A Phonocardiography based Real-Time Fetal Heart Rate Monitoring System
The mechanical activities in a pregnant woman's abdomen produce low frequency vibrations and sounds. These vibroacoustic signals, also known as fetal phonocardiogram, carry valuable physiological information that can be potentially used for fetal monitoring. The most common method for fetal monitoring is recording and monitoring of Fetal Heart Rate (FHR). In this paper, a real-time FHR monitoring system using phonocardiography is developed. A Bluetooth aided wireless Data Recording Module (DRM) is used for acquisition of the vibroacoustic signal. Wavelet based noise suppression procedures are adopted for de-noising of these signals. The de-noised signals are segmented through envelope detection and a threshold criterion. The FHR is estimated through a novel beat detection and error correction algorithm. Quantitative evaluation and testing of the developed system are done using a large number of simulated fPCG signals. The system performance observed to be more than 97% accurate, when applied to 21 real-time fPCG signals for FHR determination. The developed system for real-time estimation and analysis of fetal heart rate can set grounds for the challenging task of non-invasive continuous monitoring of the fetus.
Keywords: Fetal heart Rate, wavelet transform, de-noising, segmentation, phonocardiography, fetal monitoring.
Auscultation means interpretation of sounds produced by the fetal heart . It is a fundamental tool in the diagnosis of fetal well-being during its primary health care . In some circumstances, particularly in remote areas or developing countries, auscultation may be the only method available for fetal monitoring . However, auscultation of fetal heart sounds is not considered an easy task . This is due to the fact that fetal heart sounds are of low frequency and timing between events is very short. Therefore, a computerized fetal heart sound analysis is vital to assist the obstetricians.
Advanced techniques like ultrasound Doppler based fetal cardiotocography (fCTG) , fetal electrocardiography (fECG) , fetal magnetocardiography (fMCG)  and fetal pulseoximetry (fPOM) , can provide more direct evidence but require expensive equipments, specialized technicians to operate, experts to interpret the results, high maintenance cost, permanent placement and generally demand more resources to function properly. These requirements can only be met in the advanced hospitals and are way beyond the rural health-care centers as well as for urban clinics . On the other hand, fetal phonocardiography (fPCG) is a low cost, non-invasive (passive) and simple technique in comparison to the widely accepted Doppler ultrasound examination . It is a suitable tool for long-term surveillance of the fetus . Using this approach, measurement of important diagnostic information such as fetal heart rate, fetal movements and fetal breathing moments becomes possible . Moreover, phonocardiography is an excellent means for auscultation training to the medicos and helps to understand the haemodynamic of the fetal heart . The fPCG signal is a trace of vibrations and acoustic energy produced by the mechanical activity of fetal heart, mother's organ and other noises. It contains several non-stationary or transient characteristics such as timing of the heart sounds, their components, locations in the cardiac cycle and a fast change of frequency as time progresses .
Antepartum and intrapartum fetal monitoring is commonly based on evaluation and analysis of FHR . FHR trace is a physiological signal, represents time versus heart rate (in bpm) waveform derived from fPCG signal. It provides a picture of overall heart activity of the fetus for a considerably longer span of time . A precise examination of various FHR parameters is very useful in pre-detection of stillbirth, miscarriage, pre-term labor and other domain anomalies. Several efforts have been made for the extraction of FHR based on fetal phonocardiography. Some examples are the band pass filtering , the linear cross correlation  the variable comb filtering , or autocorrelation . All these approaches either could not deal with the noisy fPCG signals, or can only be used off-line.
In view of above considerations, this work is focused on development of a system for phonocardiography based monitoring of FHR in real time conditions. For this purpose, a Bluetooth aided wireless Data Recording Module (DRM) is used for acquisition of these signals from the maternal abdominal surface. Wavelet based noise suppression procedures are adopted for de-noising of these signals. A new most suitable wavelet named 'fetal' is developed for analysis these signals. The de-noised signals are segmented through envelope detection and threshold criteria. The FHR is estimated on the basis of time domain knowledge of fundamental components of the fetal heart. Computations are carried out in MatlabTM version 7.8.0 (R2009a).
