Font Size: a A A

Diagnostic Performance Of Coronary Computed Tomography Angiography-derived Fractional Flow Reserve(CT-FFR) Based On Deep Learning And Optimized Computational Fluid Dynamics In Myocardial Ischemia

Posted on:2022-12-20Degree:MasterType:Thesis
Country:ChinaCandidate:K XuFull Text:PDF
GTID:2504306611478554Subject:Clinical Medicine
Abstract/Summary:PDF Full Text Request
Objective We sought to validate the diagnostic performance of non-invasive fractional flow reserve(FFR)derived from coronary CT angiography(CT-FFR)based on a new deep learning and optimized computational fluid dynamics(CFD)for myocardial ischemia in stable coronary heart disease,using invasive as standard,and compare the diagnostic efficacy of vessel-level CT-FFR,lesion-level CT-FFR and △CT-FFR for myocardial ischemia.Methods Patients with stable coronary heart disease who had undergone CCTA and met the criteria for inclusion and exclusion in Zhejiang Hospital from January 2019 to June 2021 were prospectively included.All patients underwent invasive coronary angiography and FFR within 60 days after CCTA.Using invasive FFR ≤0.8 as the reference standard of myocardial ischemia,the sensitivity,specificity,positive predictive value(PPV),negative predictive value(NPV)and accuracy of CCTA,vascular-level CT-FFR(CT-FFR value based on the measurement position of invasive FFR pressure guide wire sensor),lesion-level CT-FFR(CT-FFR value 2cm distal to the maximum stenosis of target vessel),△CT-FFR(CT-FFR difference between proximal and distal adjacent to the maximum stenosis of target vessel)in the diagnosis of myocardial ischemia in patients with stable coronary heart disease were evaluated.The area under receiver operator characteristic curve(AUC)of them was also compared.And we compared their correlation and consistency with invasive FFR,and further analyze the influencing factors of their inconsistency with FFR diagnosis using multivariate logistic regression.Results Total 73 patients with 82 vessels were included in this study.Per vessel correlation and consistency analysis showed that vessel-level AccuFFRct and lesion-level AccuFFRct had good correlation and consistency with invasive FFR(Spearman correlation coefficients were 0.725 and 0.636,respectively)and consistency(mean difference was 0.001±0.057 and-0.025± 0.064,respectively).On a per-vessel basis,diagnostic sensitivity,specificity,PPV,NPV,accuracy and AUC of CCTA,vessel-level AccuFFRct,lesion-level AccuFFRct and△AccuFFRct in the diagnosis of myocardial ischemia were 82.8%、34.0%、40.7%、78.3%、51.2%、0.584,89.7%、90.6%、83.9%、94.1%、90.2%、0.945,62.1%、94.3%、85.7%、82.0%、82.9%、0.898 and 69.0%、86.8%、74.1%、83.6%、80.5%、0.862,respectively.On a per-patient basis,diagnostic sensitivity,specificity,PPV,NPV,accuracy and AUC of CCTA,vessel-level AccuFFRct,lesion-level AccuFFRct and △AccuFFRct in the diagnosis of myocardial ischemia were 82.1%、31.1%、42.6%、73.7%、50.7%、0.566,89.3%、91.1%、86.2%、93.2%、90.4%、0.939,60.7%、95.6%、89.5%、79.6%、82.2%、0.898 and 85.9%、68.9%、63.2%、88.6%、75.3%、0.856.Multivariate logistic regression analysis showed that clinical and lesions characteristics were not independent influencing factors of the diagnostic discordance between vessel-level AccuFFRct、lesion-level AccuFFRct、△AccuFFRct and FFR(P>0.05).Conclusion 1.The vessel-level AccuFFRct and lesion-level AccuFFRct have good correlation and consistency with invasive FFR;2.vessel-level AccuFFRct、lesion-level AccuFFRct and AAccuFFRct provided better diagnostic performance compared with CCTA,with vessel-level AccuFFRct being superior in identifying myocardial ischemia;△AccuFFRct was more effective in the diagnosis of myocardial ischemia with high calcification score than those with low calcification score;3.The clinical and lesion characteristics have no impact on the diagnostic efficiency of vessel-level AccuFFRct、lesion-level AccuFFRct、△AccuFFRct in this study.4.Compared with vessel-level AccuFFRct,lesion-level AccuFFRct(AccuFFRct 2 cm distal to the stenosis)has a superior specificity than vessel-level AccuFFRct,which is not dependent on the anatomic location of the lesion,and is helpful to achieve early,accurate and noninvasive functional assessment of coronary stenosis.
Keywords/Search Tags:Coronary CT angiography, Fractional flow reserve, Artificial Intelligence, Myocardial ischemia
PDF Full Text Request
Related items