Heart Sound Classification Research Articles

A novel 1-D densely connected feature selection convolutional neural network for heart sounds classification.

BackgroundHeart sound auscultation, due to it being a non-invasive, convenient, and relatively low-cost technique, remains the dominant method for detection of cardiovascular disease.MethodsIn this paper, we present a method for identifying abnormal heart sounds based on a novel Dense Feature Selection Convolution Network framework (Dense-FSNet). The Dense-FSNet is comprised of multiple, circular dense connectivity modules, called Clique Blocks. These Clique Blocks can allow low-level and high-level features to stimulate each other for cyclic updates, which subsequently enhances the information flow among layers. Inspired by the channel-wise attention mechanism from Squeeze-and-Excitation Networks (SENet), we adopted squeeze-and-excitation block to avoid the progressive growth of parameters. The effect of the model was assessed using the accuracy, specificity, sensitivity, and area under the curve (AUC) values. To improve model performance, in addition to the structures mentioned above, we incorporated a multi-scale attention mechanism into our framework.ResultsUsing this attention mechanism, our model was able to achieve adaptively spatial feature fusion by adjusting a hyper-feature that contains higher level visual information and lower-level features including edge details and context information. It is worth noting that data balance technology was also used in the process of building the model, and notable results have been achieved.ConclusionsExperience using the PhysioNet/CinC 2016 dataset shows that our proposed Dense-FSNet models achieve state of the art levels in the classification and detection of abnormal heart sounds.

Abstract 12262: The Role of Machine Learning-Based Decision Support Tools for Diagnosing and Assessing Congenital Heart Disease

Introduction: Machine learning (ML) is frequently applied in healthcare to optimize diagnostic classification through discovery of hidden patterns in data, not obvious to the human brain. ML research in CHD entails one of the most promising clinical applications, where timely and accurate diagnosis is essential. The objective of this scoping review was to summarize the accuracy and role of ML techniques used in pediatric cardiology research, specifically focusing on approaches that optimize diagnosis and assessment of underlying CHD. Methods: All original, peer-reviewed studies published in PubMed between January 2015 and February 2021 describing “(machine learning) AND ((congenital heart disease) OR (cardiovascular disease in children))” for predicting diagnostic outcomes in children with critical and non-critical CHD were included. An independent reviewer extracted and summarized primary data elements using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews guidelines. Results: This search yielded 217 journal articles of which 48 full-text articles (22%) met inclusion criteria. The most common application included optimizing the diagnosis and assessment of critical and non-critical CHD (38%). Deep learning (48%) and support vector machine (19%) were the most common ML algorithms used, with more interpretable algorithms being less common including decision trees (5%). Studies focusing on optimizing diagnostic classification reported a pooled overall diagnostic accuracy ranging from of 0.864 to 0.955, but only two of these models were externally validated. Clinical applications primarily focused on the classification of auscultatory heart sounds, transthoracic echocardiograms, and cardiac magnetic resonance images. Conclusions: These findings indicate that ML is a very promising tool for diagnosing and assessing critical and non-critical CHD, yet extensive research is still needed to build interpretable, robust, and generalizable models for clinical use, especially considering the extreme heterogeneity of complex CHD.

Heart sound classification using signal processing and machine learning algorithms

According to global statistics and the world health organization (WHO), about 17.5 million people die each year from cardiovascular disease. In this paper, the heart sounds gathered by a stethoscope are analyzed to diagnose several diseases caused by heart failure. This research’s primary process is to identify and classify the data related to the heart sounds categorized in four general groups of S1 to S4. The sounds S1 and S2 are considered as the heart’s normal sounds, and the sounds S3 and S4 are the abnormal sounds of the heart (heart murmurs), each expressing a specific type of heart disease. In this regard, the desired features are first extracted after retrieving the data by signal processing algorithms. In the next step, feature selection algorithms are used to select the compelling features to reduce the problem’s dimensions and obtain the optimal answer faster. While the existing algorithms in the literature classify the sound into two groups of normal and abnormal, in the final section, some of the most popular classification algorithms are utilized to classify the type of sound into three classes of normal, S3 and S4 categories. The proposed methodology obtained an accuracy rate of 87.5% and 95% for multiclass data (3 classes) and 98% for binary classification (normal vs. abnormal) problems.

