NeoSSNet: Real-Time Neonatal Chest Sound Separation Using Deep Learning.

Goal: Auscultation for neonates is a simple and non-invasive method of diagnosing cardiovascular and respiratory disease. However, obtaining high-quality chest sounds containing only heart or lung sounds is non-trivial. Hence, this study introduces a new deep-learning model named NeoSSNet and evaluates its performance in neonatal chest sound separation with previous methods. Methods: We propose a masked-based architecture similar to Conv-TasNet. The encoder and decoder consist of 1D convolution and 1D transposed convolution, while the mask generator consists of a convolution and transformer architecture. The input chest sounds were first encoded as a sequence of tokens using 1D convolution. The tokens were then passed to the mask generator to generate two masks, one for heart sounds and one for lung sounds. Each mask is then applied to the input token sequence. Lastly, the tokens are converted back to waveforms using 1D transposed convolution. Results: Our proposed model showed superior results compared to the previous methods based on objective distortion measures, ranging from a 2.01 dB improvement to a 5.06 dB improvement. The proposed model is also significantly faster than the previous methods, with at least a 17-time improvement. Conclusions: The proposed model could be a suitable preprocessing step for any health monitoring system where only the heart sound or lung sound is desired.

Feasibility of self-measurement telemonitoring using a handheld heart sound recorder in patients with heart failure - SELPH multicenter pilot study.

BACKGROUND: Multi-parametric assessment, including heart sounds in addition to conventional parameters, may enhance the efficacy of noninvasive telemonitoring for heart failure (HF). We sought to assess the feasibility of self-telemonitoring with multiple devices including a handheld heart sound recorder and its association with clinical events in patients with HF. METHODS: Ambulatory HF patients recorded their own heart sounds, mono­lead electrocardiograms, oxygen saturation, body weight, and vital signs using multiple devices every morning for six months. RESULTS: In the 77 patients enrolled (63 ±â€¯13 years old, 84 % male), daily measurements were feasible with a self-measurement rate of >70 % of days in 75 % of patients. Younger age and higher Minnesota Living with Heart Failure Questionnaire scores were independently associated with lower adherence (p = 0.002 and 0.027, respectively). A usability questionnaire showed that 87 % of patients felt self-telemonitoring was helpful, and 96 % could use the devices without routine cohabitant support. Six patients experienced ten HF events of re-hospitalization and/or unplanned hospital visits due to HF. In patients who experienced HF events, a significant increase in heart rate and diastolic blood pressure and a decrease in the time interval from Q wave onset to the second heart sound were observed 7 days before the events compared with those without HF events. CONCLUSIONS: Self-telemonitoring with multiple devices including a handheld heart sound recorder was feasible even in elderly patients with HF. This intervention may confer a sense of relief to patients and enable monitoring of physiological parameters that could be valuable in detecting the deterioration of HF.

Improved recovery of cardiac auscultation sounds using modified cosine transform and LSTM-based masking.

Obtaining accurate cardiac auscultation signals, including basic heart sounds (S1 and S2) and subtle signs of disease, is crucial for improving cardiac diagnoses and making the most of telehealth. This research paper introduces an innovative approach that utilizes a modified cosine transform (MCT) and a masking strategy based on long short-term memory (LSTM) to effectively distinguish heart sounds and murmurs from background noise and interfering sounds. The MCT is used to capture the repeated pattern of the heart sounds, while the LSTMs are trained to construct masking based on the repeated MCT spectrum. The proposed strategy's performance in maintaining the clinical relevance of heart sounds continues to demonstrate effectiveness, even in environments marked by increased noise and complex disruptions. The present work highlights the clinical significance and reliability of the suggested methodology through in-depth signal visualization and rigorous statistical performance evaluations. In comparative assessments, the proposed approach has demonstrated superior performance compared to recent algorithms, such as LU-Net and PC-DAE. Furthermore, the system's adaptability to various datasets enhances its reliability and practicality. The suggested method is a potential way to improve the accuracy of cardiovascular diagnostics in an era of rapid advancement in medical signal processing. The proposed approach showed an enhancement in the average signal-to-noise ratio (SNR) by 9.6 dB at an input SNR of - 6 dB and by 3.3 dB at an input SNR of 10 dB. The average signal distortion ratio (SDR) achieved across a variety of input SNR values was 8.56 dB.

