A classifier based on the artificial neural network approach for cardiologic auscultation in pediatrics

Heart Murmur Classification Using a Capsule Neural Network

The healthcare industry has made significant progress in the diagnosis of heart conditions due to the use of intelligent detection systems such as electrocardiograms, cardiac ultrasounds, and abnormal sound diagnostics that use artificial intelligence (AI) technology, such as convolutional neural networks (CNNs). Over the past few decades, methods for automated segmentation and classification of heart sounds have been widely studied. In many cases, both experimental and clinical data require electrocardiography (ECG)-labeled phonocardiograms (PCGs) or several feature extraction techniques from the mel-scale frequency cepstral coefficient (MFCC) spectrum of heart sounds to achieve better identification results with AI methods. Without good feature extraction techniques, the CNN may face challenges in classifying the MFCC spectrum of heart sounds. To overcome these limitations, we propose a capsule neural network (CapsNet), which can utilize iterative dynamic routing methods to obtain good combinations for layers in the translational equivariance of MFCC spectrum features, thereby improving the prediction accuracy of heart murmur classification. The 2016 PhysioNet heart sound database was used for training and validating the prediction performance of CapsNet and other CNNs. Then, we collected our own dataset of clinical auscultation scenarios for fine-tuning hyperparameters and testing results. CapsNet demonstrated its feasibility by achieving validation accuracies of 90.29% and 91.67% on the test dataset.

Design of ear-contactless stethoscope and improvement in the performance of deep learning based on CNN to classify the heart sound

Cardiac-related disorders are rapidly growing throughout the world. Accurate classification of cardiovascular diseases is an important research topic in healthcare. During COVID-19, auscultating heart sounds was challenging as health workers and doctors wear protective clothing, and direct contact with patients can spread the outbreak. Thus, contactless auscultation of heart sound is necessary. In this paper, a low-cost ear contactless stethoscope is designed where auscultation is done with the help of a bluetooth-enabled micro speaker instead of an earpiece. The PCG recordings are further compared with other standard electronic stethoscopes like Littman 3 M. This work is made to improve the performance of deep learning-based classifiers like recurrent neural networks (RNN) and convolutional neural networks (CNN) for different valvular heart problems using tuning of hyperparameters like learning rate of optimizers, dropout rate, and hidden layer. Hyper-parameter tuning is used to optimize the performances of various deep learning models and their learning curves for real-time analysis. The acoustic, time, and frequency domain features are used in this research. The investigation is made on the heart sounds of normal and diseased patients available from the standard data repository to train the software models. The proposed CNN-based inception network model achieved an accuracy of 99.65 ± 0.06% on the test dataset with a sensitivity of 98.8  ± 0.05% and specificity of 98.2 ± 0.19%. The proposed hybrid CNN-RNN architecture attained 91.17 ± 0.03% accuracy on test data after hyperparameter optimization, whereas the LSTM-based RNN model achieved 82.32 ± 0.11% accuracy. Finally, the evaluated results were compared with machine learning algorithms, and the improved CNN-based Inception Net model is the most effective among others.

Beyond Heart Murmur Detection: Automatic Murmur Grading From Phonocardiogram

OBJECTIVE

Murmurs are abnormal heart sounds, identified by experts through cardiac auscultation. The murmur grade, a quantitative measure of the murmur intensity, is strongly correlated with the patient's clinical condition. This work aims to estimate each patient's murmur grade (i.e., absent, soft, loud) from multiple auscultation location phonocardiograms (PCGs) of a large population of pediatric patients from a low-resource rural area.

METHODS

The Mel spectrogram representation of each PCG recording is given to an ensemble of 15 convolutional residual neural networks with channel-wise attention mechanisms to classify each PCG recording. The final murmur grade for each patient is derived based on the proposed decision rule and considering all estimated labels for available recordings. The proposed method is cross-validated on a dataset consisting of 3456 PCG recordings from 1007 patients using a stratified ten-fold cross-validation. Additionally, the method was tested on a hidden test set comprised of 1538 PCG recordings from 442 patients.

RESULTS

The overall cross-validation performances for patient-level murmur gradings are 86.3% and 81.6% in terms of the unweighted average of sensitivities and F1-scores, respectively. The sensitivities (and F1-scores) for absent, soft, and loud murmurs are 90.7% (93.6%), 75.8% (66.8%), and 92.3% (84.2%), respectively. On the test set, the algorithm achieves an unweighted average of sensitivities of 80.4% and an F1-score of 75.8%.

CONCLUSIONS

This study provides a potential approach for algorithmic pre-screening in low-resource settings with relatively high expert screening costs.

