The phenomenon of fetal heart rate variability is well known in medicine. Having a fetal beat-to-beat heart rate, it is possible to make its spectral analysis.
Abstract
The phenomenon of fetal heart rate variability is well known in medicine. Having a fetal beat-to-beat heart rate, it is possible to make its spectral analysis. The results of the spectral analysis allow prediction of certain infant diseases and estimation of general health condition of the fetus.
Taking pure high quality signals from fetus heart is a problem. It is common to measure mechanical signals of fetuses using Doppler Ultrasound, but to obtain good quality signals is a very difficult task, since they depend on many outside factors. Therefore, new signal processing algorithm was needed to overcome that problem.
This work presents an algorithm for fetus beat-to-beat heart rate extraction from Ultrasound signals, especially noise polluted. It is based on previous works, dealing with measurement, analysis and processing of fetal heart rate variability.
Basic Approach
In this work we used two different methods for Beat-to-Beat Heart Rate estimation. One is based on auto-correlation and cross-correlation. Another one uses analysis of the fetus heart signal in frequency domain. Scheme shows the basic steps of the algorithm:
Tools
We used the analog output of HP 8040A Cardiograph and A/D PC card in order to sample the Doppler echocardiogram. The sampling was made using CODAS program, and most of the signals were of mature fetal near delivery. The algorithm was developed in MATLAB .
Results
We managed to achieve results of high quality for the new algorithm, especially it has proved to be much reliable with more complicated or noisy signals than the previous ones. An example of algorithm performance is shown below, where it is compared to heart rate signal, generated in ultrasound hardware. The heart rate signal belong to noisy heart signal, and it can easily be seen that original heart rate method often can’t deal with the noise and it diverges:
After algorithm development, achieved beat-to-beat heart rate’s Power Spectrum Density was tested using auto-regression, and compared to theory:
Conclusions
The proposed algorithm was found to be efficient and reliable. It manages to recover heart rate from very noisy or weird shaped heart signals. The results of spectral analysis were also a satisfactory. All this clears the way for fetal condition evaluation system development.
Acknowledgments
We would like to thank Dr. I. Taller, Prof. D. Adam and J. Erez for their help and guidance. This project was supported by Ollendorff Research Center Fund.