Phonocardiogram Signals Research Articles

Artificial neural network-based cardiovascular disease prediction using spectral features

The major cause of the increasing world mortality rate is cardiovascular disease (CVD), killing 17.9 million people annually. Current techniques are costly, challenging to operate on, and an expert is needed to confirm the diagnosis results. Phonocardiogram (PCG) signals are heart sound recordings of heart rhythms and have many advantages over traditional auscultation methods. This work targets CVD detection through PCG signal analysis using different artificial neural networks (ANN) and fusion of spectral features. PCG signal is acquired through the subject’s heart by a self-designed PCG acquisition setup. It is then pre-processed and extracted five spectral features with the highest pair-wise differences. Five different types of ANN named narrow, wide, tri-layered, bi-layered, and medium are simulated with 99.99% accuracy. This proposed architecture is non-invasive, moderate, and reliable compared to current approaches and also offers great guidance in offering new low-cost alternatives for CVD diagnosis techniques.

Ensembled Transfer Learning and Multiple Kernel Learning for Phonocardiogram Based Atherosclerotic Coronary Artery Disease Detection.

Conventional machine learning has paved the way for a simple, affordable, non-invasive approach for Coronary artery disease (CAD) detection using phonocardiogram (PCG). It leaves a scope to explore improvement of performance metrics by fusion of learned representations from deep learning. In this study, we propose a novel, multiple kernel learning (MKL) for their fusion using deep embeddings transferred from pre-trained convolutional neural network (CNN). The proposed MKL, finds optimal kernel combination by maximizing the similarity with ideal kernel and minimizing the redundancy with other basis kernels. Experiments are performed on 960 PCG epochs collected from 40 CAD and 40 normal subjects. The transferred embeddings attain maximum subject-level accuracy of 89.25% with kappa of 0.7850. Later, their fusion with handcrafted features using the proposed MKL gives an accuracy of 91.19% and kappa 0.8238. The study shows the potential of development of high accuracy CAD detection system by using easy to acquire, non-invasive PCG signal.

The Recognition Method of Athlete Exercise Intensity Based on ECG and PCG.

Athletes usually arrange their training plans and determine their training intensity according to the coach's experience and simple physical indicators such as heart rate during exercise. However, the accuracy of this method is poor, and the training plan and exercise intensity arranged according to this method can easily cause physical damage, or the training cannot meet the actual needs. Therefore, in order to realize the reasonable arrangement and monitoring of athletes' training, a method of human exercise intensity recognition based on ECG (electrocardiogram) and PCG (Phonocardiogram) is proposed. First, the ECG and PCG signals are fused into a two-dimensional image, and the dataset is marked and divided according to the different motion intensities. Then, the training set is trained with a CNN (convolutional neural network) to obtain the prediction model of the neural network. Finally, the neural network model is used to identify the ECG and PCG signals to judge the exercise intensity of the athlete, so as to adjust the training plan according to the exercise intensity. The recognition accuracy of the model on the dataset can reach 95.68%. Compared with the use of heart rate to detect the physical state during exercise, ECG records the total potential changes in the process of depolarization and repolarization of the heart, and PCG records the waveform of the beating sound of the heart, which contains richer feature information. Combined with the CNN method, the athlete's exercise intensity prediction model constructed by extracting the features of the athlete's ECG and PCG signals realizes the real-time monitoring of the athlete's exercise intensity and has high accuracy and generalization ability.

Unsupervised detection and classification of heartbeats using the dissimilarity matrix in PCG signals

