Heart Sound Classification and Recognition Based on Eemd and Correlation Dimension

Abstract

Analysis of heart sound is of great importance to the diagnosis of heart diseases. Most of the feature extraction methods about heart sound have focused on linear time-variant or time-invariant models. While heart sound is a kind of highly nonstationary and nonlinear vibration signal, traditional methods cannot fully reveal its essential properties. In this paper, a novel feature extraction approach is proposed for heart sound classification and recognition. The ensemble empirical mode decomposition (EEMD) method is used to decompose the heart sound into a finite number of intrinsic mode functions (IMFs), and the correlation dimensions of the main IMF components (IMF1~IMF4) are calculated as feature set. Then the classical Binary Tree Support Vector Machine (BT-SVM) classifier is employed to classify the heart sounds which include the normal heart sounds (NHSs) and three kinds of abnormal signals namely mitral stenosis (MT), ventricular septal defect (VSD) and aortic stenosis (AS). Finally, the performance of the new feature set is compared with the correlation dimensions of original signals and the main IMF components obtained by the EMD method. The results showed that, for NHSs, the feature set proposed in this paper performed the best with recognition rate of 98.67%. For the abnormal signals, the best recognition rate of 91.67% was obtained. Therefore, the proposed feature set is more superior to two comparative feature sets, which has potential application in the diagnosis of cardiovascular diseases.

The electronic stethoscope

“…of cases 80 6 9 9 12 Sensitivity – 67% 67% 67% 83% Specificity 99% – – – – [ 86 ] ANN (multi-class classification) Analyzed sound/disorder Y Y Y Y Y No. of cases 4 4 4 4 4 Sensitivity – 100% 98% 100% 100% Specificity 100% – – – – [ 87 ] SVM (binary hierarchical classification) Analyzed sound/disorder Y – Y – Y No. of cases 225 – 60 – 60 Sensitivity – – 85% – 95% …”

“… Statistical domain: skewness, kurtosis, chaos [ 84 ], entropy [ 90 , 105 ], heart rate [ 96 , 111 ], etc. Time–frequency domain: mel frequency cepstral coefficients (MFCC) [ 85 , 100 , 101 , 115 – 117 ], DWT [ 67 , 86 , 92 , 93 , 97 , 102 , 106 , 118 – 128 ], STFT [ 128 , 129 ], EMD [ 87 ], etc. …”

“…The accuracy provided in [ 86 ] is 99% in classifying normal HS, MR, MS, pulmonary stenosis (PS), AR and summation gallop (SG) using ANN. The recognition rate reported in [ 87 ] was 95.56% using SVM in classification of normal HS, MS, AS and ventricular septal defect (VSD). The ANN has the ability to adapt well with complex non-linear data (such as HS) and classify it accurately and effectively.…”

“…The objective of multiple independent binary classifications (Figure 11 a) is to categorize each of the heart disorder against NHS or a different heart disorder [ 81 , 83 , 84 , 88 ]. The more commonly adopted approaches are multi-class classification (Figure 11 b) [ 82 , 85 , 86 ] and binary hierarchical classification (Figure 11 c) [ 87 , 89 ], where the former classifies the HS instances into one of more than two classes; and the latter distinguishes one class from the remaining classes at each hierarchy level.

Figure 11 a Multiple binary independent classifications, b Multi-class classification, and c Binary hierarchical classification.

…”

Lung and Heart Sounds Analysis: State-of-the-Art and Future Trends

“…58,70,93,99,106,112 However, lung sounds recorded during pneumonia, 54,65,76,79,103 pulmonary fibrosis, 56,70,76 chronic bronchitis, 56 idiopathic pulmonary fibrosis, 63 congestive heart failure, 76,79 parenchymal pathology, 75,90 and interstitial lung disease 79,87,104 appear less frequently in the literature. Regarding heart sound analysis, it is observed that murmurs 26,121,126,131,132,[135][136][137] and heart valve problems such as regurgitation 25,135 and stenosis 130,140 are the most common heart pathologies studied.…”