Measuring and Analyzing Infant Heart Rate Variability

Heart Rate (HR) is not constant , it is modulated by different systems , which cause fluctuations in HR that are called Heart Rate Variability.

Heart Rate (HR) is not constant , it is modulated by different systems , which cause fluctuations in HR that are called Heart Rate Variability. With adults and babies , Heart Rate Variability (HRV) has been shown to reflect the behavior of different parts of the autonomous nervous system .
The spectral analysis of HRV has contributed to the understanding of the sympathetic and para-sympathetic activities , development and to the balance between these two systems.
In the last years , research’s have shown that from a certain stage of development , fetal HRV shows similar properties to that of adults.
Spectral Analysis of fetal HRV can be of use in monitoring the fetals’ development , and the development of the different systems that modulate the fetal Heart Rate (HR).

The Problem
As opposed to adults , sampling the fetal HR and performing an accurate spectral analysis of the fetal HR are much more difficult , due to a number of reasons :
1) The fetal HR must be recognized from within the womb , and separated from the mothers’ HR.
2) The fetus changes wake states and positions , which affects the reading and make the signal weaker and less stationary , while adding more noise.
3) During the early gestation period , the fetal HR will be weak and hard to discern.

The Solution
In order to take a reliable measurement , we used non-invasive Doppler ultrasound to measure the HR of the fetus. The fetal heart beat was measured by sampling the fetal cardiac wall movement and the heart valve movement with a 2.7Mhz transducer (HP8040a) , which in addition to the heart beat waveform also gave instantaneous Heart Rate data.
To solve the spectral analysis problem , we used 2 methods:
1)Periodogram – the periodogram is basically the square of the signals’ fft (fast fourier transform).
2)Autoregression – estimate an autoregressive model that simulates the behavior of the HR signal , and then compute the models’ spectral power.

– HP8040A ultrasound transducer for recording the fetal HR.

– Matlab (math software) for computing the spectral densities in the 2 methods.

– Codas A/D converter ( Hardware & Software supplied) to digitize the ultrasound analog output
and convert to Matlab format.
Block Diagram

– In order to achieve the best results , sampling must be done for at least 10 minutes , the signal should be segmented and averaging should be done between the results of different segments of the signal.

– Both methods gave similar results , but not identical , as the periodogram is the more robust method , and there is no need to set model parameters , both methods should be used in conjunction , and the autoregressive model should be tuned by the periodogram.

– Choosing the autoregression parameters must be done with care , and with consideration to the expected model , as the results may change drastically.

– Prof. Dan Adam from the Biomedical faculty.

– Dr. Israel Taler from RAMBAM hospital.

– Yohanan Erez , the chief engineer of the PSPL lab.

– The Ollendorf Research Center Fund.