Table 2

Diagnostic accuracy compared with invasive coronary angiography and FFR

ModelNVesselsDiagnostic accuracy (AUC) compared with ICA-FFRCTCA (AUC)Correlation with ICA-FFRAgreement (bias and agreement unless stated)
Siemens cFFR V.1.4 (Coenen et al 2015)1061890.830.64r=0.59-0.04±0.13 (LOA: −0.31 to 0.22)
(Koo et al 2011)
FFRCT DeFACTO study (Min et al 2012)2524070.810.68r=0.630.058; (95% CI, 0.05-0.07) (mean difference)
FFRCT NXT trial (Nørgaard et al 2014)2544840.900.81r=0.820.03 (LOA: −0.12 to 0.17)
Toshiba CT-FFR (Ko et al 2017)30580.880.77r=0.570.065±0.14 (LOA: −0.20 to 0.33)
NOVEL-FLOW (Chung et al 2017)1172180.930.74r=0.760.01±0.08 (LOA: −0.14 to 0.17)
vFAI (Siogkas et al 2019)63740.970.56r=0.930.03±0.04 (−0.05 to 0.12)
  • Several algorithms have demonstrated superior diagnostic performance of CT-FFR compared with CTCA. A good correlation has been observed in most studies, with a small bias observed compared with invasive FFR.

  • AUC, area under the curve; CTCA, CT coronary angiography; FFR, fractional flow reserve; ICA-FFR, invasive coronary angiography and FFR; LOA, limits of agreement; vFAI, Virtual Functional Assessment Index.