Fetal Heart Rate Detection Algorithm And Implementation Based On Short-time Fourier Transform

Nowadays, the fetal heart rate electronic motitoring system is widely used in clinical practice, becoming a commonly used means of maternal and fetal safety. The traditional fetal heart rate electronic monitoring system mostly uses the autocorrelation algorithm to calculate the fetal instantaneous heart rate. However, due to the fetus being in the maternal abdomen, fetal heart sound signal is often disturbed by maternal body, sensor and external environment, which makes fetal heart sound signal easily submerged in the noise. In addition, activities for mother and fetus in intrauterine activities will lead to a larger fetal heart sound signal amplitude fluctuations. For traditional autocorrelation algorithm is difficult to accurately measure the fetus of instantaneous heart rate problem, this paper put forward a new kind of heart sounds based on short-time Fourier transform method to detect instantaneous heart rate. Moreover, combined with ultrasonic Doppler sensor, embedded ARM chips, PC, this paper designs and implements the heart rate detection based on short-time Fourier transform system.The main work done in this paper is as follows:(1) This paper introduces the basic theory of short-time Fourier transform in detail. The basic idea of Fourier transform is using a sliding over time analysis window to add and cut the original signal. After that, the non-stationary signal is decomposed into a series of short approximate smooth signals. Finally, the theory of Fourier transform is used to analysis each short stationary signal spectrum.(2) This paper introduces a new kind of heart sounds based on short-time Fourier transform method to detect instantaneous heart rate. The basic idea of this method is as follows. Firstly, the short-time Fourier transform is carried out on the cardiac sound. Secondly, cardiac sound feature template is extracted from2-d time-frequency diagram. And the result of cardiac sound feature template matching the target image is normalized to plot the cross-correlation curve. Finally, according to the cross-correlation curve adjacent peaks in the time interval, the instantaneous heart rate of the fetus is calculated out. Simulation result shows that the algorithm of the recognition accuracy is5%higher than the traditional algorithm.(3) This paper designs a heart rate monitoring system to realize the new method introduced in the paper. Firstly, the ultrasonic Doppler probe is used to obtain the fetal heart sound signal. After amplified and filtered in the front analog circuit, the fetal heart sound signal obtained is sent into the AD analog-to digital conversion module. Secondly, AD analog-to digital conversion module converts analog signals into digital signals and sends these digital signals into system control module for data encapsulation. In addition to achieving the function of data encapsulation, system control module is also responsible for the coordinate control of the overall system and passes the encapsulated data to PC data processing module via a serial port communication module. Finally, by using VC++6.0software, the PC data processing module completes the function of software filtering, time-frequency analysis and calculating the instantaneous heart rate and displaying the real-time cardiac sound signal and the instantaneous heart rate on the screen.