The application of finite impulse response filters to the detection of fetal electrocardiogram signals.

Abstract

An investigation is made into the potential application of linear phase digital filters to the detection of fetal electrocardiogram signals buried in noise. Such an assessment is made by applying both matched and linear phase filters to six computer simulated fetal signals and also to experimental data. The number of times that the R-wave locations are correctly located (N), the RMS error in R-wave location (RMS) and the correlation coefficient between the averaged and clean signals are computed. It is found that the averaged fetal complexes computed using these two types of filter are almost identical. However, for three of the signals, the values for N and RMS obtained using the linear phase filter are inferior to the corresponding results obtained with the matched filter. It is suggested that the averaged complex obtained using the linear phase filter could be used as an approximation to the matched filter template; it is found that this procedure results in an effectiveness of detecting R-waves that is, for the most part, comparable with the performance of a matched filter based on the QRS complex.