IEEE Access (Jan 2021)

Non-Invasive Fetal Electrocardiogram Extraction Based on Novel Hybrid Method for Intrapartum ST Segment Analysis

  • Radek Martinek,
  • Radana Kahankova,
  • Rene Jaros,
  • Katerina Barnova,
  • Adam Matonia,
  • Michal Jezewski,
  • Robert Czabanski,
  • Krzysztof Horoba,
  • Janusz Jezewski

DOI
https://doi.org/10.1109/ACCESS.2021.3058733
Journal volume & issue
Vol. 9
pp. 28608 – 28631

Abstract

Read online

This study focuses on non-invasive fetal electrocardiogram extraction based on a novel hybrid method, which combines the advantages of non-adaptive and adaptive approaches for non-invasive fetal electrocardiogram morphological analysis. Besides estimating fetal heart rate, which is the main parameter used in the clinical practice, this study provides non-invasive ST segment analysis on data from Abdominal and Direct Fetal Electrocardiogram Database consisting of simultaneous traditional - gold standard invasive fetal scalp electrode and non-invasive fetal electrocardiogram recorded during delivery. This innovative approach utilizing the combination of independent component analysis and recursive least squares algorithms has the potential to extract valuable information from non-invasive fetal electrocardiogram in order to identify eventual sign of fetal distress. This was a prospective observational study of non-invasive fetal electrocardiogram, using 4 abdominally sited electrodes, against the traditional fetal scalp electrode on 8 patients. In terms of fetal heart rate estimation, the accuracy was high for all 8 tested patients with average value equaled 0.20 beats per minute and average value of 1.96 standard deviation equaled 5.80 beats per minute. In 7 patients, it was possible to perform the ST segment analysis with high accuracy in determining T/QRS in comparison with the reference fetal scalp electrode signal with average values and 1.96 standard deviation equaled 0.008 and 0.031 respectively. This study thus demonstrates that ST segment analysis is feasible using non-invasive fECG using the proposed hybrid method.

Keywords