Phonocardiogram Signals Research Articles

Is there an Optimal Localization of Cardio-microphone Sensors for Phonocardiogram Analysis?

Heart auscultation is one of the most useful medical diagnostic tools for getting valuable information of heart valves and heart hemodynamics functions. However, the information acquired by a traditional stethoscope can be inaccurate and insufficient. Phonocardiogram (PCG) was developed to improve accuracy through visual inspection and analysis. Digitally processed, PCG can then be analyzed by automated heart sound analysis systems. But there is no standardization for PCG data acquisition unlike electrocardiogram (ECG). This study aims at analyzing the influence of cardiomicrophone localization on the chest for the study of cardiac sounds S1 and S2. For that purpose, simultaneous acquisitions of 12 PCG signals with one ECG signal were realized and a comparative analysis of delays between R waves of ECG and detected S1 and S2 sounds was conducted. Results show that there are significant differences between R-S1 (or R-S2) intervals obtained from different areas of sensor placement on the chest. For future works on PCG, studies dealing with the analysis of heart sounds or proposing new heart sounds detection algorithms may pay attention to the location and attachment of PCG sensors.

CLASSIFICATION OF UNSEGMENTED HEART SOUND RECORDING USING KNN CLASSIFIER

Due to low physical workout, high-calorie intake, and bad behavioral character, people were affected by cardiological disorders. Every instant, one out of four deaths are due to heart-related ailments. Hence, the early diagnosis of a heart is essential. Most of the approaches for automated classification of the heart sound need segmentation of Phonocardiograms (PCG) signal. The main aim of this study was to decline the segmentation process and to estimate the utility for accurate and detailed classification of short unsegmented PCG recording. Based on wavelet decomposition, Hilbert transform, homomorphic filtering, and power spectral density (PSD), the features had been obtained using the beginning 5 second PCG recording. The extracted features were classified using nearest neighbors with Euclidean distances for different values of [Formula: see text] by bootstrapping 50% PCG recording for training and 50% for testing over 100 iterations. The overall accuracy of 100%, 85%, 80.95%, 81.4%, and 98.13% had been achieved for five different datasets using KNN classifiers. The classification performance for analyzing the whole datasets is 90% accuracy with 93% sensitivity and 90% specificity. The classification of unsegmented PCG recording based on an efficient feature extraction is necessary. This paper presents a promising classification performance as compared with the state-of-the-art approaches in short time less complexity.

Classifying Heart Sounds Using Images of Motifs, MFCC and Temporal Features.

Cardiovascular disease is the leading cause of death in the world, and its early detection is a key to improving long-term health outcomes. The auscultation of the heart is still an important method in the medical process because it is very simple and cheap. To detect possible heart anomalies at an early stage, an automatic method enabling cardiac health low-cost screening for the general population would be highly valuable. By analyzing the phonocardiogram signals, it is possible to perform cardiac diagnosis and find possible anomalies at an early-term. Therefore, the development of intelligent and automated analysis tools of the phonocardiogram is very relevant. In this work, we use simultaneously collected electrocardiograms and phonocardiograms from the Physionet Challenge database with the main objective of determining whether a phonocardiogram corresponds to a "normal" or "abnormal" physiological state. Our main contribution is the methodological combination of time domain features and frequency domain features of phonocardiogram signals to improve cardiac disease automatic classification. This novel approach is developed using both features. First, the phonocardiogram signals are segmented with an algorithm based on a logistic regression hidden semi-Markov model, which uses electrocardiogram signals as a reference. Then, two groups of features from the time and frequency domain are extracted from the phonocardiogram segments. One group is based on motifs and the other on Mel-frequency cepstral coefficients. After that, we combine these features into a two-dimensional time-frequency heat map representation. Lastly, a binary classifier is applied to both groups of features to learn a model that discriminates between normal and abnormal phonocardiogram signals. In the experiments, three classification algorithms are used: Support Vector Machines, Convolutional Neural Network, and Random Forest. The best results are achieved when both time and Mel-frequency cepstral coefficients features are considered using a Support Vector Machines with a radial kernel.