SIGNIFICANCE

The proposed method represents a significant step beyond detection of murmurs, providing characterization of intensity, which may provide an enhanced classification of clinical outcomes.

Heart sounds: Past, present, and future from a technological and clinical perspective - a systematic review

The high prevalence of cardiac diseases around the world has created a need for quick, easy and cost effective approaches to diagnose heart disease. The auscultation and interpretation of heart sounds using the stethoscope is relatively inexpensive, requires minimal to advanced training, and is widely available and easily carried by healthcare providers working in urban environments or medically underserved rural areas. Since René-Théophile-Hyacinthe Laennec's simple, monoaural design, the capabilities of modern-day, commercially available stethoscopes and stethoscope systems have radically advanced with the integration of electronic hardware and software tools, however these systems are largely confined to the metropolitan medical centers. The purpose of this paper is to review the history of stethoscopes, compare commercially available stethoscope products and analytical software, and discuss future directions. Our review includes a description of heart sounds and how modern software enables the measurement and analysis of time intervals, teaching auscultation, remote cardiac examination (telemedicine) and, more recently, spectrographic evaluation and electronic storage. The basic methodologies behind modern software algorithms and techniques for heart sound preprocessing, segmentation and classification are described to provide awareness.

CNN-Based Heart Sound Classification with an Imbalance-Compensating Weighted Loss Function

Heart sound auscultation is an effective method for early-stage diagnosis of heart disease. The application of deep neural networks is gaining increasing attention in automated heart sound classification. This paper proposes deep Convolutional Neural Networks (CNNs) to classify normal/abnormal heart sounds, which takes two-dimensional Mel-scale features as input, including Mel frequency cepstral coefficients (MFCCs) and the Log Mel spectrum. We employ two weighted loss functions during the training to mitigate the class imbalance issue. The model was developed on the public PhysioNet/Computing in Cardiology Challenge (CinC) 2016 heart sound database. On the considered test set, the proposed model with Log Mel spectrum as features achieves an Unweighted Average Recall (UAR) of 89.6%, with sensitivity and specificity being 89.5% and 89.7% respectively. This work proposes a CNN-based model to enable automated heart sound classification, which can provide auxiliary assistance for heart auscultation and has the potential to screen for heart pathologies in clinical applications at a relatively low cost.

Diagnostic Accuracy of Machine Learning Models to Identify Congenital Heart Disease: A Meta-Analysis

Background: With the dearth of trained care providers to diagnose congenital heart disease (CHD) and a surge in machine learning (ML) models, this review aims to estimate the diagnostic accuracy of such models for detecting CHD. Methods: A comprehensive literature search in the PubMed, CINAHL, Wiley Cochrane Library, and Web of Science databases was performed. Studies that reported the diagnostic ability of ML for the detection of CHD compared to the reference standard were included. Risk of bias assessment was performed using Quality Assessment for Diagnostic Accuracy Studies-2 tool. The sensitivity and specificity results from the studies were used to generate the hierarchical Summary ROC (HSROC) curve. Results: We included 16 studies (1217 participants) that used ML algorithm to diagnose CHD. Neural networks were used in seven studies with overall sensitivity of 90.9% (95% CI 85.2-94.5%) and specificity was 92.7% (95% CI 86.4-96.2%). Other ML models included ensemble methods, deep learning and clustering techniques but did not have sufficient number of studies for a meta-analysis. Majority (n=11, 69%) of studies had a high risk of patient selection bias, unclear bias on index test (n=9, 56%) and flow and timing (n=12, 75%) while low risk of bias was reported for the reference standard (n=10, 62%). Conclusion: ML models such as neural networks have the potential to diagnose CHD accurately without the need for trained personnel. The heterogeneity of the diagnostic modalities used to train these models and the heterogeneity of the CHD diagnoses included between the studies is a major limitation.

Detection of Pathologic Heart Murmurs Using a Piezoelectric Sensor

This study aimed to evaluate the capability of a piezoelectric sensor to detect a heart murmur in patients with congenital heart defects. Heart sounds and murmurs were recorded using a piezoelectric sensor and an electronic stethoscope in healthy neonates (n = 9) and in neonates with systolic murmurs caused by congenital heart defects (n = 9) who were born at a hospital. Signal data were digitally filtered by high-pass filtering, and the envelope of the processed signals was calculated. The amplitudes of systolic murmurs were evaluated using the signal-to-noise ratio and compared between healthy neonates and those with congenital heart defects. In addition, the correlation between the amplitudes of systolic murmurs recorded by the piezoelectric sensor and electronic stethoscope was determined. The amplitudes of systolic murmurs detected by the piezoelectric sensor were significantly higher in neonates with congenital heart defects than in healthy neonates (p < 0.01). Systolic murmurs recorded by the piezoelectric sensor had a strong correlation with those recorded by the electronic stethoscope (ρ = 0.899 and p < 0.01, respectively). The piezoelectric sensor can detect heart murmurs objectively. Mechanical improvement and automatic analysis algorithms are expected to improve recording in the future.