Transferring Cross-Corpus Knowledge: An Investigation on Data Augmentation for Heart Sound Classification.

Human auscultation has been regarded as a cheap, convenient and efficient method for the diagnosis of cardiovascular diseases. Nevertheless, training professional auscultation skills needs tremendous efforts and is time-consuming. Computer audition (CA) that leverages the power of advanced machine learning and signal processing technologies has increasingly attracted contributions to the field of automatic heart sound classification. While previous studies have shown promising results in CA based heart sound classification with the 'shuffle split' method, machine learning for heart sound classification decreases in accuracy with a cross-corpus test dataset. We investigate this problem with a cross-corpus evaluation using the PhysioNet CinC Challenge 2016 Dataset and propose a new combination of data augmentation techniques that leads to a CNN robust for such cross-corpus evaluation. Compared with the baseline, which is given without augmentation, our data augmentation techniques combined improve by 20.0 % the sensitivity and by 7.9 % the specificity on average across 6 databases, which is a significant difference on 4 out of these (p < .05 by one-tailed z-test).

The role of machine learning applications in diagnosing and assessing critical and non-critical CHD: a scoping review.

Machine learning uses historical data to make predictions about new data. It has been frequently applied in healthcare to optimise diagnostic classification through discovery of hidden patterns in data that may not be obvious to clinicians. Congenital Heart Defect (CHD) machine learning research entails one of the most promising clinical applications, in which timely and accurate diagnosis is essential. The objective of this scoping review is to summarise the application and clinical utility of machine learning techniques used in paediatric cardiology research, specifically focusing on approaches aiming to optimise diagnosis and assessment of underlying CHD. Out of 50 full-text articles identified between 2015 and 2021, 40% focused on optimising the diagnosis and assessment of CHD. Deep learning and support vector machine were the most commonly used algorithms, accounting for an overall diagnostic accuracy > 0.80. Clinical applications primarily focused on the classification of auscultatory heart sounds, transthoracic echocardiograms, and cardiac MRIs. The range of these applications and directions of future research are discussed in this scoping review.

Classification of real-world pathological phonocardiograms through multi-instance learning.

Heart auscultation is an inexpensive and fundamental technique to effectively to diagnose cardiovascular disease. However, due to relatively high human error rates even when auscultation is performed by an experienced physician, and due to the not universal availability of qualified personnel e.g. in developing countries, a large body of research is attempting to develop automated, computational tools for detecting abnormalities in heart sounds. The large heterogeneity of achievable data quality and devices, the variety o possible heart pathologies, and a generally poor signal-to-noise ratio make this problem extremely challenging. We present an accurate classification strategy for diagnosing heart sounds based on 1) automatic heart phase segmentation, 2) state-of-the art filters drawn from the filed of speech synthesis (mel-frequency cepstral representation), and 3) an ad-hoc multi-branch, multi-instance artificial neural network based on convolutional layers and fully connected neuronal ensembles which separately learns from each heart phase, hence leveraging their different physiological significance. We demonstrate that it is possible to train our architecture to reach very high performances, e.g. an AUC of 0.87 or a sensitivity of 0.97. Our machine-learning-based tool could be employed for heart sound classification, especially as a screening tool in a variety of situations including telemedicine applications.

Unsegmented Heart Sound Classification Using Hybrid CNN-LSTM Neural Networks.