Beat-to-beat alterations of acoustic intensity and frequency at the maximum power of heart sounds are associated with NT-proBNP levels.

Background: Auscultatory features of heart sounds (HS) in patients with heart failure (HF) have been studied intensively. Recent developments in digital and electrical devices for auscultation provided easy listening chances to recognize peculiar sounds related to diastolic HS such as S3 or S4. This study aimed to quantitatively assess HS by acoustic measures of intensity (dB) and audio frequency (Hz). Methods: Forty consecutive patients aged between 46 and 87 years (mean age, 74 years) with chronic cardiovascular disease (CVD) were enrolled in the present study after providing written informed consent during their visits to the Kitasato University Outpatient Clinic. HS were recorded at the fourth intercostal space along the left sternal border using a highly sensitive digital device. Two consecutive heartbeats were quantified on sound intensity (dB) and audio frequency (Hz) at the peak power of each spectrogram of S1-S4 using audio editing and recording application software. The participants were classified into three groups, namely, the absence of HF (n = 27), HF (n = 8), and high-risk HF (n = 5), based on the levels of NT-proBNP < 300, ≥300, and ≥900 pg/ml, respectively, and also the levels of ejection fraction (EF), such as preserved EF (n = 22), mildly reduced EF (n = 12), and reduced EF (n = 6). Results: The intensities of four components of HS (S1-S4) decreased linearly (p < 0.02-0.001) with levels of body mass index (BMI) (range, 16.2-33.0 kg/m2). Differences in S1 intensity (ΔS1) and its frequency (ΔfS1) between two consecutive beats were non-audible level and were larger in patients with HF than those in patients without HF (ΔS1, r = 0.356, p = 0.024; ΔfS1, r = 0.356, p = 0.024). The cutoff values of ΔS1 and ΔfS1 for discriminating the presence of high-risk HF were 4.0 dB and 5.0 Hz, respectively. Conclusions: Despite significant attenuations of all four components of HS by BMI, beat-to-beat alterations of both intensity and frequency of S1 were associated with the severity of HF. Acoustic quantification of HS enabled analyses of sounds below the audible level, suggesting that sound analysis might provide an early sign of HF.

Automated valvular heart disease detection using heart sound with a deep learning algorithm.

Background: Insufficient clinicians' auscultation ability delays the diagnosis and treatment of valvular heart disease (VHD); artificial intelligence provides a solution to compensate for the insufficiency in auscultation ability by distinguishing between heart murmurs and normal heart sounds. However, whether artificial intelligence can automatically diagnose VHD remains unknown. Our objective was to use deep learning to process and compare raw heart sound data to identify patients with VHD requiring intervention. Methods: Heart sounds from patients with VHD and healthy controls were collected using an electronic stethoscope. Echocardiographic findings were used as the gold standard for this study. According to the chronological order of enrollment, the early-enrolled samples were used to train the deep learning model, and the late-enrollment samples were used to validate the results. Results: The final study population comprised 499 patients (354 in the algorithm training group and 145 in the result validation group). The sensitivity, specificity, and accuracy of the deep-learning model for identifying various VHDs ranged from 71.4 to 100.0%, 83.5-100.0%, and 84.1-100.0%, respectively; the best diagnostic performance was observed for mitral stenosis, with a sensitivity of 100.0% (31.0-100.0%), a specificity of 100% (96.7-100.0%), and an accuracy of 100% (97.5-100.0%). Conclusions: Based on raw heart sound data, the deep learning model effectively identifies patients with various types of VHD who require intervention and assists in the screening, diagnosis, and follow-up of VHD.

Early diastolic heart sounds caused by the atrial kick.