Background and objective: Auscultation is the first technique applied to the early diagnose of any cardiovascular disease (CVD) in rural areas and poor-resources countries because of its low cost and non-invasiveness. However, it highly depends on the physician’s expertise to recognize specific heart sounds heard through the stethoscope. The analysis of phonocardiogram (PCG) signals attempts to segment each cardiac cycle into the four cardiac states (S1, systole, S2 and diastole) in order to develop automatic systems applied to an efficient and reliable detection and classification of heartbeats. In this work, we propose an unsupervised approach, based on time-frequency characteristics shown by cardiac sounds, to detect and classify heartbeats S1 and S2.Methods: The proposed system consists of a two-stage cascade. The first stage performs a rough heartbeat detection while the second stage refines the previous one, improving the temporal localization and also classifying the heartbeats into types S1 and S2. The first contribution is a novel approach that combines the dissimilarity matrix with the frame-level spectral divergence to locate heartbeats using the repetitiveness shown by the heart sounds and the temporal relationships between the intervals defined by the events S1/S2 and non-S1/S2 (systole and diastole). The second contribution is a verification-correction-classification process based on a sliding window that allows the preservation of the temporal structure of the cardiac cycle in order to be applied in the heart sound classification. The proposed method has been assessed using the open access databases PASCAL, CirCor DigiScope Phonocardiogram and an additional sound mixing procedure considering both Additive White Gaussian Noise (AWGN) and different kinds of clinical ambient noises from a commercial database.Results: The proposed method outperforms the detection and classification performance of other recent state-of-the-art methods. Although our proposal achieves the best average accuracy for PCG signals without cardiac abnormalities, 99.4% in heartbeat detection and 97.2% in heartbeat classification, its worst average accuracy is always above 92% for PCG signals with cardiac abnormalities, signifying an improvement in heartbeat detection/classification above 10% compared to the other state-of-the-art methods evaluated.Conclusions: The proposed method provides the best detection/classification performance in realistic scenarios where the presence of cardiac anomalies as well as different types of clinical environmental noises are active in the PCG signal. Of note, the promising modelling of the temporal structures of the heart provided by the dissimilarity matrix together with the frame-level spectral divergence, as well as the removal of a significant number of spurious heart events and recovery of missing heart events, both corrected by the proposed verification-correction-classification algorithm, suggest that our proposal is a successful tool to be applied in heart segmentation.

Deep learning approach of murmur detection using Cochleagram

Primary detection of heart abnormalities can reduce sudden death due to heart related diseases to a great extent. A low cost noninvasive phonocardiogram (PCG) signal based heart murmur detection system can serve the purpose partially. In this work, initially the heart sound signal is segmented using an unsupervised technique. After that, deep neural network (DNN) architectures with three hidden layers have been proposed to classify the input heart sound signal as normal or murmur class using 64 dimensional acoustic features. Total 3482 heart sound cycles from a standard database have been used to train and test the proposed DNN model. The usual evaluation matrices of sensitivity, specificity, precision, F1-score and accuracy have been derived to evaluate the performance of the classifier. After studying five different DNN models we have achieved the highest F1-score of 98.31% and accuracy of 98.33% for the murmur class. This murmur detection system can be extremely useful in rural health cares and small hospitals to assist the general physicians in assessing heart diseases without having expertise in cardiology.

A Low-Cost Multistage Cascaded Adaptive Filter Configuration for Noise Reduction in Phonocardiogram Signal.

Phonocardiogram (PCG), the graphic recording of heart signals, is analyzed to determine the cardiac mechanical function. In the recording of PCG signals, the major problem encountered is the corruption by surrounding noise signals. The noise-corrupted signal cannot be analyzed and used for advanced processing. Therefore, there is a need to denoise these signals before being employed for further processing. Adaptive Noise Cancellers are best suited for signal denoising applications and can efficiently recover the corrupted PCG signal. This paper introduces an optimal adaptive filter structure using a Sign Error LMS algorithm to estimate a noise-free signal with high accuracy. In the proposed filter structure, a noisy signal is passed through a multistage cascaded adaptive filter structure. The number of stages to be cascaded and the step size for each stage are adjusted automatically. The proposed Variable Stage Cascaded Sign Error LMS (SELMS) adaptive filter model is tested for denoising the fetal PCG signal taken from the SUFHS database and corrupted by Gaussian and colored pink noise signals of different input SNR levels. The proposed filter model is also tested for pathological PCG signals in the presence of Gaussian noise. The simulation results prove that the proposed filter model performs remarkably well and provides 8–10 dB higher SNR values in a Gaussian noise environment and 2-3 dB higher SNR values in the presence of colored noise than the existing cascaded LMS filter models. The MSE values are improved by 75–80% in the case of Gaussian noise. Further, the correlation between the clean signal and its estimate after denoising is more than 0.99. The PSNR values are improved by 7 dB in a Gaussian noise environment and 1-2 dB in the presence of pink noise. The advantage of using the SELMS adaptive filter in the proposed filter model is that it offers a cost-effective hardware implementation of Adaptive Noise Canceller with high accuracy.

