A NEW QRS DETECTION AND ECG SIGNAL EXTRACTION TECHNIQUE FOR FETAL MONITORING

["The electrocardiogram (ECG) signal, a graphical recording of the electrical potentials generated in association with heart activity, is one of many important physiological signals commonly used in clinical practice. As in adults, the well-being and status of a fetus can be assessed from a fetal ECG signal. Detecting and analyzing the fetal ECG are the primary objectives of electronic fetal monitoring. DSP based techniques have played a significant role in obtaining and processing the fetal ECG signal. There are two primary assumptions: 1) the ECG signals of interest, i.e., the maternal ECG and the fetal ECG signals, are quasi-periodic; and 2) the abdominal ECG signal is a linear combination of the maternal ECG, the fetal ECG signals, and any interference signal. Two new algorithms for ECG signal processing are introduced. A QRS detection algorithm that forms the basis for the ensuing processing of the ECG signal is discussed first. The main components of the QRS detection are two moving average filters implementing a bandpass filter and a feature enhancement filter based on an energy ratio computation. An ECG signal separation algorithm depending on QRS detection is then developed. The principle idea of the ECG signal separation algorithm is to align every cycle of the separating ECG signal through resampling, and then to apply the Fourier transforms to extract the required components. A novel fetal ECG signal extraction approach exploiting both of these fundamental signal processing algorithms is proposed for extracting the fetal ECG signal from the composite abdominal ECG signal. The maternal ECG signal is extracted first and eliminated and then the fetal ECG signal is extracted. Experimental results demonstrate the excellent flexibility of the proposed QRS detection algorithm both in terms of detection ability and detection accuracy. The fetal ECG signal extraction approach also shows promising results with a high degree of similarity between the extracted maternal ECG and fetal ECG signals and the original maternal ECG and fetal ECG signals. The morphological characteristics are also well preserved in the extracted maternal ECG and fetal ECG signals."]