An 81-year-old male with a history of coronary artery disease, hypertension, paroxysmal atrial fibrillation and chronic kidney disease presents with asymptomatic bradycardia. Examination was notable for an early diastolic heart sound. 12-lead electrocardiogram revealed sinus bradycardia with a markedly prolonged PR interval and second-degree atrioventricular block, type I Mobitz. We review the differential diagnosis of early diastolic heart sounds and present a case of Wenckebach associated with a variable early diastolic sound on physical exam.

A novel feature-level fusion scheme with multimodal attention CNN for heart sound classification.

BACKGROUND AND OBJECTIVE: Most of the existing machine learning-based heart sound classification methods achieve limited accuracy. Since they primarily depend on single domain feature information and tend to focus equally on each part of the signal rather than employing a selective attention mechanism. In addition, they fail to exploit convolutional neural network (CNN) - based features with an effective fusion strategy. METHODS: In order to overcome these limitations, a novel multimodal attention convolutional neural network (MACNN) with a feature-level fusion strategy, in which Mel-cepstral domain as well as general frequency domain features are incorporated to increase the diversity of the features, is proposed in this paper. In the proposed method, DilationAttenNet is first utilized to construct attention-based CNN feature extractors and then these feature extractors are jointly optimized in MACNN at the feature-level. The attention mechanism aims to suppress irrelevant information and focus on crucial diverse features extracted from the CNN. RESULTS: Extensive experiments are carried out to study the efficacy of the feature level fusion in comparison to that with early fusion. The results show that the proposed MACNN method significantly outperforms the state-of-the-art approaches in terms of accuracy and score for the two publicly available Github and Physionet datasets. CONCLUSION: The findings of our experiments demonstrated the high performance for heart sound classification based on the proposed MACNN, and hence have potential clinical usefulness in the identification of heart diseases. This technique can assist cardiologists and researchers in the design and development of heart sound classification methods.

[Diagnosis of pulmonary hypertension associated with congenital heart disease based on statistical features of the second heart sound].

Aiming at the problems of obscure clinical auscultation features of pulmonary hypertension associated with congenital heart disease and the complexity of existing machine-aided diagnostic algorithms, an algorithm based on the statistical characteristics of the high-frequency components of the second heart sound signal is proposed. Firstly, an endpoint detection adaptive segmentation method is employed to extract the second heart sounds. Subsequently, the high-frequency component of the heart sound is decomposed using the discrete wavelet transform. Statistical features including the Hurst exponent, Lempel-Ziv information and sample entropy are extracted from this component. Finally, the extracted features are utilized to train an extreme gradient boosting algorithm (XGBoost) classifier, which achieves an accuracy of 80.45% in triple classification. Notably, this method eliminates the need for a noise reduction algorithm, allows for swift feature extraction, and achieves effective multi-classification using only three features. It is promising for early screening of pulmonary hypertension associated with congenital heart disease.

[Research on bark-frequency spectral coefficients heart sound classification algorithm based on multiple window time-frequency reassignment].

The multi-window time-frequency reassignment helps to improve the time-frequency resolution of bark-frequency spectral coefficient (BFSC) analysis of heart sounds. For this purpose, a new heart sound classification algorithm combining feature extraction based on multi-window time-frequency reassignment BFSC with deep learning was proposed in this paper. Firstly, the randomly intercepted heart sound segments are preprocessed with amplitude normalization, the heart sounds were framed and time-frequency rearrangement based on short-time Fourier transforms were computed using multiple orthogonal windows. A smooth spectrum estimate is calculated by arithmetic averaging each of the obtained independent spectra. Finally, the BFSC of reassignment spectrum is extracted as a feature by the Bark filter bank. In this paper, convolutional network and recurrent neural network are used as classifiers for model comparison and performance evaluation of the extracted features. Eventually, the multi-window time-frequency rearrangement improved BFSC method extracts more discriminative features, with a binary classification accuracy of 0.936, a sensitivity of 0.946, and a specificity of 0.922. These results present that the algorithm proposed in this paper does not need to segment the heart sounds and randomly intercepts the heart sound segments, which greatly simplifies the computational process and is expected to be used for screening of congenital heart disease.