Embedded platform based heart murmur classification using deep learning approach

Ubiquitous Perturbations in cardiac auscultation properties, cardiovascular diseases (CVDs) are widely recognized. In the auscultation procedure, the appearance of pathological cardiac murmurs is linked to heart disorders. A noble automated detection system using 1-D Convolutional Neural Network (CNN) for the detection of pathological heart murmurs is proposed in this study, which removes the difficult task of extracting and selecting features. It directly acts on the phonocardiogram (PCG) signals. The fundamental purpose of this research is to develop a classification model for consistent recognition of cardiac murmurs when the data-set is imbalanced. In view of this, the proposed study for the imbalanced data-set incorporates the Adaptive Synthetic (ADASYN) approach to generate synthetic data for the minority class. The outcome analysis illustrates the positive result in the identification of heart murmurs on both balanced and imbalanced data-sets. Therefore, the developed deep learning model will learn better from the minority class and classify heart murmurs accurately.

Cardi-Net: A deep neural network for classification of cardiac disease using phonocardiogram signal

Background and objectivesThe lack of medical facilities in isolated areas makes many patients remain aloof from quick and timely diagnosis of cardiovascular diseases, leading to high mortality rates. A deep learning based method for automatic diagnosis of multiple cardiac diseases from Phonocardiogram (PCG) signals is proposed in this paper. MethodsThe proposed system is a combination of deep learning based convolutional neural network (CNN) and power spectrogram Cardi-Net, which can extract deep discriminating features of PCG signals from the power spectrogram to identify the diseases. The choice of Power Spectral Density (PSD) makes the model extract highly discriminatory features significant for the multi-classification of four common cardiac disorders. ResultsData augmentation techniques are applied to make the model robust, and the model undergoes 10-fold cross-validation to yield an overall accuracy of 98.879% on the test dataset to diagnose multi heart diseases from PCG signals. ConclusionThe proposed model is completely automatic, where signal pre-processing and feature engineering are not required. The conversion time of power spectrogram from PCG signals is very low range from 0.10 s to 0.11 s. This reduces the complexity of the model, making it highly reliable and robust for real-time applications. The proposed architecture can be deployed on cloud and a low cost processor, desktop, android app leading to proper access to the dispensaries in remote areas.

Study of Amplifier Operating Point to Improve Phonocardiogram Signal Representation in a Custom-made Electronic Stethoscope

The performance of a custom-made electronic stethoscope needs to be improved through a study on the operating point of the amplifier. The aim of this study is to analyse the composition and content of the frequency components contained in the phonocardiogram signal by spectral extraction. The output is a voltage signal that represents the heart rate. The operating point of the amplifier is varied. The signal is recorded on a laptop and analysed. The results show that the operating point produces a balanced swing signal. Hence, the amplifier’s operating point does not affect the component spectrum in the recorded signal.

Effect of Hilbert-Huang transform on classification of PCG signals using machine learning

Heartbeat sounds are biological signals used in the early diagnosis of cardiovascular diseases. Digital heartbeat sound recordings, called phonocardiogram (PCG), are used in the determination and automatic classification of possible heart diseases. Healthy and pathological PCG signals are non-stationary signals and conventional feature extraction methods are insufficient in classifying these signals. In this study, PCG signals in healthy and four pathological categories are decomposed into intrinsic mode functions (IMFs) by Hilbert-Huang transform. Mel-frequency cepstral coefficient (MFCC) features were extracted from each mode to investigate the effect of the modes obtained by Hilbert-Huang transform on the classification of PCG signals. Genetic algorithm was used as feature selection method and k-nearest neighbor (KNN), multilayer perceptron (MLP), support vector machine (SVM) and deep neural network (DNN) machine learning methods were used as classifier. We have implemented multi classifications of five PCG classes (healthy, aortic stenosis, mitral stenosis, mitral regurgitation and mitral valve prolapse) by using 5-fold cross validation and 10 × 5-fold cross validation Data Analysis Protocol (DAP) framework. The results show that the DNN model provides the highest classification performance with 98.9% precision, 98.7% recall, 98.8% F1-score and 98.9% accuracy using 5-fold cross validation, and Matthews correlation coefficient of 0.981 using the DAP method.

Time–Frequency-Domain Deep Learning Framework for the Automated Detection of Heart Valve Disorders Using PCG Signals