Cardiac Auscultation, an integral part of the physical examination of a patient, is essential for early diagnosis of cardiovascular diseases (CVDs). The ability to accurately diagnose the heart sounds requires experience and expertise, which is lacking in doctors in the early years of clinical practice. Thus, there is a need for an automatic diagnostic tool that would aid medical practitioners with their diagnosis. We propose novel hybrid architectures for classification of unsegmented heart sounds to normal and abnormal classes. We propose two methods, with and without the conventional feature extraction step in the classification pipeline. We demonstrate that the F score using the approach with conventional feature extraction is 1.25 (absolute) more than using a baseline implementation on the Physionet dataset. We also introduce a mechanism to tag predictions as unsure and compare results with a varying threshold.

Segmentation-free Heart Pathology Detection Using Deep Learning.

Cardiovascular (CV) diseases are the leading cause of death in the world, and auscultation is typically an essential part of a cardiovascular examination. The ability to diagnose a patient based on their heart sounds is a rather difficult skill to master. Thus, many approaches for automated heart auscultation have been explored. However, most of the previously proposed methods involve a segmentation step, the performance of which drops significantly for high pulse rates or noisy signals. In this work, we propose a novel segmentation-free heart sound classification method. Specifically, we apply discrete wavelet transform to denoise the signal, followed by feature extraction and feature reduction. Then, Support Vector Machines and Deep Neural Networks are utilised for classification. On the PASCAL heart sound dataset our approach showed superior performance compared to others, achieving 81% and 96% precision on normal and murmur classes, respectively. In addition, for the first time, the data were further explored under a user-independent setting, where the proposed method achieved 92% and 86% precision on normal and murmur, demonstrating the potential of enabling automatic murmur detection for practical use.

Do we really need a segmentation step in heart sound classification algorithms?

Cardiac auscultation is the key screening procedure to detect and identify cardiovascular diseases (CVDs). One of many steps to automatically detect CVDs using auscultation, concerns the detection and delimitation of the heart sound boundaries, a process known as segmentation. Whether to include or not a segmentation step in the signal classification pipeline is nowadays a topic of discussion. Up to our knowledge, the outcome of a segmentation algorithm has been used almost exclusively to align the different signal segments according to the heartbeat. In this paper, the need for a heartbeat alignment step is tested and evaluated over different machine learning algorithms, including deep learning solutions. From the different classifiers tested, Gate Recurrent Unit (GRU) Network and Convolutional Neural Network (CNN) algorithms are shown to be the most robust. Namely, these algorithms can detect the presence of heart murmurs even without a heartbeat alignment step. Furthermore, Support Vector Machine (SVM) and Random Forest (RF) algorithms require an explicit segmentation step to effectively detect heart sounds and murmurs, the overall performance is expected drop approximately 5% on both cases.

A simple proposition for heart sound signal de-noising for effective components identification in normal and abnormal cases

Mass scale, simple, efficient, minimum infrastructure-based engineering approach with an inbuilt detection capability of Heart Sound Signal (HSS) can save many lives due to cardiovascular diseases in a rural area of any developing country lacking super medical facilities and expert medical practitioners. De-noising of HSS is a crucial step for reliable and accurate identification sound components such as S1, S2, and others as HSS gets easily corrupted with random internal and external noise sources. In recent years, Wavelet-based filtering methods are the most common for de-noising of HSS for which selection of correct wavelet, level of decomposition, the best threshold selection are crucial for which excellent domain expertise is inevitable. Though these methods provide excellent returns in high-frequency noise/murmurs elimination, low-frequency noise/murmurs in abnormal HSS cannot be detected adequately as the noise/murmur spectrum overlaps with HSS component. This paper introduces a simple but very effective Power Law Algorithm (PLA) that performs de-noising in the time domain for efficient Heart Sound component identification both in normal and abnormal cases. A comparative study with qualitative (based on the visual parameter) and quantitative analysis (based on SNR and FIT coefficient) on de-noised HSSs has been performed that clearly shows PLA based method improves quality components identification than using wavelet thresholding. Further, time-domain feature parameters and power spectral density representation of the PLA-based de-noised signal exhibits a definite indication of the algorithm's efficacy towards heart sound classification both for Normal or Abnormal classes.

