Single-lead ECG arrhythmia detection and classification method based on residual network

Abstract

The invention belongs to the technical field of medical information processing, and discloses a single-lead ECG arrhythmia detection and classification method and system based on a residual network, and the method comprises the steps: carrying out the segmentation of an original ECG signal: taking one second as a window length, and carrying out the segmentation of the original ECG signal; processing and the signals by using a residual error network: inputting the segmented data into the network, and the processed network output result being the identification result of the corresponding ECG signal. According to the method, the original electrocardiosignal does not need to be subjected to sub-shooting processing, and any alignment is not needed. Classification and identification of normal heart beat, left bundle branch conduction blocking, right bundle branch conduction blocking, atrial premature beat, abnormal atrial premature beat, boundary premature beat, ventricular premature beat,supraventricular premature beat, ventricular fusion heartbeat, atrial escape beat, boundary escape beat, ventricular escape beat, pacing heart beat and pacing fusion heartbeat can be realized. Test comprehensive accuracy on MIT-BIH arrhythmia database reaches 96% or above.

Description

technical field [0001] The invention belongs to the technical field of medical information processing, in particular to a residual network-based single-lead ECG arrhythmia detection and classification method. Background technique [0002] Currently, the closest prior art: [0003] In the prior art, the method for extracting the characteristic parameters of arrhythmia, the device for identifying arrhythmia, and the computer-readable medium (public number: 108852347A) need to do parameter and feature extraction, and this method does not need to perform feature extraction . [0004] In the prior art, the arrhythmia recognition and classification method based on sparse representation and neural network (public number: 108647584A) requires parameter extraction and complicated preprocessing (the original ECG needs to be sub-shot, and dimensionality reduction and other processing are required), only Can distinguish 6 types of arrhythmia. [0005] 3Robust ECG Signal Classificatio...

AI Technical Summary

Problems solved by technology

[0004] In the prior art, the arrhythmia recognition and classification method based on sparse representation and neural network (public number: 108647584A) requires parameter extraction and complicated preprocessing (the original ECG needs to be sub-shot, and dimensionality reduction and other processing are required), only Can distinguish 6 types of arrhythmia

[0005] 3Robust ECG Signal Classification for Detection of Atrial Fibrillation Using a Novel Neural Network, using a convolutional neural network, but can only identify normal, atrial fibrillation, noise and other four categories, and the comprehensive accuracy rate can only reach 82%

[0006] 4Cardiologist-Level Arrhythmia Detection with Convolutional NeuralNetworks, using a convolutional neural network and identifying in a sequential manner, but only 12 types of arrhythmia can be identified, and the comprehensive accuracy rate is lower than 80%

[0015] To sum up, how to use less lead data, or even just use one lead data, directly process the original ECG signal as a sequence without sub-beating and manual feature extraction, and realize multiple arrhythmia Efficient and accurate identification of homogeneous types is relatively difficult, but it is also worth studying

Images