Incorporating clinical parameters to improve the accuracy of angiography-derived computed fractional flow reserve.

Aims: Angiography-derived fractional flow reserve (angio-FFR) permits physiological lesion assessment without the need for an invasive pressure wire or induction of hyperaemia. However, accuracy is limited by assumptions made when defining the distal boundary, namely coronary microvascular resistance (CMVR). We sought to determine whether machine learning (ML) techniques could provide a patient-specific estimate of CMVR and therefore improve the accuracy of angio-FFR. Methods and results: Patients with chronic coronary syndromes underwent coronary angiography with FFR assessment. Vessel-specific CMVR was computed using a three-dimensional computational fluid dynamics simulation with invasively measured proximal and distal pressures applied as boundary conditions. Predictive models were created using non-linear autoregressive moving average with exogenous input (NARMAX) modelling with computed CMVR as the dependent variable. Angio-FFR (VIRTUheart™) was computed using previously described methods. Three simulations were run: using a generic CMVR value (Model A); using ML-predicted CMVR based upon simple clinical data (Model B); and using ML-predicted CMVR also incorporating echocardiographic data (Model C). The diagnostic (FFR ≤ or >0.80) and absolute accuracies of these models were compared. Eighty-four patients underwent coronary angiography with FFR assessment in 157 vessels. The mean measured FFR was 0.79 (±0.15). The diagnostic and absolute accuracies of each personalized model were: (A) 73% and ±0.10; (B) 81% and ±0.07; and (C) 89% and ±0.05, P < 0.001. Conclusion: The accuracy of angio-FFR was dependent in part upon CMVR estimation. Personalization of CMVR from standard clinical data resulted in a significant reduction in angio-FFR error.

Virtual (Computed) Fractional Flow Reserve: Future Role in Acute Coronary Syndromes.

The current management of acute coronary syndromes (ACS) is with an invasive strategy to guide treatment. However, identifying the lesions which are physiologically significant can be challenging. Non-invasive imaging is generally not appropriate or timely in the acute setting, so the decision is generally based upon visual assessment of the angiogram, supplemented in a small minority by invasive pressure wire studies using fractional flow reserve (FFR) or related indices. Whilst pressure wire usage is slowly increasing, it is not feasible in many vessels, patients and situations. Limited evidence for the use of FFR in non-ST elevation (NSTE) ACS suggests a 25% change in management, compared with traditional assessment, with a shift from more to less extensive revascularisation. Virtual (computed) FFR (vFFR), which uses a 3D model of the coronary arteries constructed from the invasive angiogram, and application of the physical laws of fluid flow, has the potential to be used more widely in this situation. It is less invasive, fast and can be integrated into catheter laboratory software. For severe lesions, or mild disease, it is probably not required, but it could improve the management of moderate disease in 'real time' for patients with non-ST elevation acute coronary syndromes (NSTE-ACS), and in bystander disease in ST elevation myocardial infarction. Its practicability and impact in the acute setting need to be tested, but the underpinning science and potential benefits for rapid and streamlined decision-making are enticing.

The importance of three dimensional coronary artery reconstruction accuracy when computing virtual fractional flow reserve from invasive angiography.

Three dimensional (3D) coronary anatomy, reconstructed from coronary angiography (CA), is now being used as the basis to compute 'virtual' fractional flow reserve (vFFR), and thereby guide treatment decisions in patients with coronary artery disease (CAD). Reconstruction accuracy is therefore important. Yet the methods required remain poorly validated. Furthermore, the magnitude of vFFR error arising from reconstruction is unkown. We aimed to validate a method for 3D CA reconstruction and determine the effect this had upon the accuracy of vFFR. Clinically realistic coronary phantom models were created comprosing seven standard stenoses in aluminium and 15 patient-based 3D-printed, imaged with CA, three times, according to standard clinical protocols, yielding 66 datasets. Each was reconstructed using epipolar line projection and intersection. All reconstructions were compared against the real phantom models in terms of minimal lumen diameter, centreline and surface similarity. 3D-printed reconstructions (n = 45) and the reference files from which they were printed underwent vFFR computation, and the results were compared. The average error in reconstructing minimum lumen diameter (MLD) was 0.05 (± 0.03 mm) which was < 1% (95% CI 0.13-1.61%) compared with caliper measurement. Overall surface similarity was excellent (Hausdorff distance 0.65 mm). Errors in 3D CA reconstruction accounted for an error in vFFR of ± 0.06 (Bland Altman 95% limits of agreement). Errors arising from the epipolar line projection method used to reconstruct 3D coronary anatomy from CA are small but contribute to clinically relevant errors when used to compute vFFR.