The rest of the paper is organized as: physiology and characteristics of fetal heart are discussed in section2. Section 3 describes the methodology applied for data acquisition, de-noising and extracting the FHR from fPCG signals. In section 4 computer implementation of the proposed methodology is discussed. The experimental results and clinical trials are discussed in section 5. The section 6 contains the discussion and conclusions of the work followed by a list of references.
2. Physiology and Characteristics of Fetal Heart:
The heart and circulatory system of a fetus begin to form soon after conception . The growing fetus is fully dependent on a special organ called the placenta for nourishment. One side of the placenta is attached to the uterus, and the other side is attached to a liquid-filled sac that holds the fetus. A special cord called the umbilical cord links the placenta to the fetus. The mother's blood flows through a thin layer of cells in the wall of the uterus, giving the fetus food and oxygen while removing any wastes like carbon dioxide. There is actually no direct contact between the circulatory systems of the mother and fetus.
The fetus does not use its own lungs until birth, so its circulatory system is different from that of a newborn baby. Before birth, the fetal heart does not have to pump blood to the lungs to pick up oxygen. In other words, the fetal heart does not need a separate pulmonary artery and aorta. In the fetal heart, these two blood vessels are connected by a blood vessel called the ductus arteriosus. After birth, the ductus closes and a separate left pulmonary artery and aorta form.
The fetal heart also has an opening between the upper chambers (the right and left atria) called the foramen ovale. It lets blood flowdirectly from the right atrium to the left atrium during fetal development, but it also closes after birth. So the ductus arteriosus and the foramen ovale are part of the fetal circulatory system before birth but disappear soonafter birth. Figure 1 shows the organization of the fetal heart and its circulatory system.
Inside the fetal heart:
- Blood enters the right atrium, the chamber on the upper right side of the heart. Most of the blood flows to the left side through a special fetal opening between the left and right atria, called the foramen ovale.
- Blood then passes into the left ventricle (lower chamber of the heart) and then to the aorta, (the large artery coming from the heart).
- From the aorta, blood is sent to the head and upper extremities. After circulating there, the blood returns to the right atrium of the heart through the superior vena cava.
- About one-third of the blood entering the right atrium does not flow through the foramen ovale, but, instead, stays in the right side of the heart, eventually flowing into the pulmonary artery.
A normal cardiac cycle contains two major sounds :
- Systolic beat S1, caused by the closure of the mitral and tricuspid valves.
- Diastolic beat S2, caused by the closure of the aortic and pulmonary valves.
Figure 2 shows a typical fetal phonocardiographic (fPCG) signal of major heart sounds. These sounds are produced by the closure of heart valves.
The frequency spectrum of mechanical vibrations due to fetal breathing and heart activity has been studied through available literature . The significant information in the overall fetal signal is contained in a spectrum between 0.1 to 70 Hz. Fetal mobility and breathing movements result in weak and low frequency sound. This sound is of very low amplitude and cannot be detected by normally used sensors. Fetal heart sound intensity is comparatively high and mostly comprises of two-frequency bands. Low frequency band is from 20 - 40 Hz, and the high frequency band is from 50 - 70 Hz. Frequency spectra of vibrations due to fetal activities are summarized in Table 1.
This analysis of the fetal heart sound frequency spectrum is very important for the design of the sensing element and associated circuitry . The frequency bands listed above are used for setting the cutoff frequency of active filters applied in the prototype instrument under discussion.
3. Development of a FHR Monitoring System:
The methodology for acquisition of these signals from maternal abdominal surface, de-noising and extraction of FHR is discussed in the following subsections.
3.1 Data Acquisition:
Fetal heart sound recording system is considerably susceptible to the ambient noise . The abdominal sensor records principally the sound originating from the fetal heartbeats, but this sound gets mixed with damped version of following unwanted sounds.
- Movement of measurement head during recording (shear noise).
- External noise originating from the environment (ambient noise).
- Acoustic noise produced by the fetal breathing movements.
- Maternal digestive sound (gut sound).
- Sound of maternal heart and breathing activity.