Deep Layer Kernel Sparse Representation Network for the Detection of Heart Valve Ailments from the Time-Frequency Representation of PCG Recordings

The heart valve ailments (HVAs) are due to the defects in the valves of the heart and if untreated may cause heart failure, clots, and even sudden cardiac death. Automated early detection of HVAs is necessary in the hospitals for proper diagnosis of pathological cases, to provide timely treatment, and to reduce the mortality rate. The heart valve abnormalities will alter the heart sound and murmurs which can be faithfully captured by phonocardiogram (PCG) recordings. In this paper, a time-frequency based deep layer kernel sparse representation network (DLKSRN) is proposed for the detection of various HVAs using PCG signals. Spline kernel-based Chirplet transform (SCT) is used to evaluate the time-frequency representation of PCG recording, and the features like L1-norm (LN), sample entropy (SEN), and permutation entropy (PEN) are extracted from the different frequency components of the time-frequency representation of PCG recording. The DLKSRN formulated using the hidden layers of extreme learning machine- (ELM-) autoencoders and kernel sparse representation (KSR) is used for the classification of PCG recordings as normal, and pathology cases such as mitral valve prolapse (MVP), mitral regurgitation (MR), aortic stenosis (AS), and mitral stenosis (MS). The proposed approach has been evaluated using PCG recordings from both public and private databases, and the results demonstrated that an average sensitivity of 100%, 97.51%, 99.00%, 98.72%, and 99.13% are obtained for normal, MVP, MR, AS, and MS cases using the hold-out cross-validation (CV) method. The proposed approach is applicable for the Internet of Things- (IoT-) driven smart healthcare system for the accurate detection of HVAs.

A primer on artificial intelligence for the paediatric cardiologist

The combination of pediatric cardiology being both a perceptual and a cognitive subspecialty demands a complex decision-making model which makes artificial intelligence a particularly attractive technology with great potential. The prototypical artificial intelligence system would autonomously impute patient data into a collaborative database that stores, syncs, interprets and ultimately classifies the patient's profile to specific disease phenotypes to compare against a large aggregate of shared peer health data and outcomes, the current medical body of literature and ongoing trials to offer morbidity and mortality prediction, drug therapy options targeted to each patient's genetic profile, tailored surgical plans and recommendations for timing of sequential imaging. The focus of this review paper is to offer a primer on artificial intelligence and paediatric cardiology by briefly discussing the history of artificial intelligence in medicine, modern and future applications in adult and paediatric cardiology across selected concentrations, and current barriers to implementation of these technologies.

Artificial Neural Network Trained to Predict High-Harmonic Flux

In this work we present the results obtained with an artificial neural network (ANN) which we trained to predict the expected output of high-order harmonic generation (HHG) process, while exploring a multi-dimensional parameter space. We argue on the utility and efficiency of the ANN model and demonstrate its ability to predict the outcome of HHG simulations. In this case study we present the results for a loose focusing HHG beamline, where the changing parameters are: the laser pulse energy, gas pressure, gas cell position relative to focus and medium length. The physical quantity which we predict here using ANN is directly related to the total harmonic yield in a specified spectral domain (20–40 eV). We discuss the versatility and adaptability of the presented method.

Principal component analysis-based features generation combined with ellipse models-based classification criterion for a ventricular septal defect diagnosis system