The damage to the heart valves causes heart valve disorders (HVDs). The detection of HVDs is crucial in a clinical study as these diseases may cause congestive heart failure, hypertrophy, and stroke. The phonocardiogram (PCG) signal reveals information regarding the mechanical activity of the heart. The early detection of HVDs using PCG signal is vital to minimize the chances of cardiac arrest and other cardiac complications. This article proposes the time–frequency-domain deep learning (TFDDL) framework for automatic detection of HVDs using PCG signals. The time–frequency (TF)-domain representations of PCG signals are evaluated using both time-domain polynomial chirplet transform (TDPCT) and frequency-domain polynomial chirplet transform (FDPCT). The deep convolutional neural network (CNN) model is used to detect four types of HVDs using the TF images of PCG signals obtained using both the TDPCT and FDPCT methods. The proposed TFDDL approach is evaluated using PCG signals from public databases. For the detection of HVDs using TDPCT- and FDPCT-based TF images of PCG signals, the suggested approach has achieved overall accuracy values of 99% and 99.48%, respectively. For the classification of normal and abnormal heart sound classes, the proposed TFDDL approach has obtained an accuracy of 85.16% using PCG signals from the Physionet challenge 2016 database. The proposed TFDDL framework is compared with TF-domain transfer learning models such as residual network (ResNet-50) and visual geometry group (VGGNet-16). The overall accuracy values obtained using VGGNet-16 and ResNet-50 are less than the proposed deep CNN model for the detection of HVDs. The proposed TFDDL model can be validated in real-time using heart sound signals recorded from different subjects for automated identification of HVDs.

Automated Heart Sound Activity Detection From PCG Signal Using Time–Frequency-Domain Deep Neural Network

The phonocardiogram (PCG) signal deciphers the mechanical activity of the heart, and it consists of the fundamental heart sounds (S1 and S2), murmurs, and other associated sounds (S3 and S4). Detection of fundamental heart sound activity (FHSA) is vital for the automated analysis of PCG signals to diagnose various heart valve diseases. This paper proposes a time-frequency domain (TFD) deep neural network (DNN) approach for automated FHSA detection using PCG signals. The modified Gaussian window-based Stockwell Transform (MGWST) is used to obtain the time-frequency representation (TFR) of PCG signals. The Shannon-Teager-Kaiser Energy (STKE), smoothing, and thresholding techniques are then employed to evaluate the segmented heart sound components. The TFD Shannon entropy (TFDSE) features are computed from the segmented heart sound components of the PCG signal. The deep neural network (DNN) developed based on the stacked autoencoders (SAEs) is used for automated identification of FHSA components. The performance of the proposed approach is evaluated using two publicly available standard databases (Database 1: Michigan heart sound and murmur database and Database 2: PhysioNet Computing in Cardiology Challenge 2016). The results demonstrate that the proposed approach has achieved the accuracy, sensitivity, specificity, and precision values of 99.55 %, 99.93 %, 99.26 %, 99.02 % for Database 1, and 95.43 %, 97.92 %, 98.32 %, 97.60 % for Database 2, respectively. It is shown that, the proposed FHSA detection approach has obtained better accuracy than existing methods.

A Multi-Modal Identification System Based on Wavelet Scattering Network and Giza Pyramid Through Two-Level Fusion of Face and Phonocardiogram Signal (Pcg)

A Multi-Modal Identification System Based on Wavelet Scattering Network and Giza Pyramid Through Two-Level Fusion of Face and Phonocardiogram Signal (Pcg)

Heart Valve Diseases Detection Based on Feature-Fusion and Hierarchical LSTM Network

Early detection and diagnosis of heart valve diseases (HVDs) can prevent cardiac arrest. This work proposes a novel feature fusion method for detecting HVDs using a Phonocardiogram (PCG) signal. In the proposed method, first, the raw PCG signal is pre-processed. Then, two feature fusion models are proposed based on mel-frequency cepstral coefficients (MFCC) and linear prediction cepstral coefficients (LPCC). One model is the series feature fusion (SFF) model. Another is the parallel feature fusion (PFF) model. We propose a hierarchical long-short term memory (HLSTM) network with a self-attention mechanism for both models. HLSTM encodes the sequential information of the fused features in different abstractions. Then, the self-attention module aggregates these encoded vectors based on their clinical relevance to detect HVDs. The efficacy of the proposed method is evaluated using two publicly available databases offered by Physionet challenge (CinC) 2016 and the GitHub repository. The CinC 2016 database is used for binary classification. The results show an overall accuracy (OA) of 98.76%and 98.29%for binary classification using SFF-HLSTM and PFF-HLSTM models. The heart sound murmur (HSM) database in the GitHub repository is used for multi-class classification. The results show an OA of 99.10 % and 98.71 % for multi-class classification using the SFF-HLSTM and PFF-HLSTM models, respectively. The promising results show that the proposed method is compelling enough for preliminary automated diagnosis of HVDs in cardiac care units.