In fetal phonocardiographic measurement, these unwanted sounds create major problem at signal processing stage. Hence advanced digital signal processing technique is required for the extraction of fetal heart sound from the acquired signals. To overcome the problems associated with the existing fetal monitoring system and their signal acquisition procedures, a phonocardiography based wireless data acquisition system is developed specially for this purpose . This system is portable, comfortable for longer periods and having sustainable battery backup. It consists of wireless data acquisition, processing and transmission hardware along with a remote monitoring station for fetal monitoring applications. The implementation of Bluetooth Wireless Technology (BWT) in continuous fetal monitoring provides not only the reduction of supply voltage, power consumption and memory capacity, but also increasing applications of the system, like battery backup and mobility . The block diagram of the data acquisition system is shown in Figure 3.
The acquired fPCG signals are then transmitted to the personal computer (PC) using Bluetooth transmitter of the sensor and received through integrated Bluetooth receiver of the PC. The signals are then saved in Wave format for its graphical display and subsequent processing. A monitoring obstetrician/gynecologist can also replay the recorded fPCG signal by connecting a headset to Line-out port of the PC.
3.2 Wavelet De-noising:
The Wavelet Transform (WT) is a two-dimensional timescale processing method for non-stationary signals with adequate scale values and shifting in time. There are many application areas of WT such as sub-band coding, data compression, characteristic point detection and noise reduction. It is capable of representing signals in different resolutions by dilating and compressing its basis functions. The major advantage of the WT is that it has a varying window size, being broad at low frequencies and narrow at high frequencies, thus leading to an optimal time-frequency resolution in all frequency ranges [28, 29]. The main idea of wavelet analysis is to measure the degree of similarity between the original waveform s(t) and the basic function of the WT also called the mother wavelet, through wavelet coefficients computation. The calculation process is performed on shifted version of the mother wavelet thus moving along the time, and on stretched or compressed version of the mother wavelet thus varying the frequency. The continuous wavelet transform (CWT) is defined as the convolution between the original signal s(t) and a wavelet .
Where s(t) is the input signal, a is the scaling factor, b is the translation parameter and ?(t) is the transforming function called mother wavelet. The mother wavelet is given by:
The DWT coefficients are usually sampled from the CWT on a dyadic grid, choosing parameters of translation b = n*2m and scale a = 2m. The mother wavelet in DWT is defined as:
DWT analyzes the signal by decomposing it into its coarse and detail information, which is accomplished by using successive high-pass and low-pass filtering operations, on the basis of the following equations:
Where and are the outputs of the high-pass and low-pass filters with impulse response h and g, respectively, after upsampling by 2 [30, 31].
Successful application of wavelet transform depends heavily on selection of the wavelet family and mother wavelet. The main criteria in choosing a family of wavelets are as follows:
- The support of wavelet function, scaling function, and their Fourier transforms: the speed of convergence at infinity to 0 of these functions when the time or the frequency goes to infinity, which quantifies both time and frequency localization.
- The symmetry, which is useful in avoiding dephasing in image processing.
- The number of vanishing moments for wavelet function or for scaling function (if it exists), which is related to reducing the polynomial degree of time series being analyzed and is useful for compression purpose.
- The regularity, which is useful for getting nice features like smoothness of the reconstructed signal or image .
The recorded fPCG signals are decomposed into time-frequency representations using Discrete Wavelet Transform (DWT) based Multi-resolution analysis. The major advantage of DWT is that, it provides high time resolution and low frequency resolution for high frequencies and high frequency resolution and low time resolution for low frequencies. Because of its great time and frequency localization ability, the DWT can reveal the local characteristics of the input signal . The standard de-noising procedure affects the signal in both frequency and amplitude, and involves following steps:
- Decomposition of the fPCG signal by applying DWT into N levels using band-pass filtering and decimation to obtain the approximation and detail coefficients.
- Thresholding of these decomposition coefficients, corresponding to artifacts, using an appropriate de-noising algorithm.
- Reconstruction of the fPCG signal from these thresholded detail and approximation coefficients using the inverse discrete wavelet transform (IDWT) .