A Case of Sinus Venosus Atrial Septal Defect: Physical Examination as a Diagnostic Clue.

An atrial septal defect (ASD) may be detected later in life due to its asymptomatic status. We report a case of superior sinus venosus ASD, a rare type of ASD, in which bedside physical examination was useful for the diagnosis. A 72-year-old male was referred to cardiology during the treatment of a cerebral infarction. On examination, a right ventricular heave, a split-second heart sound with an increased pulmonary component, and a systolic ejection murmur in the pulmonary region were noted. Transthoracic echocardiography showed a systolic pulmonary artery pressure of 50 mmHg with right heart enlargement, but there was no shunt flow. Because an agitated saline contrast study was positive, transesophageal echocardiography was performed and demonstrated direct flow between the left atrium and superior vena cava. Our report highlights the importance of considering ASD, such as sinus venosus type, even in the absence of transthoracic echocardiographic findings suggestive of this condition, when patients present with a bedside physical examination consistent with ASD.

Deep learning of heart-sound signals for efficient prediction of obstructive coronary artery disease.

Background: Due to the limitations of current methods for detecting obstructive coronary artery disease (CAD), many individuals are mistakenly or unnecessarily referred for coronary angiography (CAG). Objectives: Our goal is to create a comprehensive database of heart sounds in CAD and develop accurate deep learning algorithms to efficiently detect obstructive CAD based on heart sound signals. This will enable effective screening before undergoing CAG. Methods: We included 320 subjects suspected of CAD who underwent CAG. We employed advanced filtering techniques and state-of-the-art deep learning models (VGG-16, 1D CNN, and ResNet18) to analyze the heart sound signals and identify obstructive CAD (defined as at least one ≥50 % stenosis). To assess the performance of our models, we prospectively recruited an additional 80 subjects for testing. Results: In the test set, VGG-16 exhibited the highest performance with an area under the ROC curve (AUC) of 0.834 (95 % CI, 0.736-0.930), while ResNet-18 and CNN-7 achieved AUCs of only 0.755 (95 % CI, 0.614-0.819) and 0.652 (95 % CI, 0.554-0.770) respectively. VGG-16 demonstrated a sensitivity of 80.4 % and specificity of 86.2 % in the test set. The combined diagnostic model of VGG and DF scores achieved an AUC of 0.915 (95 % CI: 0.855-0.974), and the AUC for VGG combined with PTP scores was 0.908 (95 % CI: 0.845-0.971). The sensitivity and specificity of VGG-16 exceeded 0.85 in patients with coronary artery occlusion and those with 3 vascular lesions. Conclusions: Our deep learning model, based on heart sounds, offers a non-invasive and efficient screening method for obstructive CAD. It is expected to significantly reduce the number of unnecessary referrals for downstream screening.

Changes in brain function and heart sound in acute sleep deprivation individuals.

AIMS: Sleep deprivation (SD) has become a health problem in modern society due to its adverse effects on different aspects. However, the relationship between sleep and cardiovascular system function remains unclear. Here we explored the changes occurring in the brain and the heart sounds after SD. METHODS: Ninety healthy adult men were recruited and subjected to 36 h of Sleep Deprivation (SD). They participated in a number of tests, including measurements of the heart sound, blood oxygen, and heart rate every 2 h. By using of principal component analysis to reduced the dimensionality of heart sound data. While the ALFF and ReHo indexes were measured via fMRI before and after SD. Correlation and regression analyses were used to reveal the relationship between fMRI and heart sound changes due to SD. RESULTS: In this study, there were no abnormal values in the heart rate and blood oxygen during 36 h of SD, whereas the intensity of heart sounds fluctuated significantly increased and decreased. The ALFF was increased in bilateral pericalcarine(Calcarine), left anterior cuneus, (Precuneus_L), right superior temporal gyrus(Temporal_Sup_R), left supplementary motor area (Supp_Motor_Area_L); However, it was reduced in the right medial superior frontal gyrus (Frontal_Sup_Medial_R), right dorsolateral superior frontal gyrus (Frontal_Sup_R) and left medial frontal gyrus (Frontal_Mid_L). The regression analysis uncovered that the intensity of the heart sound in the systole, s1, and s2 phase could be explained by Calcarine_L changes. CONCLUSION: Acute sleep deprivation affects cardiac-brain axis and the specific brain regions. Calcarine_L changes during sleep deprivation are involved in regulating heart contractions.

