Research On Algorithm Of T-wave Detection Based On Morphological Guidance

With the environmental deterioration,the incidence of heart disease is also increasing year by year,which has been a serious threat to human health,especially the sudden cardiac death with sudden,high incidence and high mortality characteristics,making the number of deaths caused by sudden cardiac death is increasing year by year.Therefore,the cardiovascular disease-related research has become a hot topic in the field of medical research,which puts forward a higher requirements for technology to prevent cardiovascular disease,especially sudden cardiac death and the study of ECG signal analysis and diagnostic.Studies have shown that before the occurrence of cardiovascular disease,the ECG of patients will occur Q-T morphology and interval changes.Therefore,the T wave as the core characteristic of the Q-T band is the key to the early prediction of cardiovascular disease.As the T wave signal is weak and susceptible to noise interference as well as changing morphologically,with the deepening of cardiovascular disease research,T wave morphological classification and identification of key feature points has gradually become the focus and difficulties of research.In the ECG signal,the time-frequency characteristics of some noise are complex and the distribution is unknown.At the same time,the position and shape of the T wave are changed,which needs to the improvement of the existing method.With the advent of the mobile Internet era,telemedicine technology,especially the formation of remote ECG monitoring system,making the adoption of big data processing and improving the accuracy of diagnosis become an important research direction to early prediction of cardiovascular disease.In this paper,in the context of remote ECG monitoring system center with much noise and interference and other factors,the author considers the differences between different individuals and combines the deep learning.Using the advantages of ECG's big data characteristics to research ECG signal Morphological Classification and Automatic Detection of T Wave Point Key Feature Points.The main work is as follows:(1)ECG data preprocessing to construct T-wave candidate segments with complex redundant information.In this paper,for contradictory relationship between the T-wave morphological classification and the feature points,the author first puts forward the concept of T-band containing complex redundant information.The adaptive threshold threshold wavelet method is used to filter the ECG signal,and the candidate T-band is extracted and the training set and the test are extracted by taking the R wave crest location and using the relevant medical research results to fully consider the difference between different individuals.Then the author has done the preparatory work of the T wave automatic detection algorithm.(2)To build convolution neural network(CNN,convolutional neural network),to shape classification of the T wave.This thesis is based on the results of modern medical research,combined with ECG specific information,to carry out T wave morphological identification classification research.In this paper,the candidate T wave extracted from the ECG data preprocessing is input into the CNN model for one-dimensional ECG signal construction.The shape identification of the T wave is realized by tuning the CNN model.Because the CNN model has the advantages of sparse links exits between the hidden layers and the sharing of implicit layer training parameters,which can extract more effective T-wave abstract features,and for the location of the small offset and noise has a better robust Sex.Finally,the experimental data of QT database are used to verify the experiment.Finally,it is shown that the method can identify five T-waves with different shapes without notifying the key feature points of T wave,and the correct recognition rate of T wave is 99.2%.(3)Morphological guidance-based automatic detection algorithm of T-wave key feature points is proposed.In this paper,the author uses the result of T wave morphological classification,on the basis of characteristics of T wave morphological classification,constructs Monte Carlo Markov chain by Gibbs sampling method,with the correlation between the model and the T-band,the key feature point detection of T wave is realized.Using the standardized data in the QT database to verify the experiment,it is concluded that the proposed algorithm has a great improvement in accuracy compared with the traditional T wave detection algorithm.