Different mother wavelets belonging to different wavelet families are used in the literature for analyzing phonocardiographic signals [35-39] such as Haar, Daubechies, Coiflets, Biorthogonal, Symlets, Morlet and Discrete Meyer wavelets. Each wavelet family possesses their own unique properties that make them more appropriate for a certain range of applications. The choice of the wavelet family, mother wavelet and its order greatly affect the accuracy of the analysis. This paper deals with this problem by introducing a new wavelet which will be most suitable for analyzing the fPCG signals. The developed wavelet is orthogonal and named as 'fetal'. It has small number of coefficients in high pass subbands and allows the signal singularities, transitions and edges intact in the low pass subband. The normal fetal heart sound ranges from 20-200 Hz. First, an 8th order Butterworth low pass filter with cutoff frequency of 200 Hz and sampling frequency of 800 Hz is designed. The coefficients of this filter are used by the 'wfilters' function of Matlab to compute the four filter coefficients used by the DWT. Figure 4 shows the coefficients of these four filter.
3.3 Extraction of FHR:
The heart sound signal of normal fetal heart mainly consists of the first heart sound (S1), the systolic period, the second heart sound (S2) and the diastolic period in this sequence in time . Where, S1 and S2 are known as fundamental components of fetal heart sound. Segmentation of fetal heart sound into associated cardiac cycle is a primary step prior to the analysis of heart sounds for diagnostic purpose . It is the partitioning of fPCG signal into cardiac cycles, detection of these events and calculating interval between them. In this work, the segmentation of boundaries of these components is carried out by envelope detection of the fetal heart sound signal. The envelope of a signal is the outline of the signal. An envelope detector is a system that connects all of the peaks in the signal. The envelope of fPCG signal is generated using Hilbert Transform. For this purpose, an analytic signal of the fPCG signal is created using Hilbert Transformer. The original signal is time-delayed to match the delay caused by the Hilbert transform, which is one-half the length of the Hilbert filter (in this case it is 16). The analytic signal is a complex signal, where the real part is the original signal and the imaginary part is the Hilbert transform of the original signal.
Let is the delayed input fPCG signal, the Hilbert transform of this signal is found by using a 32-point Parks-McClellan FIR filter and is given by and can be determined as:
This signal is then multiplied by i (the imaginary unit) and added to the delayed version of the original signal to find the analytic signal which is then given by:
The envelope of the signal can be found by taking the absolute value of the analytic signal (). In order to eliminate ringing and smooth the envelope, the result is subjected to a low-pass filter. The enveloped fPCG signal is then converted into a series of rectangular pulses corresponding to the beats of fetal heart sound. The combination of two beats constitutes a heart cycle. A normal cardiac cycle consists of two major sounds, the first heart sound (S1) and second heart sound (S2). The identification of these two principal heart sounds of the cardiac cycle is known as segmentation of fPCG signal. This segmented fPCG signal is used for the determination of instantaneous fetal heart rate corresponding to every 5 seconds interval of the fPCG signal. Figure 5 shows the flow chart illustrating the basic steps for the calculation of FHR in beats per minute (BPM).
The FHR trace, a time versus heart rate waveform, so obtained contains a lot of artifacts caused by mother and fetal movement or displacement of the transducer. Hence to remove the artifacts from the FHR trace, all the FHR values corresponding to 5 seconds interval are checked and corrected if required. In this work, FHR values falling in the range of 50 to 250 BPM are considered as valid data values. If the FHR value is not in the specified range, then the corresponding data is considered as invalid and is replaced by the running average FHR value. The algorithm for removal of artifacts from the FHR trace is depicted in Figure 6.
4. System Simulation:
Computations for wavelet de-noising of the fPCG signals and calculation of FHR are carried out in the MATLABTM environment. For this purpose, Simulink Toolbox of MatlabTM version 7.8.0 (R2009a) is used. A simulink model as shown in Figure 7 is developed for the real-time monitoring of FHR from fPCG signals. This model is built by interconnecting requisite blocks, which are available in the Simulink library and their parameters are fed while designing them for simulation .
The input to this model is the original fPCG signal from the maternal abdominal surface recorded through wireless DRM as discussed earlier; whereas outputs are the FHR estimated from the corresponding fPCG signal along with the time plot of FHR. The wireless DRM is interfaced with the personal computer through the Analog Input Block. This block opens, initializes, configures, and controls an analog data acquisition device. The opening, initialization, and configuration of the device occur once at the start of the model's execution. Output of the analog input block is connected to an automatic gain control subsystem for controlling the gain of input fPCG signal. This signal is then normalized to the maximum amplitude. The normalized fPCG signal is applied as an input to the wavelet de-noising subsystem. Figure 8 shows the subsystem for wavelet de-noising.
