Analysis Of Ventricular Repolarization Variability In Coronary Heart Disease And Its Study For Assisting Diagnosis

Coronary heart disease(CHD)seriously endangers the health of Chinese residents and brings heavy burdens to the society and families.How to avoid the occurrence of CHD and reduce the harm of CHD have become the major topics that need to be studied and solved seriously.The electrocardiogram(ECG)examination is the most basic and commonly used noninvasive approach for clinical diagnosis of CHD.Since myocardial ischemia could result in abnormal ventricular repolarization(VR)and increased trans-mural dispersion of VR,the ECG markers which denote changes in VR have received widespread attention,in which representative methods include QT interval variability(QTV),TpTe interval variability(TpTeV),corrected QT(QTc)interval,and TpTe/QT ratio.In this study,the above four kinds of markers are termed as VR indicators;QTV and TpTeV are termed as VR variability(VRV)indicators representing the beat-to-beat changes of VRV in time course and the transmural dispersion of VR,respectively.Among the various methods of analyzing VRV,permutation entropy(PEn)based on the symbolic dynamics has been widely applied.However,PEn still has three short-comings.Though many researchers have successively proposed a variety of improved algorithms,these algorithms only focus on one or two of these three shortcomings and lack comprehensive consideration.Existing studies of VRV analysis for CHD mainly focus on the changes between healthy controls and CHD patients without myocardial infarction(MI)or MI patients and lacks the analysis of VRV between MI patients and CHD patients without MI.Moreover,there are many gaps in the VRV analysis between healthy individuals and CHD patients without MI or MI patients.Besides,though a lot of clinical studies have demonstrated the association of the VR indicators and the ECG morphologic changes reflecting the process of VR(e.g.,T wave inversion,ST segment elevation or depression)with myocardial ischemia and infarction and its usefulness in risk stratification and guiding prognosis for CHD,their effectiveness in assisting auto-matic CHD diagnosis based on the resting ECG is still unclear.This study was carried out around the above three problems.Taking VRV analysis as the core,this stduy proposed a comprehensive improvement scheme of PEn and sys-tematically analyzed the changes of the above VR indicators in healthy controls,MI patients and CHD patients without MI,the correlation of VRV indicators with the QTc interval and TpTe/QT ratio,the influence of gender and the degree of coronary artery stenosis on the above VR indicators,and the effectiveness of the VR index and ST-T segment morphological characteristics in assisting the automatic CHD diagnosis based on the resting ECG.The main work and innovation points are summarized as follows.(1)A comprehensively improved PEn algorithm,edge dispersion entropy(EDEn),is fisrt proposed,and then the idea of multiscale is introduced to extend EDEn to multi-scale edge dispersion entropy(MEDEn).EDEn is developed based on the concepts of both edge permutation entropy(EPEn)and dispersion entropy(DEn).EDEn not only takes account of the information of element amplitude and the whole fluctuation of the sequence,but also eliminates the influence of elements with equal amplitude.In this study,the results show that under different values of parameters,EDEn and MEDEn can correctly measure the irregularity and complexity of the sequence,respectively.As compared with PEn and EPEn,the differentiating ability of EDEn is greatly improved.In the case of low dimensions and large number of classes,the differentiating perfor-mance of EDEn is better than that of DEn.(2)The VRV changes among healthy controls,CHD patients without MI and MI patients,as well as their relationships with heart rate variability,gender,and the degree of coronary artery stenosis were systematically analyzed.This study shows that myo-cardial ischemia is associated with elevated TpTeV.When compared with myocardial ischemia,MI can further increase QTV,which is related to left ventricular remodeling after MI.Myocardial ischemia alone does not significantly affect QT-RR coupling,but MI can significantly reduce QT-RR coupling.QTV and TpTeV in time and frequency domains have stronger ability to characterize myocardial ischemia and infarction than non-linear indexes and have better effectiveness as risk assessment factors.In different populations,the correlations of QTV and TpTeV with QTc interval and TpTe/QT ratio were different.The QTc interval and TpTe/QT ratio were significantly higher in women than in men,while there were no significant gender differences in other studied VR indexes.The degree of coronary artery stenosis had little correlation with the above VR indexes.(3)Two automatic diagnosis systems for CHD were developed on the basis of the VR index and the fusion of the above index with ST-T segment waveform features.This study confirms that both the VR index and the ST-T segment waveform features could effectively improve the performence of automatic CHD diagnosis.The accuracy,sensi-tivity and specificity of the system based on feature fusion were 96.16%,95.75%and 96.40%,which are significantly better than the classification performence of using fea-tures derived from the heart rate variability,the VR index or ST-T segment waveform features alone.This study fills in some gaps in the VRV analysis of CHD patients,which not only provides more objective basis and strong technical support for the clinical application of the VRV analysis,but also provides a new idea for the study of automatic diagnosis of CHD based on resting ECG.