Heart sounds play an important role in the diagnosis of cardiac conditions. Due to the low signal-to-noise ratio (SNR), it is problematic and time-consuming for experts to discriminate different kinds of heart sounds. Thus, objective classification of heart sounds is essential. In this study, we combined a conventional feature engineering method with deep learning algorithms to automatically classify normal and abnormal heart sounds. First, 497 features were extracted from eight domains. Then, we fed these features into the designed convolutional neural network (CNN), in which the fully connected layers that are usually used before the classification layer were replaced with a global average pooling layer to obtain global information about the feature maps and avoid overfitting. Considering the class imbalance, the class weights were set in the loss function during the training process to improve the classification algorithm’s performance. Stratified five-fold cross-validation was used to evaluate the performance of the proposed method. The mean accuracy, sensitivity, specificity and Matthews correlation coefficient observed on the PhysioNet/CinC Challenge 2016 dataset were 86.8%, 87%, 86.6% and 72.1% respectively. The proposed algorithm’s performance achieves an appropriate trade-off between sensitivity and specificity.

This paper proposes a method using multidomain features and support vector machine (SVM) for classifying normal and abnormal heart sound recordings. The database was provided by the PhysioNet/CinC Challenge 2016. A total of 515 features are extracted from nine feature domains, i.e., time interval, frequency spectrum of states, state amplitude, energy, frequency spectrum of records, cepstrum, cyclostationarity, high-order statistics, and entropy. Correlation analysis is conducted to quantify the feature discrimination abilities, and the results show that “frequency spectrum of state”, “energy”, and “entropy” are top domains to contribute effective features. A SVM with radial basis kernel function was trained for signal quality estimation and classification. The SVM classifier is independently trained and tested by many groups of top features. It shows the average of sensitivity, specificity, and overall score are high up to 0.88, 0.87, and 0.88, respectively, when top 400 features are used. This score is competitive to the best previous scores. The classifier has very good performance with even small number of top features for training and it has stable output regardless of randomly selected features for training. These simulations demonstrate that the proposed features and SVM classifier are jointly powerful for classifying heart sound recordings.

Noise is generally unavoidable during recordings of heart sound signal. Therefore, noise reduction is one of the important preprocesses in the analysis of heart sound signal. This was achieved in joint cycle frequency-time-frequency domains in this study. Heart sound signal was decomposed into components (called atoms) characterized by time delay, frequency, amplitude, time width, and phase. It was discovered that atoms of heart sound signal congregate in the joint domains. On the other hand, atoms of noise were dispersed. The atoms of heart sound signal could, therefore, be separated from the atoms of noise based on fuzzy detection. In a practical experiment, heart sound signal was successfully separated from lung sounds and disturbances due to chest motion. Computer simulations for various clinical heart sound signals were also used to evaluate the performance of the proposed noise reduction. It was shown that heart sound signal can be reconstructed from simulated complex noise (perhaps non-Gaussian, nonstationary, and colored). The proposed noise reduction can recover variations in the both waveform and time delay of heart sound signal during the reconstruction. Correlation coefficient and normalized residue were used to indicate the closeness of the reconstructed and noise-free heart sound signal. Correlation coefficient may exceed 0.90 and normalized residue may be around 0.10 in 0-dB noise environment, even if the phonocardiogram signal covers only ten cardiac cycles.

Background and aims: Auscultation is a cheap and fundamental technique for detecting cardiovascular disease effectively. Doctors’ abilities in auscultation are varied. Sometimes, there may be cases of misdiagnosis, even when auscultation is performed by an experienced doctor. Hence, it is necessary to propose accurate computational tools to assist auscultation, especially in developing countries. Artificial intelligence technology can be an efficient diagnostic tool for detecting cardiovascular disease. This work proposed an automatic multiple classification method for cardiovascular disease detection by heart sound signals. Methods and results: In this work, a 1D heart sound signal is translated into its corresponding 3D spectrogram using continuous wavelet transform (CWT). In total, six classes of heart sound data are used in this experiment. We combine an open database (including five classes of heart sound data: aortic stenosis, mitral regurgitation, mitral stenosis, mitral valve prolapse and normal) with one class (pulmonary hypertension) of heart sound data collected by ourselves to perform the experiment. To make the method robust in a noisy environment, the background deformation technique is used before training. Then, 10 transfer learning networks (GoogleNet, SqueezeNet, DarkNet19, MobileNetv2, Inception-ResNetv2, DenseNet201, Inceptionv3, ResNet101, NasNet-Large, and Xception) are used for comparison. Furthermore, other models (LSTM and CNN) are also compared with our proposed algorithm. The experimental results show that four transfer learning networks (ResNet101, DenseNet201, DarkNet19 and GoogleNet) outperformed their peer models with an accuracy of 0.98 to detect the multiple heart diseases. The performances have been validated both in the original heart sound and the augmented heart sound using 10-fold cross validation. The results of these 10 folds are reported in this research. Conclusions: Our method obtained high classification accuracy even under a noisy background, which suggests that the proposed classification method could be used in auxiliary diagnosis for cardiovascular diseases.

Automated heart sound signal quality assessment is a necessary step for reliable analysis of heart sound signal. An unavoidable processing step for this objective is the heart sound segmentation, which is still a challenging task from a technical viewpoint. In this study, ten features are defined to evaluate the quality of heart sound signal without segmentation. The ten features come from kurtosis, energy ratio, frequency-smoothed envelope, and degree of sound periodicity, where five of them are novel in signal quality assessment. We have collected a total of 7893 recordings from open public heart sound databases and performed manual annotation for each recording as gold standard quality label. The signal quality is classified based on two schemes: binary classification (“unacceptable” and “acceptable”) and triple classification (“unacceptable”, “good,” and “excellent”). Sequential forward feature selection shows that the feature “the degree of periodicity” gives an accuracy rate of 73.1% in binary SVM classification. The top five features dominate the classification performance and give an accuracy rate of 92%. The binary classifier has excellent generalization ability since the accuracy rate reaches to ( ) % even if 10% of the data is used to train the classifier. The rate increases to ( ) % in 10-fold validation. The triple classification has an accuracy rate of ( ) % in 10-fold validation. The results verify the effectiveness of the signal quality assessment, which could serve as a potential candidate as a preprocessing in future automatic heart sound analysis in clinical application.