Hybrid Sneaky algorithm-based deep neural networks for Heart sound classification using phonocardiogram.

In the diagnosis of cardiac disorders Heart sound has a major role, and early detection is crucial to safeguard the patients. Computerized strategies of heart sound classification advocate intensive and more exact results in a quick and better manner. Using a hybrid optimization-controlled deep learning strategy this paper proposed an automatic heart sound classification module. The parameter tuning of the Deep Neural Network (DNN) classifier in a satisfactory manner is the importance of this research which depends on the Hybrid Sneaky optimization algorithm. The developed sneaky optimization algorithm inherits the traits of questing and societal search agents. Moreover, input data from the Phonocardiogram (PCG) database undergoes the process of feature extraction which extract the important features, like statistical, Heart Rate Variability (HRV), and to enhance the performance of this model, the features of Mel frequency Cepstral coefficients (MFCC) are assisted. The developed Sneaky optimization-based DNN classifier's performance is determined in respect of the metrics, namely precision, accuracy, specificity, and sensitivity, which are around 97%, 96.98%, 97%, and 96.9%, respectively.

AI diagnosis of heart sounds differentiated with super StethoScope.

In the aging global society, heart failure and valvular heart diseases, including aortic stenosis, are affecting millions of people and healthcare systems worldwide. Although the number of effective treatment options has increased in recent years, the lack of effective screening methods is provoking continued high mortality and rehospitalization rates. Appropriately, auscultation has been the primary option for screening such patients, however, challenges arise due to the variability in auscultation skills, the objectivity of the clinical method, and the presence of sounds inaudible to the human ear. To address challenges associated with the current approach towards auscultation, the hardware of Super StethoScope was developed. This paper is composed of (1) a background literature review of bioacoustic research regarding heart disease detection, (2) an introduction of our approach to heart sound research and development of Super StethoScope, (3) a discussion of the application of remote auscultation to telemedicine, and (4) results of a market needs survey on traditional and remote auscultation. Heart sounds and murmurs, if collected properly, have been shown to closely represent heart disease characteristics. Correspondingly, the main characteristics of Super StethoScope include: (1) simultaneous collection of electrocardiographic and heart sound for the detection of heart rate variability, (2) optimized signal-to-noise ratio in the audible frequency bands, and (3) acquisition of heart sounds including the inaudible frequency ranges. Due to the ability to visualize the data, the device is able to provide quantitative results without disturbance by sound quality alterations during remote auscultations. An online survey of 3648 doctors confirmed that auscultation is the common examination method used in today's clinical practice and revealed that artificial intelligence-based heart sound analysis systems are expected to be integrated into clinicians' practices. Super StethoScope would open new horizons for heart sound research and telemedicine.

[Detection method of early heart valve diseases based on heart sound features].

Heart valve disease (HVD) is one of the common cardiovascular diseases. Heart sound is an important physiological signal for diagnosing HVDs. This paper proposed a model based on combination of basic component features and envelope autocorrelation features to detect early HVDs. Initially, heart sound signals lasting 5 minutes were denoised by empirical mode decomposition (EMD) algorithm and segmented. Then the basic component features and envelope autocorrelation features of heart sound segments were extracted to construct heart sound feature set. Then the max-relevance and min-redundancy (MRMR) algorithm was utilized to select the optimal mixed feature subset. Finally, decision tree, support vector machine (SVM) and k-nearest neighbor (KNN) classifiers were trained to detect the early HVDs from the normal heart sounds and obtained the best accuracy of 99.9% in clinical database. Normal valve, abnormal semilunar valve and abnormal atrioventricular valve heart sounds were classified and the best accuracy was 99.8%. Moreover, normal valve, single-valve abnormal and multi-valve abnormal heart sounds were classified and the best accuracy was 98.2%. In public database, this method also obtained the good overall accuracy. The result demonstrated this proposed method had important value for the clinical diagnosis of early HVDs.

