, Dina ; Khandoker, Ahsan H.

  • Author_Institution
    Khalifa Univ., Abu Dhabi, United Arab Emirates
  • fYear
    2014
  • fDate
    17-20 Feb. 2014
  • Firstpage
    349
  • Lastpage
    352
  • Abstract
    Fetal electrocardiogram (fECG) provides clinically important information concerning the electrophysiological state of a fetus. The fECG contains activity of electrical depolarization and repolarization of fetal heart. In this paper a technique is proposed called Fast Independent Component Analysis (Fast ICA) to extract the fECG signals from the 20 sets of abdominal ECG signals. Using Fast ICA, 91.4% R peaks of fECG signals were extracted as validated by the direct fECG R peaks measured from scalp. The mother electrocardiogram (mECG) is extracted from the four channel abdominal ECGs, canceled and then the fECG is extracted. Bland Altman and correlation calculations showed no significant difference between the extracted fECG and direct fECG RR intervals. The correlation coefficient ρ was 0.84 and the p value was 0.
  • Keywords
    bioelectric potentials; electrocardiography; independent component analysis; medical signal processing; obstetrics; Bland Altman-correlation calculations; correlation coefficient; electrical depolarization; electrical repolarization; electrophysiological state; fECG RR intervals; fECG signal extraction; fast independent component analysis technique; fetal electrocardiogram; fetal heart; fetus; four channel abdominal ECG signals; noninvasive extraction; Algorithm design and analysis; Data mining; Electrocardiography; Electrodes; Independent component analysis; Scalp; Signal processing algorithms;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Biomedical Engineering (MECBME), 2014 Middle East Conference on
  • Conference_Location
    Doha
  • Type

    conf

  • DOI
    10.1109/MECBME.2014.6783275
  • Filename
    6783275