In this subsystem, the Analysis Filter Bank block decomposes the fPCG signal into a collection of subbands with smaller bandwidths and slower sample rates. This block uses a series of highpass and lowpass FIR filters to repeatedly divide the input frequency range. The fPCG signal is decomposed to the 3 levels of newly developed 'fetal' wavelet with the analysis filter bank. The Synthesis Filter Bank block reconstructs the signal decomposed by the Analysis Filter Bank block. This block takes in subbands of this signal, and uses them to reconstruct the signal by using a series of highpass and lowpass FIR filters. The reconstructed signal has a wider bandwidth and faster sample rate than the input subbands. The output of this subsystem is the de-noised fPCG signal. This de-noised fPCG signal is then fed to the envelope detection subsystem and thresholding block for segmentation of the fPCG signal. The Simulink model for envelope detection subsystem is shown in Figure 9.
In this subsystem, the de-noised fPCG signals are first given to envelope generation subsystem. Envelope of fPCG signal is generated using the Hilbert Transform and by interrelated supplementary processes. Output of the envelope generator is connected to the threshold block or relay block. The Relay block allows its output to switch between two specified values. When the relay is on, it remains on until the input drops below the value of the Switch off point parameter. When the relay is off, it remains off until the input exceeds the value of the Switch on point parameter. This block converts the envelope signal into a series of rectangular pulses. The detected rectangular pulses represent the segmented fetal heart sound signal which is then fed to the FHR calculation and error correction subsystems. Computer simulation of FHR subsystem is shown in Figure 10.
In these subsystems, combination of two counters and a pulse generator is applied to find the time separation between two successive pulses. This interval is then converted into BPM. The function of this subsystem is to calculate the value of FHR on the basis of detected and identified S1s and S2s from the de-noised fPCG signal. Finally, the estimated FHR is displayed through a numeric display along with the time plot of the FHR.
5. Experimental Results and Clinical Trials:
In order to evaluate the proposed system, a set of experiments are performed using simulated signals and clinical data, which are collected using wireless DRM.
Testing with Simulated Signal: The proposed system is evaluated using large number of simulated signals generated with the help of a signal simulation module (SSM) . This module replicates the pregnant women's abdomen by generating various signals similar to actual signals in and around the mother's womb. The variations in the parameters of this module can provide signals with different physiological conditions of the fetus. A simulated fPCG signal generated using SSM, with parameters set for normal conditions, is applied as an input to the proposed system for FHR estimation. In this system, the applied signal is first de-noised using wavelet based noise suppression procedure as discussed in the section 2.2. Corresponding waveforms of simulated fPCG signal and its de-noised form are depicted in Figure 11. The resultant de-noised fPCG signal is then fed to the FHR subsystem. Figure 12 shows the waveform at various processing steps and FHR trace generated from the simulated fPCG signal.
FHR trace of the simulated fPCG signal
Testing with Real-Time Signals: After satisfactory performance with simulated signals, the developed system has been tested on 21 different pregnant women in the clinical environment. The signals are acquired using wireless DRM placed on the mother's abdomen. Since the fPCG signal ranges from 20 to 200 Hz, the signals are sampled with sampling frequency of 8000Hz, 16 bit resolution. The acquisition is done from the subjects with gestation period between 32 to 42 weeks, an important period of gestation, with no evidence of labor. Simulation outputs presented in this section are derived from fPCG signal, of a pregnant woman, at 36 weeks of gestation with singleton pregnancy. In Figure 13 the real-time fPCG signal and its de-noised signal are depicted and Figure 14 shows the waveform at various processing steps and FHR trace of this signal.
FHR trace of the real-time fPCG signal
In addition, for qualitative evaluation of the proposed system, its results are compared with the simultaneously used ultrasound Doppler based fetal CTG. For this purpose, three quantities were measured :
- True Positive (TP): Value of FHR correctly detected by the proposed system.
- False Negative (FN): FHR not detected by the proposed system though it was present.
