Automated Aortic Root Assessment By CT In TAVR Patients

Objectives:1.The aim of the study is to evaluate the correlation,concordance and the reproducibility of the routine manual measurements with the automated CT analysis software(eSieValve),which might be used to assess the aortic root in the patients undergoing transcatheter aortic valve replacement(TAVR).2.To analysis the predictive value in the selection of the valve size by the automated CT analysis software.3.To explore the predictive value of the automatic measurement software for perivalvular leakage(PVL).Methods:1.Two hundred seventy-three patients were enrolled in the research.The aortic root and annulus measurements were performed by both manual and eSie valve.Correlation,concordance and reproducibility between the manual method,fully-automatic method,and semi-automatic method were analyzed.Student t test was used to compare continuous data paired.Pearson correlations and Bland-Altman analysis were used to compare correlation and concordance among the methods.Intraclass correlation coefficients(ICC)was used to evaluate the reproducibility among the methods.2.Valve size determined by three methods with different annulus measurements were compared with actual size chosen in the TAVR procedure.The consistent rate,underestimate rate and the overestimate rate were recorded as percentage.Chi-square test and Fisher's exact test was used to analysis the distribution of the valve size among the different methods.3.The difference between anulus value measured before the procedure and when fully-expanded during the procedure are subsequently assessed for prediction of PVL:(1)the value of the annulus measurements in nominal valve-the annulus value measured before the procedure.(2)(the value of the annulus measurements in nominal valve-the annulus value measured before the procedure)/annulus value measured before the procedure.Logistic regression and Receiver-operating characteristic(ROC)curve was used to predict the affecting factors to PVL,areas under the curve(AUC)were compared to find the better affecting factor.Results:1.The annulus measurements from semi-automatic method and routine manual method showed significant correlation(R=0.69-0.96,p<0.001).The reproducibility in the semi-automatic method was more prominent than the manual method(intraobserver ICC 0.91-0.96,intreobserver ICC 0.89-0.93).2.Compared with the valve size chosen during the procedure,the valve size chosen by the semi-automatic method showes better consistent rate in the three methods(89.04%in Corevalve,91.67%in Sapine XT and 92.59%in Sapien 3).The result of the manual method is not good as the semi-automatic method,and the consistent rate is range from 61.11%to 85.19%.The perimeter-derived aortic valve sizing method shows superiority to the other annulus measurements.The fully-auto method shows relatively poor result in this part.It was obvious that the valve size was significantly underestimated by the fully-auto method,and the maximun underestimate rate is 61.9%.3.The difference value of the area that measured before the procedure and the fully-expanded during the procedure in the semi-automatical method is the relatively better indicator in the prediction of PVL(AUC=0.75).Conclusion:1.The automated software(eSie valve)with minor correction shows good correlation,concordance and reproducibility in the measurements of the aortic anulus.2.The automated software(eSie valve)with minor correction showes better predictive value in choosing the valve size.3.The semi-automatic method shows better predictive value in PVL,and can provid more comprehensive information in the valve size chosen in the clinical work.