A new approach for the detection of abnormal heart sound signals using TQWT, VMD and neural networks
Wei Zeng,Jian Yuan,Chengzhi Yuan,Qinghui Wang,Fenglin Liu,Ying Wang
DOI: https://doi.org/10.1007/s10462-020-09875-w
IF: 9.588
2020-07-24
Artificial Intelligence Review
Abstract:Phonocardiogram (PCG) plays an important role in evaluating many cardiac abnormalities, such as the valvular heart disease, congestive heart failure and anatomical defects of the heart. However, effective cardiac auscultation requires trained physicians whose work is tough, laborious and subjective. The objective of this study is to develop an automatic classification method for anomaly (normal vs. abnormal) detection of PCG recordings without any segmentation of heart sound signals. Hybrid signal processing and artificial intelligence tools, including tunable Q-factor wavelet transform (TQWT), variational mode decomposition (VMD), phase space reconstruction (PSR) and neural networks, are utilized to extract representative features in order to model, identify and detect abnormal patterns in the dynamics of PCG system caused by heart disease. First, heart sound signal is decomposed into a set of frequency subbands with a number of decomposition levels by using the TQWT method. Second, VMD is employed to decompose the subband of the heart sound signal into different intrinsic modes, in which the first four intrinsic modes contain the majority of the heart sound signal's energy and are considered to be the predominant intrinsic modes. They are selected to construct the reference variable for analysis. Third, phase space of the reference variable is reconstructed, in which the properties associated with the nonlinear PCG system dynamics are preserved. Three-dimensional PSR together with Euclidean distance has been utilized to derive features, which demonstrate significant difference in PCG system dynamics between normal and abnormal heart sound signals. Finally, PhysioNet/CinC Challenge heart sound database is used for evaluation and the synthetic minority over-sampling technique method is applied to balance the datasets. By using the 10-fold cross-validation style, experimental results demonstrate that the proposed features with dynamical neural networks based classifier yield classification performance with sensitivity, specificity, overall score and accuracy values of 97.73<span class="mathjax-tex">\(\%\)</span>, 98.05<span class="mathjax-tex">\(\%\)</span>, 97.89<span class="mathjax-tex">\(\%\)</span>, and 97.89<span class="mathjax-tex">\(\%\)</span>, respectively. The results verify the effectiveness of the proposed method which can serve as a potential candidate for the automatic anomaly detection in the clinical application.
computer science, artificial intelligence