- False Positive (FP): FHR wrongly detected by the proposed system though it was not present.
It can be observed that in most of the cases, accuracy of the developed system is almost at same level with the ultrasound Doppler based instrument. Based on the data in Table 1, the overall accuracy of the system is observed to be around 97 %. The developed system for FHR monitoring provides reliable results even for early stages of pregnancy. From the results, it can be concluded that the presented system is viable and can effectively be used for long-term FHR monitoring of the fetus. Furthermore, the system also provides the time domain waveform of the fetal heart sound also known to be fetal phonocardiogram. This waveform displays the information regarding timing and location of the cardiac cycle.
6. Discussion and Conclusion:
Fetal heart rate monitoring is a popular diagnostic means for fetal well-being assessment. The Doppler ultrasound based fetal cardiotocography is not suitable for taking measurements over longer periods of time because of its invasiveness, need of experienced personnel to operate and requirement of specialized equipment . Therefore, the fetal heart sound diagnosis or fetal phonocardiography (fPCG) is the most viable, economical and a better alternative to detect the physiological condition of the unborn.
The proposed system extracts the FHR from the fPCG signal while addressing several issues such as:
- The system can be used for long-term monitoring of the fetus during its antepartum and intrapartum life.
- The energy requirement of the technique is negligible, which offers great portability of the system.
- The system can be used in noisy hospital and home environments.
- The developed system is simple enough to be used by the mothers themselves.
- Most importantly, the system with the proposed technique is highly cost effective; and hence can be used in low cost home monitoring application.
- The developed system is based on phonocardiography technique; in which no energy is emitted hence it eliminates the risk of harmful exposure to both mother and unborn.
In conclusion, the developed phonocardiography integrated real-time FHR monitoring system is user-friendly, non-invasive, portable and can be used for long duration monitoring. Hence it can be embedded in homecare devices or wearable systems for non-invasive fetal monitoring.
The authors of this paper would like to thank the experts and technical staff of Government Women Hospital, Gondia (M.S.), India for their kind support. We also thank to the pregnant women who contribute in the real-time experimental testing and clinical trials of the developed system.
1. P. Vrady, L. Wildt, Z. Beny and A. Hein, An advanced method in fetal phonocardiography, Computer Methods and Programs in Biomedicine, Vol. 7, p. 283 (2003).
2. E. H. Hon and S. T. Lee, Electronic evaluation of the fetal heart rate patterns preceding fetal death: further observations, Amer J Obstet Gynecol, Vol. 8(7), p. 814 (1965).
3. V. S. Chourasia and A. K. Tiwari, A Review and Comparative Analysis of Recent Advancements in Fetal Monitoring Techniques, CriticalTM Reviews in Biomedical Engineering, Vol. 36(5-6), p. 335 (2008).
4. C. B. Martin, Electronic fetal monitoring: a brief summary of its development, problem and prospects, Eur J Obstet Gynecol Reprod Biol, Vol. 78(2), p. 133 (1999).
5. S. I. Kim, E. G. Lee, Y. H. Lee, M. I. Park and D. S. Lee, Real time processing algorithm for high resolution fetal heart rate estimation, Proc of the 16th Annual Int Conf of the IEEE EMBS, Vol. 2, p. 1288, 3-6 Nov (1994).
6. J. Karin, M. Hirsch, O. Segal and S. Akselrod, Non invasive fetal ECG monitoring. Computers in Cardiology, p. 365-368 (1994).
7. H. J. M. ter Brake, A. P. Rijpma, J. G. Stinstra, J. Borgmann, H. J. Holland, H. J. G. Krooshoop, M. J. Peters, J. Flokstra, H. W. P. Quartero and H. Rogalla, Fetal magnetocardiography: clinical relevance and feasibility, Physica C: Superconductivity, Vol. 368(1-4), 10-17 March (2002).
8. U. Elchalal, A. Weissman, Y. Abramov, D. Abramov and D. Weinstein, Intrapartum fetal pulse oximetry: present and future, Int J Gynaecol Obstet, Vol. 50(2), p. 131 (1995).
