Heart Sound Signals Based on CNN Classification Research

Heart sound signal can provide complex physiological and pathological information while diagnosing CHD(congenital heart disease). The main procedure includes pretreatment, Envelope extraction, classification and recognition. Pretreatment includes normalization, de-noising, envelope extraction, etc. Then, segment the heartbeat signal through Hilbert Envelope to confirm the first heart sound and the second heartbeat. After that, take logarithm of the periodic signal of the heartbeat to get Hilbert Energy Spectrum. Take the result as characteristic variable and put it into convolution neural network to train the study. The main innovation of this paper is using the convolution neural network to identify the heart sound pattern. The experiment results have a very high correct rate by using CNN (convolutional neural network) compared with other methods.

[1]  Wei Ping Cui,et al.  Improved Wavelet Threshold Denoising Method , 2014 .

[2]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[3]  Gerald Penn,et al.  Convolutional Neural Networks for Speech Recognition , 2014, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[4]  Trevor Darrell,et al.  Caffe: Convolutional Architecture for Fast Feature Embedding , 2014, ACM Multimedia.

[5]  Dong Yu,et al.  Exploring convolutional neural network structures and optimization techniques for speech recognition , 2013, INTERSPEECH.

[6]  Li Deng,et al.  A deep convolutional neural network using heterogeneous pooling for trading acoustic invariance with phonetic confusion , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[7]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[8]  A Matsuyama,et al.  The Application of Wavelet Analysis to ECG Signals , 2004 .