Operator-dependent variability of angiography-derived fractional flow reserve and the implications for treatment.

AIMS: To extend the benefits of physiologically guided percutaneous coronary intervention to many more patients, angiography-derived, or 'virtual' fractional flow reserve (vFFR) has been developed, in which FFR is computed, based upon the images, instead of being measured invasively. The effect of operator experience with these methods upon vFFR accuracy remains unknown. We investigated variability in vFFR results based upon operator experience with image-based computational modelling techniques. METHODS AND RESULTS: Virtual fractional flow reserve was computed using a proprietary method (VIRTUheart) from the invasive angiograms of patients with coronary artery disease. Each case was processed by an expert (>100 vFFR cases) and a non-expert (<20 vFFR cases) operator and results were compared. The primary outcome was the variability in vFFR between experts and non-experts and the impact this had upon treatment strategy (PCI vs. conservative management). Two hundred and thirty-one vessels (199 patients) were processed. Mean non-expert and expert vFFRs were similar overall [0.76 (0.13) and 0.77 (0.16)] but there was significant variability between individual results (variability coefficient 12%, intraclass correlation coefficient 0.58), with only moderate agreement (κ = 0.46), and this led to a statistically significant change in management strategy in 27% of cases. Variability was significantly lower, and agreement higher, for expert operators; a change in their recommended management occurred in 10% of repeated expert measurements and 14% of inter-expert measurements. CONCLUSION: Virtual fractional flow reserve results are influenced by operator experience of vFFR processing. This had implications for treatment allocation. These results highlight the importance of training and quality assurance to ensure reliable, repeatable vFFR results.

The new role of diagnostic angiography in coronary physiological assessment.

The role of 'stand-alone' coronary angiography (CAG) in the management of patients with chronic coronary syndromes is the subject of debate, with arguments for its replacement with CT angiography on the one hand and its confinement to the interventional cardiac catheter laboratory on the other. Nevertheless, it remains the standard of care in most centres. Recently, computational methods have been developed in which the laws of fluid dynamics can be applied to angiographic images to yield 'virtual' (computed) measures of blood flow, such as fractional flow reserve. Together with the CAG itself, this technology can provide an 'all-in-one' anatomical and functional investigation, which is particularly useful in the case of borderline lesions. It can add to the diagnostic value of CAG by providing increased precision and reduce the need for further non-invasive and functional tests of ischaemia, at minimal cost. In this paper, we place this technology in context, with emphasis on its potential to become established in the diagnostic workup of patients with suspected coronary artery disease, particularly in the non-interventional setting. We discuss the derivation and reliability of angiographically derived fractional flow reserve (CAG-FFR) as well as its limitations and how CAG-FFR could be integrated within existing national guidance. The assessment of coronary physiology may no longer be the preserve of the interventional cardiologist.

The relationship between coronary stenosis morphology and fractional flow reserve: a computational fluid dynamics modelling study.

Aims: International guidelines mandate the use of fractional flow reserve (FFR) and/or non-hyperaemic pressure ratios to assess the physiological significance of moderate coronary artery lesions to guide revascularization decisions. However, they remain underused such that visual estimation of lesion severity continues to be the predominant decision-making tool. It would be pragmatic to have an improved understanding of the relationship between lesion morphology and haemodynamics. The aim of this study was to compute virtual FFR (vFFR) in idealized coronary artery geometries with a variety of stenosis and vessel characteristics. Methods and results: Coronary artery geometries were modelled, based upon physiologically realistic branched arteries. Common stenosis characteristics were studied, including % narrowing, length, eccentricity, shape, number, position relative to branch, and distal (myocardial) resistance. Computational fluid dynamics modelling was used to calculate vFFRs using the VIRTUheart™ system. Percentage lesion severity had the greatest effect upon FFR. Any ≥80% diameter stenosis in two views (i.e. concentric) was physiologically significant (FFR ≤ 0.80), irrespective of length, shape, or vessel diameter. Almost all eccentric stenoses and all 50% concentric stenoses were physiologically non-significant, whilst 70% uniform concentric stenoses about 10 mm long straddled the ischaemic threshold (FFR 0.80). A low microvascular resistance (MVR) reduced FFR on average by 0.05, and a high MVR increased it by 0.03. Conclusion: Using computational modelling, we have produced an analysis of vFFR that relates stenosis characteristics to haemodynamic significance. The strongest predictor of a positive vFFR was a concentric, ≥80% diameter stenosis. The importance of MVR was quantified. Other lesion characteristics have a limited impact.