Characterization of $S_1$ and $S_2$ Heart Sounds Using Stacked Autoencoder and Convolutional Neural Network
- 15 October 2018
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 68 (9) , 3211-3220
Abstract
This paper proposes a new technique for the identification of fundamental heart sounds (HSs), namely, S₁ and S₂ from the cardiac cycle, the first and foremost step in the automated HSs analysis for the detection of pathological events, without incorporating time-interval informations between S₁ and S₂ or electrocardiogram signal as reference. The motive of this paper is to demonstrate that the reliable S₁ and S₂ classification performances based on the combinatory feature (CF) derived from higher order moments and cepstral-based domain can still be achieved, under the circumstances where the timing interval information might not be easily understood due to cardiac abnormalities. Using deep neural networks approach, a stacked autoencoder (SAE) based on the CF is proposed for the classification of fundamental HSs. Experiments are conducted on both publicly available and recorded HSs signals for the validation of the proposed method. The SAE using the proposed CF achieves better classification results in recognizing S₁ and S₂ in comparison to well-known classifiers such as deep belief neural network, support vector machine, Naive Bayes, linear discriminant analysis, and boosting ensemble. The proposed method shows higher classification rate in terms of accuracy, sensitivity, and specificity by considering CF, which uses Mel-frequency cepstral coefficients and its derivative features. A second approach for addressing the problem of S₁ and S₂ identifications is carried out by employing 1-D convolutional neural network that uses the signals directly to learn the relevant features by its own for the recognition.Keywords
This publication has 28 references indexed in Scilit:
- Bearing Health Monitoring Based on Hilbert–Huang Transform, Support Vector Machine, and RegressionIEEE Transactions on Instrumentation and Measurement, 2014
- Equalization of Crackle Sounds to Compensate Thorax AttenuationIEEE Transactions on Instrumentation and Measurement, 2014
- Efficient Heart Sound Segmentation and Extraction Using Ensemble Empirical Mode Decomposition and Kurtosis FeaturesIEEE Journal of Biomedical and Health Informatics, 2013
- Conditional Mutual Information-Based Feature Selection Analyzing for Synergy and RedundancyETRI Journal, 2011
- Segmentation of heart sound recordings by a duration-dependent hidden Markov modelPhysiological Measurement, 2010
- Conditional Infomax Learning: An Integrated Framework for Feature Extraction and Fusion2006
- On the Use of Variable Complementarity for Feature Selection in Cancer Classification2006
- Interpolated smoothed pseudo Wigner-Ville distribution for accurate spectrum analysisIEEE Transactions on Instrumentation and Measurement, 1996
- Using mutual information for selecting features in supervised neural net learningIEEE Transactions on Neural Networks, 1994
- Application of Wigner-Ville distribution to measurements on transient signalsIEEE Transactions on Instrumentation and Measurement, 1994