Heart sound classification based on sub-band envelope and convolution neural network

Automatic classification of heart sounds plays an important role in the early diagnosis of congenital heart disease. A kind of heart sound classification algorithms based on sub-band envelope feature and convolution neural network was proposed in this paper, which did not need to segment the heart sounds according to cardiac cycle accurately. Firstly, the heart sound signal was divided into some frames. Then, the frame level heart sound signal was filtered with Gammatone filter bank to obtain the sub-band signals. Next, the sub-band envelope was extracted by Hilbert transform. After that, the sub-band envelope was stacked into a feature map. Finally, type Ⅰ and type Ⅱ convolution neural network were selected as classifier. The result shown that the sub-band envelope feature was better in type Ⅰ than type Ⅱ. The algorithm is tested with 1 000 heart sound samples. The test results show that the overall performance of the algorithm proposed in this paper is significantly improved compared with other similar algorithms, which provides a new method for automatic classification of congenital heart disease, and speeds up the process of automatic classification of heart sounds applied to the actual screening.

A novel multi-branch architecture for state oftheartrobust detection ofpathological phonocardiograms.

Heart auscultation is an inexpensive and fundamental technique to effectively diagnose cardiovascular disease. However, due to relatively high human error rates even when auscultation is performed by an experienced physician, and due to the not universal availability of qualified personnel, e.g. in developing countries, many efforts are made worldwide to propose computational tools for detecting abnormalities in heart sounds. The large heterogeneity of achievable data quality and devices, the variety of possible heart pathologies, and a generally poor signal-to-noise ratio make this problem very challenging. We present an accurate classification strategy for diagnosing heart sounds based on (1) automatic heart phase segmentation, (2) state-of-the art filters drawn from the field of speech synthesis (mel-frequency cepstral representation) and (3) an ad hoc multi-branch, multi-instance artificial neural network based on convolutional layers and fully connected neuronal ensembles which separately learns from each heart phase hence implicitly leveraging their different physiological significance. We demonstrate that it is possible to train our architecture to reach very high performances, e.g. an area under the curve of 0.87 or a sensitivity of 0.97. Our machine-learning-based tool could be employed for heartsound classification, especially as a screening tool in a variety of situations including telemedicine applications. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

Classification of heart sounds based on quality assessment and wavelet scattering transform

Automatic classification of heart sound plays an important role in the diagnosis of cardiovascular diseases. In this study, a heart sound sample classification method based on quality assessment and wavelet scattering transform was proposed. First, the ratio of zero crossings (RZC) and the root mean square of successive differences (RMSSD) were used for assessing the quality of heart sound signal. The first signal segment conforming to the threshold standard was selected as the current sample for the continuous heart sound signal. Using the wavelet scattering transform, the wavelet scattering coefficients were expanded according to the wavelet scale dimension, to obtain the features. Support vector machine (SVM) was used for classification, and the classification results for the samples were obtained using the wavelet scale dimension voting approach. The effects of RZC and RMSSD on the results are discussed in detail. On the database of PhysioNet Computing in Cardiology Challenge 2016 (CinC 2016), the proposed method yields 92.23% accuracy (Acc), 96.62% sensitivity (Se), 90.65% specificity (Sp), and 93.64% measure of accuracy (Macc). The results show that the proposed method can effectively classify normal and abnormal heart sound samples with high accuracy.

