Electrocardiogram Signature For Myocardial Ischemia Acquisition And Intelligent Diagnostic Research

With the richness and diversity of human production and life style,the aging degree of population is increasing day by day,the prevalence of cardiovascular disease continues to increase,and it has become the number one killer of human health.Electrocardiogram,as a way to record human heart activity,is often used in the clinical diagnosis of cardiovascular disease.The most common arrhythmia and myocardial infarction in cardiovascular disease are mainly induced by myocardial ischemia.Therefore,the detection of myocardial ischemia is of great significance for the diagnosis of these two kinds of diseases.Cardiac ischemia is mainly reflected in the change of ST-T segment of electrocardiogram.Therefore,this paper studies the detection of ECG characteristic wave position and the expression of ischemic morphological information,as follows:1.Aiming at the glitch interference and zero-crossing dislocation problems existing in the existing ECG signal signature wave detection algorithm,a T-wave detection algorithm fusing stationary and continuous wavelet transform are proposed.By using the multi-scale information of continuous wavelet transform,the main components of T wave in ECG signal are obtained.Then,the stationary wavelet is used to smooth the T-wave candidate segment,and eliminate the influence of the jagged burr on the peak point detection in the waveform.Finally,time-shift correction is performed on the zero-crossing point of the T wave to ensure that the zero-crossing point can accurately correspond to its peak point during the process of restoring the zerohcrossing point to the original signal,thereby improving the detection accuracy of the T wave.Validated by the MIT-BIH Arrhythmia ECG database,the error rate,sensitivity and correct prediction of the T wave of the ECG signal reached 0.25%,99.87% and 99.88%,respectively.The results of experiment have proved that the algorithm proposed in this paper could accurately detect the position of the characteristic wave of the ECG signal.It provides a solid foundation for the interception of the candidate segments of myocardial ischemia characteristic wave.2.Aiming at the problems of insufficient expression of pathological features and high complexity of training models in existing myocardial ischemia detection algorithms,a multigranular cascade forest-based myocardial ischemia intelligent detection algorithm is proposed.The multi-granularity sliding window is used to scan the ischemic candidate segment frame by frame,and the ischemic features at different moments are merged to achieve data enhancement.Then,random forests with different split attributes are used to deeply express the pathological features of ischemia,so as to realize the deep extraction of pathological information.Averaging processing and gain comparison are added to each cascade of the cascade layer.The averaging processing can avoid the clustering of multiple linear feature vectors in series and cause unnecessary redundant information.The gain comparison can effectively solve the model training layer.The complexity of the number realizes the adaptive ability of the algorithm model in this paper.Validated by Long-term ST-T database,the classification accuracy of 95.89%,95.40% and 94.93% were achieved under the three experimental situations of different training sets accounting for 90%,70% and 30%.Experimental results prove that the algorithm proposed in this paper can truly and effectively express the ischemic characteristics.