Cardiac anomaly detection method based on CycleGAN and BiLSTM neural network method

Abstract

The invention discloses an arrhythmia category detection method based on a CycleGAN and a BiLSTM neural network. The method comprises the following steps: step 1, preprocessing a multi-lead electrocardiosignal data sample: 1.1, collecting electrocardiosignals; 1.2, carrying out de-noising processing on the electrocardiosignal; 1.3, carrying out normalization processing on the electrocardiosignals; 1.4, generating a multi-lead electrocardiosignal sample; step 2, performing data expansion on the preprocessed training database 2 by using a CycleGAN (cyclic generative adversarial network); and step 3, performing cardiac anomaly detection: 3.1, constructing a BiLSTM neural network; 3.2, a training process of the BiLSTM neural network; and 3.3, carrying out classification evaluation. The invention relates to the technical field of electrocardiogram anomaly detection. According to the multi-lead arrhythmia detection method based on the CycleGAN and BiLSTM neural network method, the problem of sample imbalance in the prior art is solved, and a good deep learning model can present a better expression result.

Description

technical field [0001] The invention relates to the technical field of electrocardiogram abnormality detection, in particular to a heartbeat abnormality detection method based on CycleGAN and BiLSTM neural network methods. Background technique [0002] Arrhythmia refers to the formation or conduction disorder of cardiac excitation due to various reasons, so that the activity frequency of the whole or part of the heart is too fast or too slow, and the rhythm is irregular. Clinically, arrhythmias are usually classified according to electrophysiology. Due to the complex structure of the heart, there are many types of arrhythmias. According to scientific research, about 540,000 people die of sudden cardiac death in my country every year, and nearly 90% of sudden deaths are caused by arrhythmia. It can be seen that arrhythmia is very harmful to the human body. [0003] At present, the electrocardiogram has the characteristics of high accuracy, affordable price, and no side effe...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the deficiencies of the prior art, the present invention provides a multi-lead arrhythmia detection method based on CycleGAN and BiLSTM neural network method, which solves the above-mentioned sample imbalance problem and makes a good deep learning model present a better performance results

Images