Characterization of [Formula Omitted] and [Formula Omitted] Heart Sounds Using Stacked Autoencoder and Convolutional Neural Network

• Published in: IEEE transactions on instrumentation and measurement Vol. 68; no. 9; p. 3211
• Main Authors: Mishra, Madhusudhan, Menon, Hrishikesh, Mukherjee, Anirban
• Format: Journal Article
• Language: English
• Published: New York The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 01.01.2019
• Subjects: Abnormalities; Acoustics; Artificial neural networks; Classification; Discriminant analysis; Electrocardiography; Feature recognition; Heart; Neural networks; Support vector machines
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2018.2872387

This paper proposes a new technique for the identification of fundamental heart sounds (HSs), namely, [Formula Omitted] and [Formula Omitted] 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 [Formula Omitted] and [Formula Omitted] or electrocardiogram signal as reference. The motive of this paper is to demonstrate that the reliable [Formula Omitted] and [Formula Omitted] 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 [Formula Omitted] and [Formula Omitted] 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 [Formula Omitted] and [Formula Omitted] 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.