In this study, we present an improved computer-aided-diagnosis (CAD) system to distinguish between normal heart sound and one affected with murmur. The proposed system is based on nonlinear characteristics of the original heart sound high frequency oscillations. Specifically, the original signal is decomposed by discrete wavelet transform (DWT) and analyzed by complexity measures in a straightforward manner to describe its overall characteristics, rather than to describe characteristics of the sounds related to the turbulent flow during the different phases of a heartbeat. The complexity measures include Hurst exponent, Lempel-Ziv information, and Shannon entropy. They are computed from the high frequency oscillations which are obtained by wavelet transform. These nonlinear characteristics are employed to train nonlinear support vector machines (SVM) classifier. The latter was tuned by Bayesian optimization. Tested on a new large dataset, the proposed CAD system outperforms existing models that were validated on the same database. The proposed approach is fast, effective, and promising in clinical milieu.
Complexity measures of high oscillations in phonocardiogram as biomarkers to distinguish between normal heart sound and pathological murmur
Bekiros S.
2022-01-01
Abstract
In this study, we present an improved computer-aided-diagnosis (CAD) system to distinguish between normal heart sound and one affected with murmur. The proposed system is based on nonlinear characteristics of the original heart sound high frequency oscillations. Specifically, the original signal is decomposed by discrete wavelet transform (DWT) and analyzed by complexity measures in a straightforward manner to describe its overall characteristics, rather than to describe characteristics of the sounds related to the turbulent flow during the different phases of a heartbeat. The complexity measures include Hurst exponent, Lempel-Ziv information, and Shannon entropy. They are computed from the high frequency oscillations which are obtained by wavelet transform. These nonlinear characteristics are employed to train nonlinear support vector machines (SVM) classifier. The latter was tuned by Bayesian optimization. Tested on a new large dataset, the proposed CAD system outperforms existing models that were validated on the same database. The proposed approach is fast, effective, and promising in clinical milieu.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