In this study, a simple and efficient diagnostic system, which adopts a novel methodology consisting of principal component analysis (PCA)-based feature generation and ellipse models-based classification criterion, is proposed for the diagnosis of a ventricular septal defect (VSD). The three stages corresponding to the diagnostic system implementation are summarized as follows. In stage 1, the heart sound is collected by 3M-3200 electronic stethoscope and is preprocessed using the wavelet decomposition. In stage 2, the PCA-based diagnostic features, [[Formula: see text]], are generated from time-frequency feature matrix ([Formula: see text]). In the matrix TFFM, the time domain features [Formula: see text] are firstly extracted from the time domain envelope [Formula: see text] for the filtered heart sound signal [Formula: see text], and frequency domain features, [Formula: see text], are subsequently extracted from a frequency domain envelope ([Formula: see text]) for each heart sound cycle automatically segmented via the short time modified Hilbert transform (STMHT). In stage 3, support vector machines-based classification boundary curves for the dataset [Formula: see text] are first generated, and least-squares-based ellipse models are subsequently built for the classification boundary curve. Finally, based on the ellipse models, the classification criterion is defined for the diagnosis of VSD sounds. The proposed diagnostic system is validated by sounds from the internet and by sounds from clinical heart diseases. Moreover, comparative analysis to validate the usefulness of the proposed diagnostic system, mitral regurgitation and aortic stenosis sounds are used as examples for detection. As a result, the higher classification accuracy, which is achieved by this study compared to the other methods, is [Formula: see text], [Formula: see text], [Formula: see text] and [Formula: see text] for diagnosing small VSD, moderate VSD, large VSD and normal sounds, respectively.

Estimating pressure gradients by auscultation: How technology (echocardiography) can help improve clinical skills

AIM

To extend our previously-published experience in estimating pressure gradients (PG) via physical examination in a large patient cohort.

METHODS

From January 1, 1997 through December 31, 2009, an attending pediatric cardiologist compared clinical examination (EXAM) with Doppler-echo (ECHO), in 1193 patients with pulmonic stenosis (PS, including tetralogy of Fallot), aortic stenosis (AS), and ventricular septal defect (VSD). EXAM PG estimates were based primarily on a murmur’s pitch, grade, and length. ECHO peak instantaneous PG was derived from the modified Bernoulli equation. Patients were 0-38.4 years old (median 4.8).

RESULTS

For all patients, EXAM correlated highly with ECHO: ECHO = 0.99 (EXAM) + 3.2 mmHg; r = +0.89; P < 0.0001. Agreement was excellent (mean difference = -2.9 ± 16.1 mmHg). In 78% of all patients, agreement between EXAM and ECHO was within 15 mmHg and within 5 mmHg in 45%. Clinical estimates of PS PG were more accurate than of AS and VSD. A palpable precordial thrill and increasing loudness of the murmur predicted higher gradients (P < 0.0001). Weight did not influence accuracy. A learning curve was evident, such that the most recent quartile of patients showed ECHO = 1.01 (EXAM) + 1.9, r = +0.92, P < 0.0001; during this time, the attending pediatric cardiologist had been > 10 years in practice.

CONCLUSION

Clinical examination can accurately estimate PG in PS, AS, or VSD. Continual correlation of clinical findings with echocardiography can lead to highly accurate diagnostic skills.

Initial Field Test of a Cloud-Based Cardiac Auscultation System to Determine Murmur Etiology in Rural China

A system for collection, distribution, and long distant, asynchronous interpretation of cardiac auscultation has been developed and field-tested in rural China. We initiated a proof-of-concept test as a critical component of design of a system to allow rural physicians with little experience in evaluation of congenital heart disease (CHD) to obtain assistance in diagnosis and management of children with significant heart disease. The project tested the hypothesis that acceptable screening of heart murmurs could be accomplished using a digital stethoscope and internet cloud transmittal to deliver phonocardiograms to an experienced observer. Of the 7993 children who underwent school-based screening in the Menghai District of Yunnan Province, Peoples Republic of China, 149 had a murmur noted by a screener. They had digital heart sounds and phonocardiograms collected with the HeartLink tele auscultation system, and underwent echocardiography by a cardiology resident from the First Affiliated Hospital of Kunming Medical University. The digital phonocardiograms, stored on a cloud server, were later remotely reviewed by a board-certified American pediatric cardiologist. Fourteen of these subjects were found to have CHD confirmed by echocardiogram. Using the HeartLink system, the pediatric cardiologist identified 11 of the 14 subjects with pathological murmurs, and missed three subjects with atrial septal defects, which were incorrectly identified as venous hum or Still's murmur. In addition, ten subjects were recorded as having pathological murmurs, when no CHD was confirmed by echocardiography during the field study. The overall test accuracy was 91% with 78.5% sensitivity and 92.6% specificity. This proof-of-concept study demonstrated the feasibility of differentiating pathologic murmurs due to CHD from normal functional heart murmurs with the HeartLink system. This field study is an initial step to develop a cost-effective CHD screening strategy in low-resource settings with a shortage of trained medical professionals and pediatric heart programs.

