Preliminary Study On Quantitative Analysis Of Coronary CTA Myocardial Image Texture To Predict Myocardial Infarction

Background: Myocardial infarction is the leading cause of mortality in coronary heart disease,and one of the leading causes of mortality and disability worldwide.After myocardial infarction,a large number of myocardial cells undergo necrosis and apoptosis,which reduces cardiac function and gradually leads to heart failure,posing a great threat to human health.It has become an important social public problem and is attracting great attention from the whole society,especially the medical community.Therefore,the early diagnosis of myocardial infarction,the time of coronary artery occlusion and the early PCI reperfusion of myocardium are of great significance for the improvement and recovery of cardiac structure and function,the improvement of survival rate and the improvement of quality of life.Conventional coronary artery CTA can quickly show the vascular involvement and plays an important role in the diagnosis and management of myocardial infarction.However,it is still insufficient to evaluate the local blood perfusion of myocardial tissue.Texture analysis is a new technology emerging this year.By quantifying the gray scale variation rule of pixels in medical images,it can make up for the deficiency of coronary artery CTA to some extent and improve the accuracy of the diagnosis of myocardial infarction.Objective: To explore the feasibility of predicting myocardial infarction by texture analysis of coronary artery CTA myocardial image.Methods: Selected from the outpatient and inpatient of department of cardiology of our hospital from February 2016 to October 2019,25 patients with myocardial infarction(MI)(13 males;average age of 55±11 years)and 25 patients with no cardiac abnormalities(15 males,average age of 40±20 years)are selected as the control group.All patients had done CT coronary artery imaging and SPECT myocardial perfusion,The original images of SPECT myocardial perfusion were fused with the original coronary CTA images by a post-processing software.and finding the location and range of myocardial perfusion abnormalities in the fusion image,on the basis of image fusion and combining with the abnormal resting myocardial perfusion imaging(perfusion reduction/defect)location and range,Roentgenologists manually Delineated abnormal perfusion areas of interest(ROI)in coronary CTA images on the same position.in the control group,in the same position outline the scope of the same ROI.Using Ma Zda texture analysis software wavelets ransform(WAV),grey-level co-occurrence matrix(GLCM),absolute gradient(GRA),run-length matrix(RUN),autoregressive model(ARM)and histogram analysis methods,The texture features of the myocardial infarction group and the control group were extracted from the CTA images,and 256 texture parameters were obtained.Four statistical methods: Fiher correlation coefficient(Fisher),minimum classification misjudgment rate + average correlation coefficient(POE+ACC),mutual trust coefficient(MI)and the combination of the first three(Fisher+POE+ACC+MI,FPM)were used to screen out the best texture feature parameters that distinguish acute myocardial infarction from control group.Then the optimum parameters of the texture in the Ma Zda texture analysis software used to own dimensionality of classification on the procedures of bl1,the methods inciude the raw data analysis(RDA),principal component analysis(PCA),nonlinear differential analysis method(NDA)and linear discrimination analysis(LDA),the final evaluation optimal texture feature parameters under different combination statistical methods to identify the misjudgment rate(R)of myocardial infarction,And the receiver operating characteristic curve(ROC)was established for statistically significant texture parameters,and its efficacy in diagnosing acute myocardial infarction was compared and analyzed.Results:(1)statistical analysis of general data showed that there was no statistical significance in age and gender between the myocardial infarction group and the normal group,P>0.05,and there were differences in smoking and drinking history,history of hypertension,history of hyperlipidemia,history of diabetes and body mass index(BMI > 30 kg/m 2),P<0.05.There was no statistical significance in heart rate,troponin positive rate,st-segment elevation myocardial infarction rate,involved vessels,number of segments of the involved myocardium,and whether it was transmural between men and women during CTA,P<0.05.(2)different combinations filter the dominant texture,The miscarriage rates are: Fisher+RDA 42%.Fisher+PCA 42%.Fisher+LDA 22%.Fisher+NDA 16%;POE+ACC+RDA 36%.POE+ACC+PCA 32%.POE+ACC+PCA 22%.POE+ACC+PCA 14%;MI+RDA 38%.MI+PCA 32%.MI+LDA 28%.MI+ NDA 12%;FPM+RDA34%.FPM+PCA30%.FPM+LDA6%.FPM+NDA4%.Compared with other methods,FPM++NDA combination has the lowest misdiagnosis rate in differential diagnosis,and the discrimination ability and the statistical extraction and dimensionality reduction methods of dominant texture features are as well.(3)the logistic shows the AUC of S(0,2)Entropy=0.80,95%CI =0.663-0.900,the cut-off value is >3.14,the sensitivity=0.800,the specificity = 0.88,and the Jordan index =0.68.The AUC of Perc.01% = 0.861,95%CI = 0.733-0.942,the cut-off value >249.65,the sensitivity = 100%,the specificity = 0.72,and the Jordan index = 0.72.The AUC of S(0,2)Sum Averg= 0.839,95%CI = 0.708-0.928,the cut-off value >32.00,the sensitivity = 100%,the specificity = 0.56,and the Jordan index = 0.56.The AUC of the three combinations = 0.958,95%CI = 0.860-0.994,the sensitivity =0.920,the specificity =0.92,and the Jordan index = 0.84.Conclusion :(1)studies show that it is feasible to predict myocardial infarction by quantitative analysis of myocardial image texture of coronary artery CTA;(2)S(0,2)Entropy,perc.01%,S(0,2)Sum Averg are the independent factors that distinguish the myocardial infarction group from the control group,and Perc.01% has the best diagnostic performance;The combined diagnostic effectiveness of three parameters is greater than that of a single parameter;(3)the values of S(0,2)Entropy,perc.01%,S(0,2)Sum Averg are all increased during myocardial infarction;(4)myocardial texture analysis combined with coronary artery CTA examination can simultaneously evaluate coronary vascular disease and myocardial infarction status,which has important reference significance for "one-stop" cardiac examination.