A Method for Fetal Heart Rate Extraction Based on Time-Frequency Analysis

A three-stage method for fetal heart rate extraction, from abdominal ECG recordings, is proposed. In the first stage the maternal R-peaks and fiducial points (QRS onset and offset) are detected, using time-frequency analysis, and the maternal QRS complexes are eliminated. The second stage locates the positions of the candidate fetal R-peaks, using complex wavelets and pattern matching theory techniques. In the third stage, the fetal R-peaks that overlap with the maternal QRS complexes are found. The method is validated using a dataset of 4 long duration recordings and the obtained results indicate high detection ability of the method (96% accuracy)

1. Introduction

The fetal electrocardiogram (FECG) can be derived from the abdominal ECG (AECG) and be used for the extraction of fetal heart rate (FHR), which is a marker for the cardiac condition of the fetus [1]. Various research efforts have been carried out in the area of FECG and FHR extraction, including subtraction of an averaged pattern [2], matched filtering [3], adaptive filtering [4]–[6], orthogonal basis functions [7], fractals [8], FIR [9], dynamic neural networks [10], temporal structure [11], fuzzy logic [12], frequency tracking [13], polynomial networks [14], and real-time signal processing [15]. The wavelet transform (WT) is another approach that has been proposed for FECGs processing. Several techniques for noise removal and detection of fetal waveforms have been used, involving Gabor-8 wavelets and Lipschitz exponent's theory [16], bi-orthogonal quadratic spline wavelet, modulus maxima theory [17] and complex continuous wavelets [18].