An open access database for the evaluation of heart sound algorithms

In the past few decades, analysis of heart sound signals (i.e. the phonocardiogram or PCG), especially for automated heart sound segmentation and classification, has been widely studied and has been reported to have the potential value to detect pathology accurately in clinical applications. However, comparative analyses of algorithms in the literature have been hindered by the lack of high-quality, rigorously validated, and standardized open databases of heart sound recordings. This paper describes a public heart sound database, assembled for an international competition, the PhysioNet/Computing in Cardiology (CinC) Challenge 2016. The archive comprises nine different heart sound databases sourced from multiple research groups around the world. It includes 2435 heart sound recordings in total collected from 1297 healthy subjects and patients with a variety of conditions, including heart valve disease and coronary artery disease. The recordings were collected from a variety of clinical or nonclinical (such as in-home visits) environments and equipment. The length of recording varied from several seconds to several minutes. This article reports detailed information about the subjects/patients including demographics (number, age, gender), recordings (number, location, state and time length), associated synchronously recorded signals, sampling frequency and sensor type used. We also provide a brief summary of the commonly used heart sound segmentation and classification methods, including open source code provided concurrently for the Challenge. A description of the PhysioNet/CinC Challenge 2016, including the main aims, the training and test sets, the hand corrected annotations for different heart sound states, the scoring mechanism, and associated open source code are provided. In addition, several potential benefits from the public heart sound database are discussed.

Paediatric heart sound signal analysis towards classification using multifractal spectra

Healthy versus unhealthy heart sound computer-aided classification tools are very popular for supporting clinical decisions. In this paper a new method is proposed for the classification of heart sound recordings from a statistical standpoint without detection and localization of fundamental heart sounds (S1, S2). This study analyzes the possibility of detecting healthy heart sound signal from a large set of measurements, corresponding to different pathologies, such as aortic regurgitation, mitral regurgitation, aortic stenosis and ventricular septal defects. The proposed method employs singularity spectra analysis and long-term dependency of irregular structures. Healthy signals are firstly separated from the rest of the recordings. In the second step, the signals with a click syndrome, used here as a reference, are detected in the unhealthy group. Innocent murmurs have not been considered in this paper. Each auscultatory recording is classified into one of the following classes: healthy; click syndrome; and other heart dysfunctions. The results of the proposed method provided high recall and precision values for each of the three classes. Since the presence of additive noise may affect the classification, we also analyzed the possibility of classifying signals in such circumstances. The method was tested, verified and showed high accuracy.

New approach on analysis of pathologic cardiac murmurs based on WPD energy distribution

In this paper, an approach on analysis of the pathologic cardiac murmurs for congenital heart defects was proposed based on the wavelet packet (WP) technique. Considering the difference of the energy intensity distributions for the innocent and pathologic murmurs in frequency domain, the WP decomposition was introduced and the WP energies at each frequency band were calculated and compared. Based on the analysis of a large amount of clinic heart sound data, the murmurs energy distributions were divided into five frequency bands, and the relative evaluation indexes for cardiac murmurs (ICM) were proposed for analysis of the pathologic murmurs. Finally, the threshold values between the innocent and pathologic cardiac murmurs were determined based on the statistical results of the normal heart sounds. The analysis results validate the proposed evaluation indexes and the corresponding thresholds.

Current trends and perspectives for automated screening of cardiac murmurs

Although in high income countries rheumatic heart disease is now rare, it remains a major burden in low and middle income countries. In these world areas, physicians and expert sonographers are rare, and screening campaigns are usually performed by nomadic caregivers who can only recognise patients in an advanced phase of heart failure with high economic and social costs. Therefore, great interest exists regarding the possibility of developing a simple, low-cost procedure for screening valvular heart disease. With the development of computer science, the cardiac sound signal can be analysed in an automatic way. More precisely, a panel of features characterising the acoustic signal are extracted and sent to a decision-making software able to provide the final diagnosis. Although no system is currently available in the market, the rapid evolution of these technologies recently led to the activation of clinical trials. The aim of this note is to review the state of advancement of this technology (trends in feature selection and automatic diagnostic strategies), data available regarding performance of the technology in the clinical setting and finally what obstacles still need to be overcome before automated systems can be clinically/commercially viable.

Evaluation of cardiac auscultation skills in pediatric residents

Auscultation skills are in decline, but few studies have shown which specific aspects are most difficult for trainees. We evaluated individual aspects of cardiac auscultation among pediatric residents using recorded heart sounds to determine which elements pose the most difficulty.

METHODS

Auscultation proficiency was assessed among 34 trainees following a pediatric cardiology rotation using an open-set format evaluation module, similar to the actual clinical auscultation description process.