A Subject-Driven Unsupervised Hidden Semi-Markov Model and Gaussian Mixture Model for Heart Sound Segmentation

The analysis of heart sounds is a challenging task, due to the quick temporal onset between successive events and the fact that an important fraction of the information carried by phonocardiogram (PCG) signals lies in the inaudible part of the human spectrum. For these reasons, computer-aided analysis of the PCG can dramatically improve the quantity of information recovered from such signals. In this paper, a hidden semi-Markov model (HSMM) is used to automatically segment PCG signals. In the proposed models, the emission probability distributions are approximated via Gaussian mixture model (GMM) priors. The choice of GMM emission probability distributions allow to apply re-estimation routines to automatically adjust the HSMM emission probability distributions to each subject. Building on the proposed method for fine tuning emission distributions, a novel subject-driven unsupervised heart sound segmentation algorithm is proposed and validated over the publicly available PhysioNet dataset. Perhaps surprisingly, the proposed unsupervised method achieved results in line with state-of-the-art supervised approaches, when applied to long heart sounds.

Classification System for Heart Sounds Based on Random Forests

In the last century, cardiovascular illnesses are the first death cause in developed countries. For this reason, many efforts have been made in order to develop sophisticated techniques for the early diagnoses of cardiac disorders. The Phonocardiogram (PCG) signals contain very useful information about the condition of the heart. By analyzing these signals, early detection and diagnosis of heart diseases can be done. It is also very useful in the case of infants, where ECG recording and other techniques are difficult to implement. In this study, a classification method is proposed to classify normal and abnormal heart sound signals using random forests algorithm. The proposed framework was applied to a database of 100 heart sound signals which collected from the web site, firstly all the signals were processed using the wavelet technique to eliminate the noise from the signal, features were extracted from the enhanced signals and the most significant features was selected using the RFs. Finally the random forests classifier was used to perform the classification process. The system achieved 93.24% accuracy in distinguishing between normal and abnormal heart sound signals.

Classification of coronary artery diseased and normal subjects using multi-channel phonocardiogram signal

Coronary artery disease (CAD) is one of the leading causes of mortality and morbidity. There is a need to develop a simple, reliable, and non-invasive screening tool to diagnose CAD. Prior studies reported that turbulent blood flow due to stenosed coronary arteries causes weak CAD murmurs. Analysis of phonocardiogram (PCG) signals can be useful to detect these murmurs. In this work, we propose a new multi-channel PCG-based system to classify CAD-affected and normal subjects, and it does not require any additional reference signal, such as an electrocardiogram (ECG) signal. The proposed system simultaneously acquires PCG signals from four different auscultation sites on the chest. It extracts five different features from time and frequency domains of the PCG signals. The two-class classification is done in a machine learning framework by employing an artificial neural network (ANN) classifier. The classification performances are evaluated for each channel as well as for their combinations. Experimental results show that the proposed sub-band-based spectral features perform well for both clean and noisy data. An accuracy of 82.57% is obtained using the combination of the signals acquired from tricuspid, mitral, and midaxillary regions. The multi-channel system gives more than 4% relative improvement over the best performance obtained by its single-channel counterpart, and the proposed features outperform earlier used features.

Dual-Input Neural Network Integrating Feature Extraction and Deep Learning for Coronary Artery Disease Detection Using Electrocardiogram and Phonocardiogram

Electrocardiogram (ECG) and phonocardiogram (PCG) signals reflect the electrical and mechanical activities of the heart, respectively. Although studies have documented that some abnormalities in ECG and PCG signals are associated with coronary artery disease (CAD), only few researches have combined the two signals for automatic CAD detection. This paper aims to differentiate between CAD and non-CAD groups using simultaneously collected ECG and PCG signals. To entirely exploit the underlying information in these signals, a novel dual-input neural network that integrates the feature extraction and deep learning methods is developed. First, the ECG and PCG features are extracted from multiple domains, and the information gain ratio is used to select important features. On the other hand, the ECG signal and the decomposed PCG signal (at four scales) are concatenated as a five-channel signal. Then, the selected features and the five-channel signal are fed into the proposed network composed of a fully connected model and a deep learning model. The results show that the classification performance of either feature extraction or deep learning is insufficient when using only ECG or PCG signal, and combining the two signals improves the performance. Further, when using the proposed network, the best result is obtained with accuracy, sensitivity, specificity, and G-mean of 95.62%, 98.48%, 89.17%, and 93.69%, respectively. Comparisons with existing studies demonstrate that the proposed network can effectively capture the combined information of ECG and PCG signals for the recognition of CAD.

