Diagnostic performance of a Lattice Boltzmann-based method for CT-based fractional flow reserve

EuroIntervention. 2018 Feb 20;13(14):1696-1704. doi: 10.4244/EIJ-D-17-00019.

Abstract

Aims: Fractional flow reserve (FFR) estimated from coronary computed tomography angiography (CT-FFR) offers non-invasive detection of lesion-specific ischaemia. We aimed to develop and validate a fast CT-FFR algorithm utilising the Lattice Boltzmann method for blood flow simulation (LBM CT-FFR).

Methods and results: Sixty-four patients with clinically indicated CTA and invasive FFR measurement from three institutions were retrospectively analysed. CT-FFR was performed using an onsite tool interfacing with a commercial Lattice Boltzmann fluid dynamics cloud-based platform. Diagnostic accuracy of LBM CT-FFR ≤0.8 and percent diameter stenosis >50% by CTA to detect invasive FFR ≤0.8 were compared using area under the receiver operating characteristic curve (AUC). Sixty patients successfully underwent LBM CT-FFR analysis; 29 of 73 lesions in 69 vessels had invasive FFR ≤0.8. Total time to perform LBM CT-FFR was 40±10 min. Compared to invasive FFR, LBM CT-FFR had good correlation (r=0.64), small bias (0.009) and good limits of agreement (-0.223 to 0.206). The AUC of LBM CT-FFR (AUC=0.894, 95% confidence interval [CI]: 0.792-0.996) was significantly higher than CTA (AUC=0.685, 95% CI: 0.576-0.794) to detect FFR ≤0.8 (p=0.0021). Per-lesion specificity, sensitivity, and accuracy of LBM CT-FFR were 97.7%, 79.3%, and 90.4%, respectively.

Conclusions: LBM CT-FFR has very good diagnostic accuracy to detect lesion-specific ischaemia (FFR ≤0.8) and can be performed in less than one hour.

MeSH terms

  • Aged
  • Area Under Curve
  • Computed Tomography Angiography / methods*
  • Female
  • Fractional Flow Reserve, Myocardial*
  • Humans
  • Male
  • Middle Aged
  • Myocardial Ischemia / diagnostic imaging*
  • Retrospective Studies