RESULTS

Diagnostic accuracy for distinguishing normal from abnormal cases was 73%. Findings most commonly correctly identified included pathological systolic and diastolic murmurs and widely split second heart sounds. Those least likely to be identified included continuous murmurs and clicks. Accuracy was low for identifying specific diagnoses.

CONCLUSIONS

Given time constraints for clinical skills teaching, this suggests that focusing on distinguishing normal from abnormal heart sounds and murmurs instead of making specific diagnoses may be a more realistic goal for pediatric resident auscultation training.

Neural network approaches to grade adult depression

Depression is a common but worrying psychological disorder that adversely affects one's quality of life. It is more ominous to note that its incidence is increasing. On the other hand, screening and grading of depression is still a manual and time consuming process that might be biased. In addition, grades of depression are often determined in continuous ranges, e.g., 'mild to moderate' and 'moderate to severe' instead of making them more discrete as 'mild', 'moderate', and 'severe'. Grading as a continuous range is confusing to the doctors and thus affecting the management plan at large. Given this practical issue, the present paper attempts to differentiate depression grades more accurately using two neural net learning approaches-'supervised', i.e., classification with Back propagation neural network (BPNN) and Adaptive Network-based Fuzzy Inference System (ANFIS) classifiers, and 'unsupervised', i.e., 'clustering' technique with Self-organizing map (SOM), built in MATLAB 7. The reason for using the supervised and unsupervised learning approaches is that, supervised learning depends exclusively on domain knowledge. Supervision may induce biasness and subjectivities related to the decision-making. Finally, the performance of BPNN and ANFIS are compared and discussed. It was observed that ANFIS, being a hybrid system, performed much better compared to the BPNN classifier. On the other hand, SOM-based clustering technique is also found useful in constructing three distinct clusters. It also assists visualizing the multidimensional data clusters into 2-D.

Autonomous auscultation of the human heart employing a precordial electro-phonocardiogram and ensemble empirical mode decomposition

The research presented in this paper serves to provide a tool to autonomously screen for cardiovascular disease in the rural areas of Africa. With this tool, cardiovascular disease can potentially be detected in its initial stages, which is essential for effective treatment. The autonomous auscultation system proposed here utilizes recorded heart sounds and electrocardiogram signals to automatically distinguish between normal and abnormal heart conditions. Patients that are identified as abnormal by the system can then be referred to a specialist consultant, which will save a lot of unnecessary referrals. In this study, heart sound and electrocardiogram signals were recorded with the prototype precordial electro-phonocardiogram device, as part of a clinical study to screen patients for cardiovascular disease. These volunteers consisted of 28 patients with a diagnosed cardiovascular disease and, for control purposes, 34 persons diagnosed with healthy hearts. The proposed system employs wavelets to first denoise the recorded signals, which is then followed by segmentation of heart sounds. Frequency spectrum information was extracted as diagnostic features from the heart sounds by means of ensemble empirical mode decomposition and auto regressive modelling. The respective features were then classified with an ensemble artificial neural network. The performance of the autonomous auscultation system used in concert with the precordial electro-phonocardiogram prototype showed a sensitivity of 82% and a specificity of 88%. These results demonstrate the potential benefit of the precordial electro-phonocardiogram device and the developed autonomous auscultation software as a screening tool in a rural healthcare environment where large numbers of patients are often cared for by a small number of inexperienced medical personnel.

Cardiac sound murmurs classification with autoregressive spectral analysis and multi-support vector machine technique

In this paper, a novel cardiac sound spectral analysis method using the normalized autoregressive power spectral density (NAR-PSD) curve with the support vector machine (SVM) technique is proposed for classifying the cardiac sound murmurs. The 489 cardiac sound signals with 196 normal and 293 abnormal sound cases acquired from six healthy volunteers and 34 patients were tested. Normal sound signals were recorded by our self-produced wireless electric stethoscope system where the subjects are selected who have no the history of other heart complications. Abnormal sound signals were grouped into six heart valvular disorders such as the atrial fibrillation, aortic insufficiency, aortic stenosis, mitral regurgitation, mitral stenosis and split sounds. These abnormal subjects were also not included other coexistent heart valvular disorder. Considering the morphological characteristics of the power spectral density of the heart sounds in frequency domain, we propose two important diagnostic features Fmax and Fwidth, which describe the maximum peak of NAR-PSD curve and the frequency width between the crossed points of NAR-PSD curve on a selected threshold value (THV), respectively. Furthermore, a two-dimensional representation on (Fmax, Fwidth) is introduced. The proposed cardiac sound spectral envelope curve method is validated by some case studies. Then, the SVM technique is employed as a classification tool to identify the cardiac sounds by the extracted diagnostic features. To detect abnormality of heart sound and to discriminate the heart murmurs, the multi-SVM classifiers composed of six SVM modules are considered and designed. A data set was used to validate the classification performances of each multi-SVM module. As a result, the accuracies of six SVM modules used for detection of abnormality and classification of six heart disorders showed 71-98.9% for THVs=10-90% and 81.2-99.6% for THVs=10-50% with respect to each of SVM modules. With the proposed cardiac sound spectral analysis method, the high classification performances were achieved by 99.9% specificity and 99.5% sensitivity in classifying normal and abnormal sounds (heart disorders). Consequently, the proposed method showed relatively very high classification efficiency if the SVM module is designed with considering THV values. And the proposed cardiac sound murmurs classification method with autoregressive spectral analysis and multi-SVM classifiers is validated for the classification of heart valvular disorders.

