Synthetic database for testing algorithms of fetal ECG extraction from abdominal ECG

Abstract

The fetal electrocardiogram can be obtained from abdominal electrocardiogram using standard digital signal processing. The algorithms designed for this purpose cannot be evaluated by performance index, since the fetal electrocardiogram is not known a priori. In this paper we review the different models to generate synthetic or semi-synthetic databases, and the performance index to evaluate algorithms. We describe the development of a semi-synthetic database to measure the performance of the jadeR algorithm and PCA algorithm, before different combinations of signal to interference and signal to noise, with Gaussian white noise and pink noise. It is found that both algorithms are susceptible to noise, especially Gaussian white noise. The PCA algorithm has a better performance in cases of low noise and high maternal interference.