Fractional flow reserve from intravascular ultrasound imaging: Computational fluid dynamics to the rescue?

Abstract

Key Points• IVUS measurements of luminal area have only modest correlations to FFR.• This study used IVUS 3D anatomical data in a computational fluid dynamic model to estimate FFR with a reasonable degree of accuracy.• Further studies are needed to refine the models to make them more accurate and practical for routine use.Interventionalists like it simple, procedurally speaking. Early on, interventionalists hoped that intravascular ultrasound imaging (IVUS) could provide both anatomic and physiologic information and be a one-stop shop for lesion assessment. Unfortunately, this has never been proven to be the case. Although IVUS accurately measures lesion morphology and vessel dimensions, translating the imaging data into translesional hemodynamics has proven difficult, if not impossible. Minimal lumen area (MLA) by either IVUS or optical coherence tomography (OCT) does not correspond strongly with fractional flow reserve (FFR), except in the extreme ranges of lesion severity. In fact, no single 2D anatomic parameter, or even combination of parameters, can be equated to FFR. 1 The main reasons for this "anatomical-functional mismatch" involve both the specific morphologic features of the stenosis that produce resistance to flow, including lesion length, entrance angle, shape of the lumen, exit angle, as well as the characteristics of the subtended myocardium, including myocardial mass and microvascular resistance. Intracoronary imaging may be able to quantify lesion characteristics, but has to estimate distal myocardial factors indirectly, using vessel size and length. Nevertheless, working toward the goal of generating physiologic data from vascular imaging, Bezerra et al 2 apply computational fluid dynamics (CFD) to the reconstructed 3D IVUS image tube. Using a sophisticated mathematical model, translesional pressure is produced by the CFD algorithm, employing several assumptions about coronary flow and resistance. Simulated hyperemic translesional pressures (assuming 3× flow rates) yield an FFR derived from IVUS (IVUS FFR ).The authors found that in 34 vessels, IVUS FFR significantly correlated with invasive pressure wire FFR (r = 0.79; P < 0.001) with good agreement, and a mean difference of −0.008 ± 0.067 (P = 0.47).Overall accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of IVUS FFR were 91%, 89%, 92%, 80%, and 96%, respectively, to detect FFR < 0.80. Compared to invasive FFR,FFR had a much larger area under the ROC curve (0.93; 95%CI: 0.83-1.00)-better than MLA (0.81; 95%CI: 0.62-1.00) or %DS by visual assessment (0.70; 95%CI: 0.52-0.87). 1 | LIMITATIONS AND ASSUMPTIONS TO GENERATING IVUS F F R For CFD calculations, the imputed variables of basal pressure, flow, and resistance are critical to accuracy. The authors estimated stroke volume from pulse pressure, age, and weight, applying a formula with only moderate predictive power. In this initial proof of concept study, the validity of the estimated variables could have been verified by directly measuring right ...

The assessment of intermediate coronary lesions using intracoronary imaging

“…Firstly, further validation work is required in more diverse lesion and patient cohorts. Secondly, computational time for CFD is still relatively high and this precludes use in a high paced cath lab environment (94). Nevertheless, there is clear potential for the future and there are several industry led collaborations attempting to deliver this technology to clinicians soon.…”