Support Vectors Machine-based identification of heart valve diseases using heart sounds

Taking into account that heart auscultation remains the dominant method for heart examination in the small health centers of the rural areas and generally in primary healthcare set-ups, the enhancement of this technique would aid significantly in the diagnosis of heart diseases. In this context, the present paper initially surveys the research that has been conducted concerning the exploitation of heart sound signals for automated and semi-automated detection of pathological heart conditions. Then it proposes an automated diagnosis system for the identification of heart valve diseases based on the Support Vector Machines (SVM) classification of heart sounds. This system performs a highly difficult diagnostic task (even for experienced physicians), much more difficult than the basic diagnosis of the existence or not of a heart valve disease (i.e. the classification of a heart sound as 'healthy' or 'having a heart valve disease'): it identifies the particular heart valve disease. The system was applied in a representative global dataset of 198 heart sound signals, which come both from healthy medical cases and from cases suffering from the four most usual heart valve diseases: aortic stenosis (AS), aortic regurgitation (AR), mitral stenosis (MS) and mitral regurgitation (MR). Initially the heart sounds were successfully categorized using a SVM classifier as normal or disease-related and then the corresponding murmurs in the unhealthy cases were classified as systolic or diastolic. For the heart sounds diagnosed as having systolic murmur we used a SVM classifier for performing a more detailed classification of them as having aortic stenosis or mitral regurgitation. Similarly for the heart sounds diagnosed as having diastolic murmur we used a SVM classifier for classifying them as having aortic regurgitation or mitral stenosis. Alternative classifiers have been applied to the same data for comparison (i.e. back-propagation neural networks, k-nearest-neighbour and naïve Bayes classifiers), however their performance for the same diagnostic problems was lower than the SVM classifiers proposed in this work.

Machine learning method for knowledge discovery experimented with otoneurological data

We have been interested in developing an otoneurological decision support system that supports diagnostics of vertigo diseases. In this study, we concentrate on testing its inference mechanism and knowledge discovery method. Knowledge is presented as patterns of classes. Each pattern includes attributes with weight and fitness values concerning the class. With the knowledge discovery method it is possible to form fitness values from data. Knowledge formation is based on frequency distributions of attributes. Knowledge formed by the knowledge discovery method is tested with two vertigo data sets and compared to experts' knowledge. The experts' and machine learnt knowledge are also combined in various ways in order to examine effects of weights on classification accuracy. The classification accuracy of knowledge discovery method is compared to 1- and 5-nearest neighbour method and Naive-Bayes classifier. The results showed that knowledge bases combining machine learnt knowledge with the experts' knowledge yielded the best classification accuracies. Further, attribute weighting had an important effect on the classification capability of the system. When considering different diseases in the used data sets, the performance of the knowledge discovery method and the inference method is comparable to other methods employed in this study.

Computer-aided auscultation of the heart: from anatomy and physiology to diagnostic decision support

There is a clear and present need for computer-aided auscultation of the heart which arises from the highly informative nature of heart sounds, the inherent difficulty of auscultation and increasing pressure in healthcare for rapid, accurate, objective, documented and cost-effective patient evaluation and diagnostic decision making. There are advanced signal processing technologies that hold promise for developing computer-aided auscultation solutions that are intuitive, efficient, informative and accurate. Computer-aided auscultation offers an objective, quantitative and cost-effective tool for acquiring and analyzing heart sounds, providing archival records that support the patient evaluation and referral decision as well as serial comparisons for patient monitoring. There is the further promise of new quantitative acoustic measures and auscultatory findings that have more precise correlation with underlying physiological parameters. These solutions are being developed with the benefits of a rich literature of clinical studies in phonocardiography, the added insights derived from echocardiography, and advances in signal processing technology.