Automated classification of valvular heart diseases using FBSE-EWT and PSR based geometrical features

The growing prevalence and high mortality rate due to valvular heart diseases (VHD) are concerned. Therefore, its accurate, rapid, and early diagnosis is important. This study processes the phonocardiogram (PCG) signals for automatic detection of VHD associated with aortic stenosis (AS), mitral stenosis (MS), mitral valve prolapse (MVP) and mitral regurgitation (MR) which is originated by the MVP. In the proposed approach, we have used Fourier-Bessel series expansion-based empirical wavelet transform (FBSE-EWT) for capturing the non-stationary nature of PCG signals. It follows by selecting significant Fourier-Bessel intrinsic mode functions (FBIMFs) and extraction of geometrical features from two-dimensional phase space reconstruction (2D-PSRs) of the FBIMFs. To reduce the real-time processing load, the feature set dimension is reduced by using metaheuristics optimization-based features selection (FS) algorithms, namely, the salp swarm optimization algorithm (SSOA), emperor penguin optimization algorithm (EPOA), and tree growth optimization algorithm (TGOA). These FS methods have been tested and compared with machine learning classifiers. The result indicates the effectivity of FBSE-EWT based features extraction and used FS methods in classifying the intended categories of PCG signals. With the reduced features set, obtained with SSOA, the proposed approach has resulted in the highest classification accuracies of 98.53 %, 98.84 %, 99.07 % and 99.70 % respectively for the five classes, four classes, three classes, and two classes problems. Thus, besides aiding the cardiologists, this approach is also useful for developing wearable cardiac devices (as it uses a reduced feature set) for heart health monitoring and diagnosis purposes.

Computer Aided Diagnosis (CAD) untuk Phonocardiogram (PCG) Berbasis Fast Fourier Tranform

Data from WHO 2015 shows that 70% of deaths in the world are caused by non-infection diseases, 45% are caused by heart and blood vessel disease, namely 17.7 million from 39.5 million deaths. Riset Kesehatan Dasar (Riskesdas) reported that in 2018, there were 15 of 1000 people, or 2,784,064 individuals in Indonesia suffering from heart disease. Symptoms of heart abnormalities often come suddenly. Therefore, early recognition can help to avoid heart attacks. Doctors currently use a heart sound / phonocardiogram (PCG) to assess the performance of the heart using a stethoscope. The PCG’s diagnosis is very influenced by the subjectivity of doctors because of its relatively weak and physical limitations. So that the possibility of a False Positive Result happening is quite high. To minimize this risk, a Computer Aided Diagnosis (CAD) PCG signal was developed. Several studies have proposed a PCG diagnostic method using the wavelet or Welch method and based on Neural Network are more complex. In this study, a simple diagnosis method is proposed so that the computation is easier and faster with good accuracy. The PCG signal is amplified twice, then the Fast Fourier Transform (FFT) process is carried out to obtain the characteristics of the fundamental frequency and max amplitude. The classification stage uses the Multi Layer Perceptron (MLP). From testing of 55 data PCG, the results obtained accuracy of 90%, sensitivity of 80%, PPV of 100% and NPV of 83.33%.
 Keywords: PCG, CAD, FFT, frekuensi fundamental, classification

Abstract 14001: Using the S2 Sound as an Early Marker of Aortic Valve Calcification

Introduction: Heart valve diseases (HVDs) often remain undiagnosed until late stages, especially in underserved populations, leading to significant comorbidities. Artificial intelligence has improved murmur detection in digital phonocardiograms (PCG). However, the presence of murmurs depends on hemodynamic factors unrelated to HVD. Screening tests that directly measure features of valvular remodeling regardless of symptoms would improve HVD diagnosis. We used a mouse model of chronic kidney disease (CKD) induced aortic valve (AoV) remodeling to determine whether S2 sound changes can be used to detect early stages of AoV calcification. Methods: Eight-week old C57BL/6J mice (n=24) were assigned the following diets for 12 weeks: 1) control group fed a normal chow diet, 2) CKD group fed an adenine-supplemented diet to induce CKD, and 3) CKD+HP group fed the CKD diet for 6 weeks, then an adenine and high phosphate (HP)-supplemented diet for another 6 weeks to induce AoV calcification. PCG signals were recorded every 6 weeks. Endpoint AoV calcification and echocardiogram measures of valvular function were assessed. We developed a clustering method based on time domain principal component analysis of S2 sounds collected at week 6 (W6) to classify week 12 (W12) S2 sounds and identify S2 sound alterations due to HP. Results: The high phosphate diet induced AoV calcification in CKD+HP mice (1a). Of the total number of S2 sounds per mouse, an average of 84.18% and 81.15% S2 sounds (1b and 1c) were accurately classified as control or CKD mice with a specificity and sensitivity of 90.91% and 100%, respectively. CKD+HP mice had an average of 59.22% W12 S2 sounds that neither classified as control nor CKD. Conclusions: Mice that have AoV calcification show distinct S2 sounds features that were absent in the control and CKD groups. In this study, we demonstrate the feasibility of using the S2 sound not only to detect AoV calcification prior to expected hemodynamic changes but also to detect CKD.