9. L. D. Lathauwer, B. D. Moor and J. Vandewalle, Fetal electrocardiogram extraction by blind source subspace separation, IEEE Transactions on Biomedical Engineering, Vol. 47(5), p. 567 (2000).
10. J. Chen, K. Phua, Y. Song and L. Shue, A portable phonocardiographic fetal heart rate monitor. Proceedings of International Symposium of the IEEE on Circuits and Systems (2006).
11. F. Kovacs, Cs. Horvath, M. Torok and G. Hosszu, Long-term phonocardiographic fetal home monitoring for telemedicine systems, 27th Annual Int Conf of the IEEE EMBS, p. 3946 (2005).
12. N. Colley, D. G. Talberg and D. P. Southall, Biophysical profile in the fetus from a phonocardiographic sensor, European Journal of Obstetrics & Gynecology and Reproductive Biology, Vol. 23, p. 261 (1986).
13. E. van Vollenhoven and J. G. J. Chin, Phonocardiography: Past, Present and Future, Acfa Cardiologica, Vol. 48(4), p. 337 (1993).
14. Jimenez-Gonzalez and C. J. James, Blind Source Separation to Extract Foetal Heart Sounds from Noisy Abdominal Phonograms: A Single Channel Method, 4th IET Int Conf on Advances in Medical, Signal and Information Processing, MEDSIP 2008, p. 1-4 (2008).
15. National Institute of Child Health and Human Development Research Planning Workshop, Electronic fetal heart rate monitoring: Research guidelines for interpretation, American Journal of Obstetrics and Gynecology, Vol. 177(5), p. 1385 (1997).
16. M. G. Signorini, G. Margenes, S. Ceruttia and D. Arduini, Linear and nonlinear parameters for the analysis of fetal heart rate signal from cardiotocographic recordings, IEEE Trans Biomed Eng, Vol. 50(3), p. 365 (2003).
17. M. Godinez, A. Jimenez, M. R. Ortiz and M. Pena, On-line fetal heart rate monitor by phonocardiography, Proceedings of the 25th Annual Int Conf of the IEEE EMBS, Vol. 4, p. 3141, 17-21 Sept (2003).
18. D. M. Holburn and T. D. Rowsell, Real time analysis of fetal phonography signals using the TMS320, IEEE Colloquium on Biomedical Applications of Digital Signal Processing, Vol. 7/1 -7/12, 30 Nov (1989).
19. H. E. Bassil and J. H. Dripps, Real time processing and analysis of fetal phonocardiographic signals, Clin. Phys. Physiol. Meas., 10 (Suppl B), p. 67 (1989).
20. Y. Song, W. Xie, J. F. Chen and K. S. Phua, Passive Acoustic Maternal Abdominal Fetal Heart Rate Monitoring Using Wavelet Transform, Computers in Cardiology, Vol. 33, p. 581 (2006).
21. C. W. George Jr., The Fetal and Maternal Circulation of the Placenta, Clinical Obstetrics and Gynecology, Vol. 6(1), p. 17, March (1963).
22. M. El-Hanjouri, W. Alkhaldi, N. Hamdy and 0. A. Alim, Heart diseases diagnosis using HMM, Proceedings of the IEEE on Electrotechnical Conference (MELECON 2002), p. 489, 7-9 May (2002).
23. J. T. E. McDonnell, Knowledge-based interpretation of foetal phonocardiographic signals, IEE Proceedings, Vol. 137(5), p. 311 (1990).
24. H. G. Goovaerts, O. Rompelman and H. P. van Geijn, A transducer for detection of fetal breathing movements, IEEE Transactions on Biomedical Engineering, Vol. 36(4), p. 471, April (1989).
25. V. Padmanabhan, R. Fischer, J. L. Semmlow and W. Welkowitz, High Sensitivity PCG Transducer for Extended Frequency Applications, Proceedings of the Annual Int Conf of the IEEE EMBS, Images of the Twenty-First Century, Vol. 1, p. 57, 9-12 Nov (1989).
26. V. S. Chourasia and A. K Tiwari, Development of Wireless Data Acquisition System for Fetal Phonocardiographic signals using BluetoothTM, Submitted to Int Conference on Signal Processing and Communications (SPCOM2010), Bangalore, to be held on 18-21 July 2010.