Heart sound classification based on log Mel-frequency spectral coefficients features and convolutional neural networks

In view of the important role of heart sound signals in diagnosing and preventing congenital heart disease, a novel method about feature extraction and classification of heart sound signals was put forward in this study. Firstly, the heart sound signals were de-noised by using the wavelet algorithm. Subsequently, the improved duration-dependent hidden Markov model (DHMM) was used to segment the heart sound signal according to the heart cycle. Then, the dynamic frame length method was used to extract log Mel-frequency spectral coefficients (MFSC) features from the heart sound signal based on the heart cycle. Afterward, the convolution neural network (CNN) was used to classify the MFSC features. Finally, the majority voting algorithm was used to get the optimal classification results. In this paper, two-classification and multi-classification models were built. An accuracy of 93.89% for two-classification and an accuracy of 86.25% for multi-classification were achieved using the novel method.

Deep Learning Methods for Heart Sounds Classification: A Systematic Review.

The automated classification of heart sounds plays a significant role in the diagnosis of cardiovascular diseases (CVDs). With the recent introduction of medical big data and artificial intelligence technology, there has been an increased focus on the development of deep learning approaches for heart sound classification. However, despite significant achievements in this field, there are still limitations due to insufficient data, inefficient training, and the unavailability of effective models. With the aim of improving the accuracy of heart sounds classification, an in-depth systematic review and an analysis of existing deep learning methods were performed in the present study, with an emphasis on the convolutional neural network (CNN) and recurrent neural network (RNN) methods developed over the last five years. This paper also discusses the challenges and expected future trends in the application of deep learning to heart sounds classification with the objective of providing an essential reference for further study.

Abnormal heart sound classification using phonocardiography signals

An intelligent support system is needed to assist in the identification of abnormalities of a human heart. The integration of signal processing with machine learning techniques is a new research trend in the studies of heart sound analysis. This paper proposes a heart sound feature dimension reduction and classification methods using supervised machine learning algorithms, by utilising the first (S1) and the second (S2) heart sounds, produced due to vibrations during the closure of heart valves. The features of S1 and S2 heart sounds are extracted in both time and frequency domains. Time domain features are based on S1 and S2 sound distance, amplitude, sound peak area, sound peak cycle duration and intensity, whilst 20 Mel Frequency Cepstral Coefficients (MFCCs) filter-bank energy for 12 coefficients represent the frequency domain features. Statistical values of the selected features are further used to increase the number of heart sound features. Due to the size of the extracted features, Linear Discriminant Analysis (LDA) dimensionality reduction technique has been used to select the best features for normal and abnormal heart sound classification using an Artificial Neural Network (ANN) model. It has been shown that the proposed LDA/ANN heart sound classification model achieved 90%, 83.33%, and 93.33% classification accuracies using the time domain, frequency domain and combined time-frequency domain features, respectively. The results using the proposed method are significantly better than previous classification methods by other researchers, with minimal complexity. This work provides a step forward in providing clinical informatics tool to assist clinician in providing early detections of abnormal heart conditions.

MACHINE-ASSISTED DIAGNOSIS OF CHD BASED ON VARIANT LOGIC THEORY AND MACHINE LEARNING MODELS

Cardiac auscultation is a basic means of initial diagnosis of congenital heart disease (CHD). It is significant to analyze Phonocardiogram (PCG) by using signal processing and machine learning techniques for the purpose of machine-assisted diagnosis of CHD. A novel method of machine-assisted diagnosis of CHD was proposed in this paper. First, the duration of the hidden Markov model (DHMM) was applied to locate and segment PCG into each cardiac cycle. Then, the envelope of the PCG was extracted by using Viola integral. After that, the variant logic theory was applied to extract the features and convert the envelope data of the heart sound signal into visual analysis measurement data. Finally, the classifier of the support vector machine (SVM) was used to classify the normal and abnormal heart sounds. There were 1000 cases used in this study. It was divided into a training set of 600 cases, a test set of 200 cases, and validation set of 200 cases. An accuracy of 0.965, a specificity of 0.898, and a sensitivity of 0.937 were achieved using the novel signal processing techniques. The results showed that DHMM and variant logic theory models were suitable for heart sound classification.