Elimination of Interference in Phonocardiogram Signal Based on Wavelet Transform and Empirical Mode Decomposition

The recording of the phonocardiogram (PCG) signal is an important biological index in the diagnosis and treatment in the area of cardiac diseases and disorders. However, the signal is often corrupted by the wide range of artifacts which affect the quality of the PCG waveform and complicate the process of the diagnosis determination. The study focuses on the well-known signal processing approaches (wavelet transform and empirical mode decomposition) used for elimination of the overlapping interference in PCG signal and evaluates the accuracy of methods by assessment of the power of the remaining noise and comparing a filtered PCG signal with an original one based on correlation and Bland-Altman analysis.

A new heart sounds segmentation approach based on the correlation between ECG and PCG signals

During the cardiac cycle, both electrical and mechanical events are present. In fact, the electrocardiogram (ECG) allows the exploration of the electrical activity of the heart, while the phonocardiogram (PCG) constitutes a complementary tool to record the mechanical activity of the heart. The aim of our work is to develop a new approach for heart sounds segmentation based on the correlation between ECG and PCG signals for measurement of systolic time interval. Two different groups of simultaneous recordings of ECG and PCG signals were tested in this study. The performance of automatic heart sounds localisation process is compared with that of manual measurements made by two experts. The results obtained are interesting especially for pathological cases heart sound.

Variability of Cardiac Electromechanical Delay With Application to the Noninvasive Detection of Coronary Artery Disease

Heart rate variability (HRV), systolic period variability (SPV), and diastolic period variability (DPV) have shown potential for assessing cardiac function. It is unknown whether the time delay between the myocardial electrical and mechanical activities (i.e., electromechanical delay, EMD) also possesses variability, and if it does, whether this EMD variability (EMDV) could render additional value for cardiac function assessment. In this paper, we extracted the beat-to-beat EMD from 5-min simultaneously recorded electrocardiogram and phonocardiogram signals in 30 patients with coronary artery disease (CAD) and 30 healthy control subjects, and studied its variability using the same methods as applied for HRV including time-domain measures [mean and standard deviation (SD)], frequency-domain measures [normalized low- and high-frequency (LFn, HFn) and LF/HF)], and nonlinear measures [sample entropy (SampEn), permutation entropy (PE), and dynamical patterns]. In addition, we examined whether the addition of EMDV could offer improved performance for distinguishing between the two groups compared to using the HRV, SPV, and DPV features. Support vector machine with 10-fold cross-validation was used for classification. Results showed increased SD of SPV, increased mean, SD and decreased SampEn of EMDV in CAD patients. Besides, the dynamical pattern analysis showed that CAD patients had significantly increased fluctuated patterns and decreased monotonous patterns in EMDV. In particular, the addition of EMDV indices dramatically increased the classification accuracy from 0.729 based on HRV, SPV, and DPV features to 0.958. Our results suggest promising of the EMDV analysis that could potentially be helpful for detecting CAD noninvasively.

A Comparative Analysis of Fetal Phonocardiograph Acoustical Performance

This paper deals with comparative analysis of elementary parts of fetal phonocardiograph by means of time-frequency analysis of fetal phonocardiogram (fPCG) signal. Elementary parts of fetal phonocardiograph are tested, analyzed and subsequently evaluated using synthetic data and a phantom of the pregnant woman abdomen. Based on these results, the optimal configuration of fetal phonocardiograph was designed and tested on a pregnant woman at 34th weeks of pregnancy.

