Diagnostic accuracy of quantitative flow ratio for assessment of coronary stenosis significance from a single angiographic view: A novel method based on bifurcation fractal law

Abstract

Objectives We aimed to evaluate the diagnostic accuracy of computation of fractional flow reserve (FFR) from a single angiographic view in patients with intermediate coronary stenosis. Background Computation of quantitative flow ratio (QFR) from a single angiographic view might increase the feasibility of routine use of computational FFR. In addition, current QFR solutions assume a linear tapering of the reference vessel size, which might decrease the diagnostic accuracy in the presence of the physiologically significant bifurcation lesions. Methods An artificial intelligence algorithm was proposed for automatic delineation of lumen contours of major epicardial coronary arteries including their side branches. A step‐down reference diameter function was reconstructed based on the Murray bifurcation fractal law and used for QFR computation. Validation of this Murray law‐based QFR (μQFR) was performed on the FAVOR II China study population. The μQFR was computed separately in two angiographic projections, starting with the one with optimal angiographic image quality. Hemodynamically significant coronary stenosis was defined by pressure wire‐derived FFR ≤0.80. Results The μQFR was successfully computed in all 330 vessels of 306 patients. There was excellent correlation (r = 0.90, p < .001) and agreement (mean difference = 0.00 ± 0.05, p = .378) between μQFR and FFR. The vessel‐level diagnostic accuracy for μQFR to identify hemodynamically significant stenosis was 93.0% (95% CI: 90.3 to 95.8%), with sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio of 87.5% (95% CI: 80.2 to 92.8%), 96.2% (95% CI: 92.6 to 98.3%), 92.9% (95% CI: 86.5 to 96.9%), 93.1% (95% CI: 88.9 to 96.1%), 23.0 (95% CI: 11.6 to 45.5), 0.13 (95% CI: 0.08 to 0.20), respectively. Use of suboptimal angiographic image view slightly decreased the diagnostic accuracy of μQFR (AUC = 0.97 versus 0.92, difference = 0.05, p < .001). Intra‐ and inter‐observer variability for μQFR computation was 0.00 ± 0.03, and 0.00 ± 0.03, respectively. Average analysis time for μQFR was 67 ± 22 s. Conclusions Computation of μQFR from a single angiographic view has high feasibility and excellent diagnostic accuracy in identifying hemodynamically significant coronary stenosis. The short analysis time and good reproducibility of μQFR bear potential of wider adoption of physiological assessment in the catheterization laboratory.

Functional Assessment of Cerebral Artery Stenosis by Angiography-Based Quantitative Flow Ratio: A Pilot Study

“…The new method for computing QFR took side branches into account, closer to the natural bifurcation physiology. A step-down reference diameter function based on the Murray bifurcation fractal law, instead of assuming linear tapering, was applied to reconstruct a more accurate reference vessel size (Tu et al, 2021). Although multiangle angiography can theoretically provide a more comprehensive description of lesion morphology, the suboptimal angiography view is often shortened and overlapped due to the curvature of cerebral arteries.…”

“…The image with the lesion segment fully expanded and clearly delineated was selected as the analysis frame. The detailed computational algorithms of the QFR have been described previously (Tu et al, 2021). They can be summarized as follows:…”

“…QFR has been widely validated showing a significant diagnostic performance and prognostic value for coronary artery stenosis (Westra et al, 2018;Biscaglia et al, 2019;Tang et al, 2021), validated in the latest multicenter randomized controlled trial (Xu et al, 2021). Recently, a new method for computing QFR from a single angiographic view, based on artificial intelligence and the Murray bifurcation fractal law, exhibited high feasibility and excellent diagnostic accuracy (Tu et al, 2021).…”

Long‐term prognostic value of dynamic function assessment of intermediate coronary lesion with computational physiology

“…The availability of QFR can be improved by specifying appropriate equipment and modes before angiography and following specific angiographic guidelines; however, the extent to which this can be improved remains to be assessed. Next‐generation QFR systems, which require a single projection and incorporate more automated processes, might increase QFR availability in retrospective studies 28 . In addition, some baseline characteristics were not evenly distributed between the included and excluded patients.…”

“…Next-generation QFR systems, which require a single projection and incorporate more automated processes, might increase QFR availability in retrospective studies. 28 In addition, some baseline characteristics were not evenly distributed between the included and excluded patients. The extent to which these considerations may affect the external validity of the present results is unknown.…”