Electrocardiogram And Echocardiography Detection Methods For Auxiliary Diagnosis

The physiological information of the heart can provide a favorable basis for medical diagnosis,and its related technologies are extensively used in clinical examination practice.With the rapid development of artificial intelligence,the intelligent diagnosis methods based on deep learning are being gradually applied in the medical field,and remarkable results have been achieved.By analyzing the characteristics of cardiac physiological information,this paper focuses on studying the detection methods of cardiac physiological information based on electrocardiogram and echocardiogram by starting from the clinical test information on the heart,so as to improve the recognition accuracy of detection algorithm,shorten the time consumed by the manual diagnosis process,and provide more effective help for heart disease diagnosis.The main work is as follows:The intelligent classification and recognition of electrocardiogram(ECG)signals are studied.A time network model based on pre convolution is proposed aiming at the problems of difficulty in parallel data processing,deficiency in long time memory and lack of fast and accurate location of target recognition in one-dimensional time series signal recognition and classification.In this method,a pre convolution optimization module is designed to extract the deep features of ECG signals and a network of large range data sensing layer groups is used to increase the memory length of the network.A network model for fast target detection based on two-dimensional coordinate ECG images is proposed by taking the advantages of two-dimensional image form in information processing and expression.This method converts the ECG signal from one-dimensional time series form to two-dimensional picture form,which can not only identify and analyze a complete beat in the heart conduction cycle,but also use the clinical ECG signals of sinus rhythm to recognize and analyze the waveform one by one,so as to improve the efficiency of ECG information aided diagnosis.The problem of target region recognition in echocardiogram is studied.A target region recognition model based on dense pixel feature detection network is proposed for the phenomenon of repeated storages and calculations in the process of region recognition of ultrasonic images.This method uses two-dimensional cardiac ultrasound technology to extract the regions with dense pixel features from the cardiac ultrasonic images at the end of diastole and the end of systole respectively under the clinical apical four chamber view,and then correctly identify the left ventricular region and related important parameters,so as to provide effective assistance in improving the accuracy of clinical cardiac function evaluation.The joint recognition of electrocardiogram and echocardiogram is studied.A multi-source information recognition model with the combination of Doppler electrocardiogram and echocardiogram as the research object is proposed in order to solve the problems of complex cardiac system that needs to combine multiple types of information analysis,the poor real-time correlation between different types of data,and the time-consuming manual diagnosis process etc.The mitral valve blood flow spectrogram,aortic valve blood flow spectrogram,and pulmonary valve blood flow spectrogram are collected in the same view as the electrocardiogram to identify the key targets and related parameters which are used for the analysis of the coordination of the three types of heart movement between the left atrium and left ventricle,between the left and right ventricles and in the left ventricle.This can improve the efficiency of combined clinical analysis on the electrocardiogram and echocardiogram data.