A ECG feature extraction method based on pcanet

Abstract

The technical solution of the present invention includes a method for extracting electrocardiographic features based on PCANet, which is characterized in that it includes the following steps: S10, performing preprocessing on the electrocardiogram to obtain a training set and a set to be classified, S20, using PCANet to separate the training set and the set to be classified The classification set carries out the feature extraction of the heart beat, S30, train the classifier using the heart beat feature extracted from the training set and use it for the classification of the heart beat feature of the set to be classified; the beneficial effect of the present invention is: it is robust to the noise of the electrocardiogram signal, Simplify the steps of noise removal, have a better classification effect on unbalanced heart beats, improve the efficiency and accuracy of ECG feature extraction, reduce the pressure on doctors to identify ECGs, and reduce the probability of doctors' misdiagnosis.

Description

technical field [0001] The invention relates to a PCANet-based ECG feature extraction method, which belongs to the field of medical signal processing. Background technique [0002] At present, with the development of computer technology, pattern recognition technologies such as data mining and deep learning have been gradually applied to medical signal processing. The currently known field of pattern recognition includes technical fields such as electrocardiogram, electroencephalogram, and medical image processing. In the field of electrocardiogram, electrocardiogram-assisted diagnosis and treatment equipment has made great progress. It can mine deep information in electrocardiogram and perform efficient automatic identification. [0003] ECG automatic recognition technology includes three steps, namely preprocessing, feature extraction and classification, that is, the preprocessed cardiac beats are mined for deep features through the feature extraction step, and these feat...

AI Technical Summary

Problems solved by technology

First of all, when we sample the heartbeat through instruments and equipment, the heartbeat waveform we obtain has noise pollution due to the influence of other organs in the body. Generally, the noise of the electrocardiogram (ECG) signal includes baseline drift noise, power frequency noise, etc. , in the prior art, it is necessary to take necessary measures to remove noise from the heartbeat in the preprocessing stage. Currently, there are many noise removal methods such as wavelet analysis and median filter, but the removal of noise will inevitably cause signal noise loss of useful information

Secondly, although many classification algorithms that exist today can obtain better evaluation standards for training and classifying a variety of heart beats that are balanced in number, when training on heart beats with an unbalanced number of samples, they often have significant advantages for heart beats that are superior in number. Classification advantages, but the recognition accuracy of heart beats with less training is low

Images