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.

Heart Sound Classification Network Based on Convolution and Transformer.

Electronic auscultation is vital for doctors to detect symptoms and signs of cardiovascular diseases (CVDs), significantly impacting human health. Although progress has been made in heart sound classification, most existing methods require precise segmentation and feature extraction of heart sound signals before classification. To address this, we introduce an innovative approach for heart sound classification. Our method, named Convolution and Transformer Encoder Neural Network (CTENN), simplifies preprocessing, automatically extracting features using a combination of a one-dimensional convolution (1D-Conv) module and a Transformer encoder. Experimental results showcase the superiority of our proposed method in both binary and multi-class tasks, achieving remarkable accuracies of 96.4%, 99.7%, and 95.7% across three distinct datasets compared with that of similar approaches. This advancement holds promise for enhancing CVD diagnosis and treatment.

Design and application of a flexible nano cardiac sound sensor based on P(VDF-TrFE)/KNN/GR composite piezoelectric film for heart disease diagnosis.

The paper proposes a flexible micro-nano composite piezoelectric thin film. This flexible piezoelectric film is fabricated through electrospinning process, utilizing a combination of 12 wt% poly(vinylidene fluoride-co-trifluoroethylene)(P(VDF-TrFE)), 8 wt% potassium sodium niobate (KNN) nanoparticles, and 0.5 wt% graphene (GR). Under cyclic loading, the composite film demonstrates a remarkable increase in open-circuit voltage and short-circuit current, achieving values of 36.1 V and 163.7 uA, respectively. These values are 5.8 times and 3.6 times higher than those observed in the pure P(VDF-TrFE) film. The integration of this piezoelectric film into a wearable flexible heartbeat sensor, coupled with the RepMLP classification model, facilitates heartbeat acquisition and real-time automated diagnosis. After training and validation on a dataset containing 2000 heartbeat samples, the system achieved an accuracy of approximately 99% in two classification of heart sound signals (normal and abnormal). This research substantially enhances the output performance of the piezoelectric film, offering a novel and valuable solution for the application of flexible piezoelectric films in physiological signal detection.

Deep Learning Algorithms to Detect Murmurs Associated With Structural Heart Disease.

Background The success of cardiac auscultation varies widely among medical professionals, which can lead to missed treatments for structural heart disease. Applying machine learning to cardiac auscultation could address this problem, but despite recent interest, few algorithms have been brought to clinical practice. We evaluated a novel suite of Food and Drug Administration-cleared algorithms trained via deep learning on >15 000 heart sound recordings. Methods and Results We validated the algorithms on a data set of 2375 recordings from 615 unique subjects. This data set was collected in real clinical environments using commercially available digital stethoscopes, annotated by board-certified cardiologists, and paired with echocardiograms as the gold standard. To model the algorithm in clinical practice, we compared its performance against 10 clinicians on a subset of the validation database. Our algorithm reliably detected structural murmurs with a sensitivity of 85.6% and specificity of 84.4%. When limiting the analysis to clearly audible murmurs in adults, performance improved to a sensitivity of 97.9% and specificity of 90.6%. The algorithm also reported timing within the cardiac cycle, differentiating between systolic and diastolic murmurs. Despite optimizing acoustics for the clinicians, the algorithm substantially outperformed the clinicians (average clinician accuracy, 77.9%; algorithm accuracy, 84.7%.) Conclusions The algorithms accurately identified murmurs associated with structural heart disease. Our results illustrate a marked contrast between the consistency of the algorithm and the substantial interobserver variability of clinicians. Our results suggest that adopting machine learning algorithms into clinical practice could improve the detection of structural heart disease to facilitate patient care.