A multi-classification method for heart sounds based on deep convolutional neural network

Abstract

The invention discloses a heart sound multi-classification method based on a deep convolutional neural network and relates to the field of heart sound classification based on deep learning. The methodcomprises the following steps: 1) processing acquired original heart sound data so as to obtain N segments of heart sound signals; 2) inputting the N segments of heart sound signals into a heart sound classification model based on a two-dimensional convolutional neural network and a one-dimensional convolutional neural network, and classifying to obtain 2N classification results according to frequency domain and time domain features; and 3) training the 2N classification results by adopting a Lasso framework so as to obtain corresponding weights, and multiplying the weights by the 2N classification results, thereby obtaining the final classification result. According to the method disclosed by the invention, the problems that low resolution performance is caused by adopting the two-dimensional convolutional neural network only in the existing heart sound classification method and low classification accuracy is caused by adopting one multi-classifier for classifying are solved, and theeffect of improving the heart sound multi-classification accuracy is achieved.

Description

technical field [0001] The invention relates to the field of heart sound classification based on deep learning, in particular to a heart sound multi-classification method based on a deep convolutional neural network. Background technique [0002] In 2014, the world's authoritative medical journal "The Lancet" released the "Global Burden of Disease Report 2013", which evaluated the death situation in 188 countries from 1990 to 2013. According to the report, the three diseases with the highest death rate in China Stroke, coronary heart disease and chronic obstructive pulmonary disease accounted for 46% of all deaths in 2013, which shows that two of the three deadliest health killers in China are cardiovascular diseases; heart sounds are used to detect heart health It is a cheap and non-invasive method that is widely used, which is convenient for timely detection of heart problems and early treatment; changes in the physical structure of the heart will cause changes in the hear...

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is: the present invention provides a kind of heart sound multi-classification method based on deep convolutional neural network, solves the problem that the existing heart sound classification method only adopts two-dimensional convolutional network to cause low resolution performance, and adopts a multi-classifier to perform Classification leads to the problem of low classification accuracy

Images