Time-frequency and complexity analyses for differentiation of physiologic murmurs from heart murmurs caused by aortic stenosis in Boxers

OBJECTIVE

To investigate whether time-frequency and complexity analyses of heart murmurs can be used to differentiate physiologic murmurs from murmurs caused by aortic stenosis (AS) in Boxers.

ANIMALS

27 Boxers with murmurs.

PROCEDURES

Dogs were evaluated via auscultation and echocardiography. Analyses of time-frequency properties (TFPs; ie, maximal murmur frequency and duration of murmur frequency > 200 Hz) and correlation dimension (T(2)) of murmurs were performed on phonocardiographic sound data. Time-frequency property and T(2) analyses of low-intensity murmurs in 16 dogs without AS were performed at 7 weeks and 12 months of age. Additionally, TFP and T(2) analyses were performed on data obtained from 11 adult AS-affected dogs with murmurs.

RESULTS

In dogs with low-intensity murmurs, TFP or T(2) values at 7 weeks and 12 months did not differ significantly. For differentiation of physiologic murmurs from murmurs caused by mild AS, duration of murmur frequency > 200 Hz was useful and the combination assessment of duration of frequency > 200 Hz and T(2) of the murmur had a sensitivity of 94% and a specificity of 82%. Maximal murmur frequency did not differentiate dogs with AS from those without AS.

CONCLUSIONS AND CLINICAL RELEVANCE

Results suggested that assessment of the duration of murmur frequency > 200 Hz can be used to distinguish physiologic heart murmurs from murmurs caused by mild AS in Boxers. Combination of this analysis with T(2) analysis may be a useful complementary method for diagnostic assessment of cardiovascular function in dogs.

Phono-spectrographic analysis of heart murmur in children

BACKGROUND

More than 90% of heart murmurs in children are innocent. Frequently the skills of the first examiner are not adequate to differentiate between innocent and pathological murmurs. Our goal was to evaluate the value of a simple and low-cost phonocardiographic recording and analysis system in determining the characteristic features of heart murmurs in children and in distinguishing innocent systolic murmurs from pathological.

METHODS

The system consisting of an electronic stethoscope and a multimedia laptop computer was used for the recording, monitoring and analysis of auscultation findings. The recorded sounds were examined graphically and numerically using combined phono-spectrograms. The data consisted of heart sound recordings from 807 pediatric patients, including 88 normal cases without any murmur, 447 innocent murmurs and 272 pathological murmurs. The phono-spectrographic features of heart murmurs were examined visually and numerically. From this database, 50 innocent vibratory murmurs, 25 innocent ejection murmurs and 50 easily confusable, mildly pathological systolic murmurs were selected to test whether quantitative phono-spectrographic analysis could be used as an accurate screening tool for systolic heart murmurs in children.

RESULTS

The phono-spectrograms of the most common innocent and pathological murmurs were presented as examples of the whole data set. Typically, innocent murmurs had lower frequencies (below 200 Hz) and a frequency spectrum with a more harmonic structure than pathological cases. Quantitative analysis revealed no significant differences in the duration of S1 and S2 or loudness of systolic murmurs between the pathological and physiological systolic murmurs. However, the pathological murmurs included both lower and higher frequencies than the physiological ones (p < 0.001 for both low and high frequency limits). If the systolic murmur contained intensive frequency components of over 200 Hz, or its length accounted for over 80 % of the whole systolic duration, it was considered pathological. Using these criteria, 90 % specificity and 91 % sensitivity in screening were achieved.

CONCLUSION

Phono-spectrographic analysis improves the accuracy of primary heart murmur evaluation and educates inexperienced listener. Using simple quantitative criterias a level of pediatric cardiologist is easily achieved in screening heart murmurs in children.

Using voice-recognition technology to eliminate cardiac cycle segmentation in automated heart sound diagnosis

Advanced digital signal processing has the potential to revolutionize the stethoscope through the use of mathematical algorithms to interpret heart sound acoustic information. In this study, a novel classification algorithm that does not require cardiac cycle segmentation was used for identifying differences between normal and diseased heart sounds. The heart sound signals were not separated into systole and diastole. A recordable electronic stethoscope was used to record the heart sounds of 163 echocardiogram patients. Mel-cepstrum and Principal Components Analysis were applied to the 60 recorded heart sounds and decision spaces were developed. The algorithm was tested using 100 novel patients. The specificity of the algorithm is 72.4% and the sensitivity is 63.4%.