Classification of Heart Sound Signal Using Multiple Features

Cardiac disorders are critical and must be diagnosed in the early stage using routine auscultation examination with high precision. Cardiac auscultation is a technique to analyze and listen to heart sound using electronic stethoscope, an electronic stethoscope is a device which provides the digital recording of the heart sound called phonocardiogram (PCG). This PCG signal carries useful information about the functionality and status of the heart and hence several signal processing and machine learning technique can be applied to study and diagnose heart disorders. Based on PCG signal, the heart sound signal can be classified to two main categories i.e., normal and abnormal categories. We have created database of 5 categories of heart sound signal (PCG signals) from various sources which contains one normal and 4 are abnormal categories. This study proposes an improved, automatic classification algorithm for cardiac disorder by heart sound signal. We extract features from phonocardiogram signal and then process those features using machine learning techniques for classification. In features extraction, we have used Mel Frequency Cepstral Coefficient (MFCCs) and Discrete Wavelets Transform (DWT) features from the heart sound signal, and for learning and classification we have used support vector machine (SVM), deep neural network (DNN) and centroid displacement based k nearest neighbor. To improve the results and classification accuracy, we have combined MFCCs and DWT features for training and classification using SVM and DWT. From our experiments it has been clear that results can be greatly improved when Mel Frequency Cepstral Coefficient and Discrete Wavelets Transform features are fused together and used for classification via support vector machine, deep neural network and k-neareast neighbor(KNN). The methodology discussed in this paper can be used to diagnose heart disorders in patients up to 97% accuracy. The code and dataset can be accessed at “https://github.com/yaseen21khan/Classification-of-Heart-Sound-Signal-Using-Multiple-Features-/blob/master/README.md”.

Automatic Identification of S1 and S2 Heart Sounds Using Simultaneous PCG and PPG Recordings

Accurate and reliable identification of the first (S1) and second (S2) sounds of the phonocardiogram (PCG) is still a challenging task due to the presence of the S3 and S4, murmurs, high-pitched sounds, physiological interference, and other environmental noises. This paper presents an automated method for identification of the S1 and S2 sounds by simultaneously recording, processing, and fusing the extracted fiducial points of the PCG and photoplethysmogram (PPG) signals. The method consists of four main stages: the PCG and PPG signal sensing with Arduino interfacing, the heart sound delineation (HSD) and pulse waveform delineation (PWD) using variational mode decomposition (VMD), and the S1/S2 identification based on the estimated timings of the pulse onset, peak, upward slope zerocrossing, and downward slope zerocrossing of the pulsatile waveform. The accuracy and robustness of the HSD and PWD algorithms are evaluated using both the benchmark databases and simultaneous PCG and PPG recordings. Then, the proposed S1/S2 identification method is evaluated using the simultaneous PCG and PPG recordings. The VMD-based HSD algorithm achieves an average sensitivity (SE) = 99.80%, positive predictivity (PP) = 99.0%, and accuracy (ACC) = 98.81% whereas the Gaussian derivative filter (GDF)-based HSD and bandpass filter (BPF)-based HSD algorithms achieve the average SE = 99.75%, PP = 93.59%, and ACC = 93.37% and the average SE = 99.67%, PP = 84.97%, and ACC = 84.73%, respectively, for the recorded PCG signals with SNR value of 5 dB. The proposed S1/S2 identification method achieves a correct identification rate of 99.75% which outperforms the GDF+systole interval(SI)/diastole interval(DI) and the BPF+SI/DI feature-based methods having 66% and 59.25%, respectively, in the case of noisy recordings with a SNR value of 5 dB. The proposed method has great potential in improving the identification accuracy and robustness in the presence of other sounds and murmurs, and different kinds of environmental noises.

Spectral and SNR improvement analysis of normal and abnormal heart sound signals using different windows

The primary function of a coronary care unit is to monitor the rhythm of cardiac patients. The cardiac abnormality of the patients should be identified during the early stage. The heart sound recoding is the phonocardiogram (PCG) signal. Stethoscopes are commonly used to hear the heart sound for diagnosis. In common, the heart murmurs can be caused by a high blood pressure, heart attack, pregnancy, rheumatic fever, anemia or thyrotoxicosis. This study analyzes PCG signals spectrally with different windows with signal-to-noise ratio (SNR) computations. The two PCG signals of normal and abnormal subjects in each were analyzed in this study. The normal subject produce clear “lub” and “dub” sounds where the abnormal subject produce a kind of whistling or swishing sound in middle which may be due to the problem of septal defect in a heart. The spectral leakage reduction by different windows helps to increase the SNR value. Simulation results of SNR with different windows are obtained using Matlab R2016b tool.