27. D. Raskovic, T. Martin and E. Jovanov, Medical monitoring applications for wearable computing, The computer Journal, Vol. 47(4), p. 495 (2004).
28. X. Yin, P. Han, J. Zhang, F. Zhang and N. Wang, Application of wavelet transform in signal de-noising, Proceedings of the Second International Conference of the IEEE on Machine Learning and Cybernetics, Vol. 1, p. 436 (2003).
29. I. Daubechies, Ten lectures on wavelets, Society for Industrial and Applied Mathematics (SIAM), New York (1992).
30. M. R. Raghuveer and A. S. Bopardikar, Wavelet Transforms: Introduction to Theory and Applications, Pearson Education (Singapore) Pte. Ltd., Indian Branch, Third edition, p. 25 (2002).
31. S. G. Mallat, A Theory of Multiresolution Signal Decomposition: The Wavelet Representation, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 11(7), p. 674 (1989).
32. M. Misiti, Y. Misiti, G. Oppenheim and J. Poggi, MATLAB Wavelet Toolbox User's Guide, The MathWorks Inc (1997).
33. S. Degang, Y. Kai and T. Haifeng, Application of wavelet analysis in acoustic signal processing, The Eighth International Conference on Electronic Measurement and Instruments, p. 772 (2007).
34. V. S. Chourasia and A. K. Mittra, Wavelet-based de-noising of fetal phonocardiographic signals, Int. J. Medical Engineering and Informatics, Vol. 2(2), p. 139 (2010).
35. H. Liang, S. Lukkminen and I. Hartimo, A Heart Sound Segmentation Algorithm using Wavelet Decomposition and Reconstruction, Proc. of the 19th Annual Int Conf of the IEEE EMBS, Chicago, USA, p. 1630, Oct 30-Nov 2 (1997).
36. J. N. Torry, Heart Sound Analysis comparing Wavelet and Autoregressive Techniques, Computers in Cardiology, Vol. 30, p. 657 (2003).
37. H. Liang and I. Hartimo, A Heart Sound Feature Extraction Algorithm based on Wavelet Decomposition and Reconstruction, Proc of the 20th Annual Int Conf of the IEEE EMBS, Vol. 20(3), p. 1539 (1998).
38. A. Jimenez, M. R. Ortiz, M. A. Pena, S. Charleston, A. T. Aljama and R. Gonzalez, The Use of Wavelet Packets to Improve the Detection of Cardiac Sounds from the Fetal Phonocardiogram, Computers in Cardiology, Vol. 26, p. 463 (1999).
39. J. J. Lee, S. M. Lee, I. Y. Kim, H. K. Min and S. H. Hong, Comparison between short time Fourier and wavelet transform for feature extraction of heart sound, In: Proc. of IEEE TENCON 99, Cheju Island, South Korea, Vol. 2, p. 1547 (1999).
40. H. Liang, S. Lukkarinen and I. Hartimo, Heart sound segmentation algorithm based on heart sound envelogram, Computers in Cardiology, p. 105 (1997).
41. V. S. Chourasia and A. K. Mittra, Computer aided analysis and segmentation of fetal heart sound, Proc of the Int Conf on Biomedical Instrumentation and Health Care Engineering (ICOBIAHC'09), Chennai, p. 120, 8-10 Aug (2009).
42. Simulink, Simulation and model based design user's guide, Mathwork Incorporation; www.mathworks.com.
43. V. S. Chourasia and A. K. Tiwari, Development of a Signal Simulation Module for Testing of Phonocardiography Based Prenatal Monitoring Systems, Proceedings of 5th Annual conf of the IEEE (INDICON 2009), Ahmedabad, 18-20 Dec (2009).
44. E. C. Karvounis and D. I. Fotiadis, Maternal and fetal heart rate extraction from abdominal recordings using multi-scale principal components analysis, Proc of the 29th Annual Int Conf of the IEEE EMBS, Lyon, France, p. 6507, 22-26 August (2007).
45. H. Kieler, S. Cnattingius, B. Haglund, J. Palmgren and O. Axelsson, Ultrasound and Adverse effects, Ultrasound in Obstetrics and Gynaecology, Vol. 20(1), p. 102 (2002).