Heart sounds classification using convolutional neural network with 1D-local binary pattern and 1D-local ternary pattern features

Cardiovascular diseases are one of those diseases which should be considered to be serious health threats for humans since today it is seen that majority of deaths are caused by heart disorders. There are advanced techniques such as electrocardiograph (ECG), computerized tomography in order to detect heart disorders. However, these methods are usually costly and time-consuming. Besides, auscultating the heart sounds by doctors for the detection of deformation of the heart sounds is an alternative method. Nevertheless, analysis of heart sounds performed by auscultating the patient mostly depends on the practitioner’s expertise and skills. Therefore, record and analysis of heart sounds via computers is a preferred method. In this paper, a new heart sound classification model is proposed based on Local Binary Pattern (LBP) and Local Ternary (LTP) Pattern features and deep learning. In the first step of the model, One Dimensional Local Binary Pattern (1D-LBP) and One Dimensional Local Ternary Pattern (1D-LTP) are used to extract features from signals. In the second step, the features obtained as a result of the application of 1D-LBP and 1D-LTP are combined and a hybrid feature vector is obtained from a signal. In the third step, ReliefF based feature selection method is used to identify the most relevant features. Finally, the most discriminative features are given as input to the designed One Dimensional Convolutional Neural Network (1D-CNN) in order to complete the classification process. Experiments are made with the help of two popular datasets that are used in the context to determine the efficiency of distinct techniques. These datasets are PASCAL and PhysioNet 2016. We obtain 91.66% and 91.78% accuracy rates for the PASCAL and PhysioNet 2016 datasets respectively. As a consequence, when we compare our results that we obtain from the proposed method in this study with those of the up-to-date methods’ results used in the literature, it is obvious that our technique surpasses them according to the classification accuracy rate.

Heart sound recognition technology based on convolutional neural network

ABSTRACT The mortality rate of heart disease continues to rise each year: developing mechanisms to reduce mortality from heart disease is a top concern in today’s society. Heart sound auscultation is a crucial skill used to detect and diagnose heart disease. In this study, we propose a heart sound signal classification algorithm based on a convolutional neural network. The algorithm is based on heart sound data collected in the clinic and from medical books. The heart sound signals were first preprocessed into a grayscale image of 5 seconds. The training samples were then used to train and optimize the convolutional neural network; obtaining a training result with an accuracy of 95.17% and a loss value of 0.23. Finally, the convolutional neural network was used to test the test set samples. The results showed an accuracy of 94.80%, sensitivity of 94.29%, specificity of 95.54%, precision of 93.44%, F1_score of 93.84%, and an AUC of 0.943. Compared with other algorithms, the accuracy and sensitivity of the algorithms were improved. This shows that the method used in this study can effectively classify heart sound signals and could prove useful in assisting heart sound auscultation.

An innovative method for cardiovascular disease detection based on nonlinear geometric features and feature reduction combination

Cardiovascular is arguably the most dominant death cause in the world. Heart functionality can be measured in various ways. Heart sounds are usually inspected in these experiments as they can unveil a variety of heart related diseases. This study tackles the lack of reliable models and high training times on a publicly available dataset. The heart sound set is provided by Physionet consisting of 3153 recordings, from which five seconds were fixed to evaluate to the developed method. In this work, we propose a novel method based on feature reduction combination, using Genetic Algorithm (GA) and Principal Component Analysis (PCA). The authors present eight dominant features in heart sound classification: mean duration of systole interval, the standard deviation of diastole interval, the absolute amplitude ratio of diastole to S2, S1 to systole and S1 to diastole, zero crossings, Centroid to Centroid distance (CCdis) and mean power in the 95–295 Hz range. These reduced features are then optimized respectively with two straightforward classification algorithms weighted k-NN with a lower-dimensional feature space and Linear SVM that uses a linear combination of all features to create a robust model, acquiring up to 98.15% accuracy, holding the best stats in the heart sound classification on a largely used dataset. According to the experiments done in this study, the developed method can be further explored for real world heart sound assessments.