Validation of beat by beat fetal heart signals acquired from four-channel fetal phonocardiogram with fetal electrocardiogram in healthy late pregnancy

Fetal heart rate monitoring is an essential obstetric procedure, however, false-positive results cause unnecessary obstetric interventions and healthcare cost. In this study, we propose a low cost and non-invasive fetal phonocardiography based signal system to measure the fetal heart sounds and fetal heart rate. Phonocardiogram (PCG) signals contain acoustic information reflecting the contraction and relaxation of the heart. We have developed a four-channel recording device with four separated piezoelectric sensors harnessed by a cloth sheet to record abdominal phonogram signals. A multi-lag covariance matrix based eigenvalue decomposition technique was used to extract fetal and maternal heart sounds as well as maternal breathing movement. In order to validate the fetal heart sounds extracted by PCG signal processing, 10 minutes’ simultaneous recordings of fetal Electrocardiogram (fECG) and abdominal phonogram from 15 pregnant women (27 ± 5-year-old) with fetal gestation ages between 33 and 40 weeks were obtained and processed. Highly significant (p < 0.01) correlation (r = 0.96; N = 270) was found between beat to beat fetal heart rate (FHRECG) from fECG and the same (FHRPCG) from fetal PCG signals. Bland–Altman plot of FHRECG and FHRPCG shows good agreement (<5% difference). We conclude that the proposed beat to beat fetal heart rate measurement system would be useful for monitoring fetal neurological wellbeing as a better alternative to traditional cardiotocogram based antenatal fetal heart rate monitoring.

Phonocardiogram Signal Denoising Based on Nonnegative Matrix Factorization and Adaptive Contour Representation Computation

—This letter introduces a new technique for phono-cardiogram (PCG) signal denoising based non-negative matrix factorization (NMF) of its spectrogram and adaptive contour representation computation (ACRC) of its short-time Fourier transform (STFT). More precisely, NMFs on PCG and synchronous electrocardiogram (ECG) spectrograms are first used to filter out high-energy noises from PCG. Then, ACRC is performed on a low-pass filtered version of the STFT of the resulting signal to identify relevant time-frequency (TF) components which are subsequently used for signal retrieval. Numerical experiments conducted on a real database of noisy PCG signals (SiSEC2016) illustrate the superiority of the proposed method over state-of-the-art techniques.

A PHONOCARDIOGRAM-BASED NOISE-ROBUST REAL-TIME HEART RATE MONITORING ALGORITHM FOR OUTPATIENTS DURING NORMAL ACTIVITIES

For outpatients who need continuous monitoring of heart rate (HR) variation, it is important that HR can be monitored during normal activities such as speaking and walking. In this study, a noise-robust real-time HR monitoring algorithm based on phonocardiogram (PCG) signals is proposed. PCG signals were recorded using an electronic stethoscope; electrocardiogram (ECG) signals were recorded simultaneously with HR references. The proposed algorithm consisted of pre-processing, peak/nonpeak classification, voice noise processing, walking noise processing, and HR calculation. The performance of the algorithm was evaluated using PCG/ECG signals from 11 healthy participants. For comparison, the absolute errors between manually extracted ECG-based HR values and automatically calculated PCG-based HR values were calculated for the proposed algorithm and the comparison algorithm in two different test protocols. Experimental results showed that the average absolute errors of the proposed algorithm were 72.03%, 22.92%, and 36.39% of the values of the comparison algorithm for resting-state, speaking-state, and walking-state data, respectively, in protocol-1. In protocol-2, the average absolute error was 36.99% of that of the comparison algorithm. A total of 1102 cases in protocol-1 and 783 in protocol-2 had an absolute error [Formula: see text] beats per minute (BPM) using the comparison algorithm and an absolute error [Formula: see text] BPM using the proposed algorithm. On the basis of these results, we anticipate that the proposed algorithm can potentially improve the performance of continuous real-time HR monitoring during activities of normal life, thereby improving the safety of outpatients with cardiovascular diseases.