Robust optimization of convolutional neural networks with a uniform experiment design method: a case of phonocardiogram testing in patients with heart diseases

BackgroundHeart sound measurement is crucial for analyzing and diagnosing patients with heart diseases. This study employed phonocardiogram signals as the input signal for heart disease analysis due to the accessibility of the respective method. This study referenced preprocessing techniques proposed by other researchers for the conversion of phonocardiogram signals into characteristic images composed using frequency subband. Image recognition was then conducted through the use of convolutional neural networks (CNNs), in order to classify the predicted of phonocardiogram signals as normal or abnormal. However, CNN requires the tuning of multiple hyperparameters, which entails an optimization problem for the hyperparameters in the model. To maximize CNN robustness, the uniform experiment design method and a science-based methodical experiment design were used to optimize CNN hyperparameters in this study.ResultsAn artificial intelligence prediction model was constructed using CNN, and the uniform experiment design method was proposed to acquire hyperparameters for optimal CNN robustness. The results indicate Filters ({X}_{1}), Stride ({X}_{3}), Activation functions ({X}_{6}), and Dropout ({X}_{7}) to be significant factors considerably influencing the ability of CNN to distinguish among heart sound states. Finally, the confirmation experiment was conducted, and the hyperparameter combination for optimal model robustness was Filters ({X}_{1}) = 32, Kernel Size ({X}_{2}) = 3 × 3, Stride ({X}_{3}) = (1,1), Padding ({X}_{4}) as same, Optimizer ({X}_{5}) as the stochastic gradient descent, Activation functions ({X}_{6}) as relu, and Dropout ({X}_{7}) = 0.544. With this combination of parameters, the model had an average prediction accuracy rate of 0.787 and standard deviation of 0.ConclusionIn this study, phonocardiogram signals were used for the early prediction of heart diseases. The science-based and methodical uniform experiment design was used for the optimization of CNN hyperparameters to construct a CNN with optimal robustness. The results revealed that the constructed model exhibited robustness and an acceptable accuracy rate. Other literature has failed to address hyperparameter optimization problems in CNN; a method is subsequently proposed for robust CNN optimization, thereby solving this problem.

Unsupervised Heart Sound Decomposition and State Estimation with Generative Oscillation Models.

This paper proposes a new generative probabilistic model for phonocardiograms (PCGs) that can simultaneously capture oscillatory factors and state transitions in cardiac cycles. Conventionally, PCGs have been modeled in two main aspects. One is a state space model that represents recurrent and frequently appearing state transitions. Another is a factor model that expresses the PCG as a non-stationary signal consisting of multiple oscillations. To model these perspectives in a unified framework, we combine an oscillation decomposition with a state space model. The proposed model can decompose the PCG into cardiac state dependent oscillations by reflecting the mechanism of cardiac sounds generation in an unsupervised manner. In the experiments, our model achieved better accuracy in the state estimation task compared to the empirical mode decomposition method. In addition, our model detected S2 onsets more accurately than the supervised segmentation method when distributions among PCG signals were different.

Toward Automated Analysis of Fetal Phonocardiograms: Comparing Heartbeat Detection from Fetal Doppler and Digital Stethoscope Signals.

Longitudinal fetal health monitoring is essential for high-risk pregnancies. Heart rate and heart rate variability are prime indicators of fetal health. In this work, we implemented two neural network architectures for heartbeat detection on a set of fetal phonocardiogram signals captured using fetal Doppler and a digital stethoscope. We test the efficacy of these networks using the raw signals and the hand-crafted energy from the signal. The results show a Convolutional Neural Network is the most efficient at identifying the S1 waveforms in a heartbeat, and its performance is improved when using the energy of the Doppler signals. We further discuss issues, such as low Signal-to-Noise Ratios (SNR), present in the training of a model based on the stethoscope signals. Finally, we show that we can improve the SNR, and subsequently the performance of the stethoscope, by matching the energy from the stethoscope to that of the Doppler signal.