Wavelet Packet Algorithm to Feature Extraction of Heart Sound

Abstract:

In order to extract pathological features of heart sound signal accurately, an algorithm for extracting the sub-band energy is developed based on the wavelet packet. The db6 wavelet is taken as the mother function, and the best wavelet packet basis of heart sound signal is picked out. Then, various heart sound signals are decomposed into four levels and the wavelet packet coefficients of the best basis are obtained. According to the equal-value relation between wavelet packet coefficients and signal energy in time domain, the normalized sub-band energy of the best basis is extracted as the feature vector. Based on BP network, seven identification models for seven kinds of heart sound were trained separately. Then, these models were tested by using 70 heart sounds, and the mean of identification accuracy is 72.9%.

• Periodical: Advanced Materials Research (Volumes 317-319)
• Pages: 1211-1214
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.317-319.1211
• Online since: August 2011
• Authors: Guo Hua Zhang, Shi Xuan Liu
• Keywords: Feature Extraction, Heart Sound, Identification Model, Wavelet Packet (WP)

References

[1] Zhidong Zhao, Zhijin Zhao, Song Zhang: Space Medicine & Medical Engineering. 17(2004)452- 456.

[2] A.P. Yoganathan, R. Gupta, F.E. Udwadia: Medical and Biological Engineering and Computing. 14(1976)69-73.

[3] J. Herold, R. Schroeder, F. Nasticzky: Medical and Biological Engineering and Computing. 43(2005)451-456.

[4] Y. Mizuno-Matsumoto, S. Ukai, R. Ishii: Brain Topography. 17(2005)237-252.

[5] Yong Li, Xiaorong Gao and Aiwen Guo: Journal of Tsinghua University. 41(2001)77-80.

[6] K. Kim, J. Hong and J. Lim: Lecture Notes in Computer Science. 3658(2005)241-248.

[7] Yi Liu, Caiming Zhang and Yuhua Peng: Chinese Journal of Computers. 29(2006)769-777.

[8] E. Nanou, V. Tsiafakis, E. Kapareliotis: The Environmentalist. 25(2005)173-179. Table 1 Test results of identification models Test sample A B C D E F G 10 10 10 10 10 10 10 Identification results A 8 0 0 0 0 2 0 B 0 7 0 1 0 1 0 C 0 1 8 0 1 0 1 D 1 1 0 7 0 1 0 E 0 0 1 1 7 0 0 F 1 0 1 0 2 6 1 G 0 1 0 1 0 0 8 Accuracy 80% 70% 80% 70% 70% 60% 80% Mean 72.9% Note: A, normal heart sound; B, splitting of first heart sound; C, splitting of second heart sound; D, soft first heart sound; E, loud second heart sound; F, early systolic murmur; G, complete left bundle branch block. Fig.1. Space subdivision of wavelet packet and best basis of heart sound A(a) normal heart sound; B(b) splitting of first heart sound; C(c) splitting of second heart sound; D(d) soft first heart sound; E(e) loud second heart sound; F(f) early systolic murmur; G(g) complete left bundle branch block. Fig.2. Heart sound signals and normalized feature vectors