Analysis of PCG signals using quality assessment and homomorphic filters for localization and classification of heart sounds

Background and Objective: Accurate localization of heart beats in phonocardiogram (PCG) signal is very crucial for correct segmentation and classification of heart sounds into S1 and S2. This task becomes challenging due to inclusion of noise in acquisition process owing to number of different factors. In this paper we propose a system for heart sound localization and classification into S1 and S2. The proposed system introduces the concept of quality assessment before localization, feature extraction and classification of heart sounds. Methods: The signal quality is assessed by predefined criteria based upon number of peaks and zero crossing of PCG signal. Once quality assessment is performed, then heart beats within PCG signal are localized, which is done by envelope extraction using homomorphic envelogram and finding prominent peaks. In order to classify localized peaks into S1 and S2, temporal and time-frequency based statistical features have been used. Support Vector Machine using radial basis function kernel is used for classification of heart beats into S1 and S2 based upon extracted features. The performance of the proposed system is evaluated using Accuracy, Sensitivity, Specificity, F-measure and Total Error. The dataset provided by PASCAL classifying heart sound challenge is used for testing. Results: Performance of system is significantly improved by quality assessment. Results shows that proposed Localization algorithm achieves accuracy up to 97% and generates smallest total average error among top 3 challenge participants. The classification algorithm achieves accuracy up to 91%. Conclusion: The system provides firm foundation for the detection of normal and abnormal heart sounds for cardiovascular disease detection.

Detection of Third Heart Sound Using Variational Mode Decomposition

In this paper, a novel approach for the separation of heart and lung sounds (HLS) is proposed based on the nonlinear decomposition technique, variational mode decomposition (VMD) of phonocardiogram signal. Subjective validation and objective quantification are used to assess the performance of the method. The proposed technique is found to perform better than the conventional algorithms such as empirical mode decomposition (EMD), ensemble EMD, and different variants of the EMD. As the third heart sound, $S_{3}$ is an important clinical sign of cardiac failure in elderly patients, a new technique is proposed for its detection. This method is built on the VMD and the smoothed pseudo Wigner–Ville distribution. In total, 40 sets of cardiac cycles containing $S_{3}$ are obtained from publicly available databases and are used to evaluate the performance of the proposed method in noiseless condition. It is able to detect the $S_{3}$ correctly even when the normalized amplitude of $S_{3}$ is 14.1%, whereas the existing method based on the Hilbert variation decomposition requires at least 16.13% of the normalized amplitude of $S_{3}$ in comparison to the normalized amplitude of the highest peak present in the cardiac cycle. In addition to this, the result shows that the proposed method can detect $S_{3}$ in noisy cases up to SNR level of −5 dB.

Segmentation of Phonocardiograms Signal

Heart sound is a kind of bio-sound, mainly through the media pass sound signals. The measures of the heart sound signals involved in acoustics, fluid mechanics research. when added some noise for selecting pure PCG signal by using adaptive white Gaussian noise. Than denoise the PCG signal by using Discrete Wavelet transform and decomposes a signal into a 4 level of basic functions. These basic functions are called Discrete wavelet transform (DWT), which transforms a discrete time signal to a discrete wavelet representation. Experiments are conducted on 23 different recordings of heart sound where experiment working on 2 normal signal and 19 Abnormal signals where working on the accuracy of the PCG signal that is depending on the training and testing data. Segmentation process based on Shannon entropy method for low amplitude and GSF is based on high amplitude. Toward this objective, after preprocessing the PCG signal, for feature extraction of the PCG signal fixed windows were moved on the preprocessed signal, and in each analysis window, two frequency-and amplitude-based features were calculated from the excerpted segment. In order to recognize the delineated PCG sounds, ?rst, S1 and S2 were detected. Then, a new DS was regenerated from the signal whose S1 and S2 were eliminated to detect occasional S3 and S4 sounds. Finally, probable murmurs and souf?es were spotted. The proposed algorithm was applied to 6 beat PCG. signals gathered from patients with different valve diseases. This feature Extraction of the PCG signal is use PCA for reducing 216 dimensions to 7 dimensions of the PCG signal.&nbsp;and the classifier is used SVM method which is found the normal and abnormal heart sound of the PCG signal.