Computed numerical analysis of the biomechanical effects on coronary atherogenesis using human hemodynamic and dimensional variables

Machine Learning for Aiding Blood Flow Velocity Estimation Based on Angiography

Computational fluid dynamics (CFD) is widely employed to predict hemodynamic characteristics in arterial models, while not friendly to clinical applications due to the complexity of numerical simulations. Alternatively, this work proposed a framework to estimate hemodynamics in vessels based on angiography images using machine learning (ML) algorithms. First, the iodine contrast perfusion in blood was mimicked by a flow of dye diffusing into water in the experimentally validated CFD modeling. The generated projective images from simulations imitated the counterpart of light passing through the flow field as an analogy of X-ray imaging. Thus, the CFD simulation provides both the ground truth velocity field and projective images of dye flow patterns. The rough velocity field was estimated using the optical flow method (OFM) based on 53 projective images. ML training with least absolute shrinkage, selection operator and convolutional neural network was conducted with CFD velocity data as the ground truth and OFM velocity estimation as the input. The performance of each model was evaluated based on mean absolute error and mean squared error, where all models achieved or surpassed the criteria of 3 × 10−3 and 5 × 10−7 m/s, respectively, with a standard deviation less than 1 × 10−6 m/s. Finally, the interpretable regression and ML models were validated with over 613 image sets. The validation results showed that the employed ML model significantly reduced the error rate from 53.5% to 2.5% on average for the v-velocity estimation in comparison with CFD. The ML framework provided an alternative pathway to support clinical diagnosis by predicting hemodynamic information with high efficiency and accuracy.

Accuracy of computational pressure-fluid dynamics applied to coronary angiography to derive fractional flow reserve: FLASH FFR

Aims

Conventional fractional flow reserve (FFR) is measured invasively using a coronary guidewire equipped with a pressure sensor. A non-invasive derived FFR would eliminate risk of coronary injury, minimize technical limitations, and potentially increase adoption. We aimed to evaluate the diagnostic performance of a computational pressure-flow dynamics derived FFR (caFFR), applied to coronary angiography, compared to invasive FFR.

Methods and results

The FLASH FFR study was a prospective, multicentre, single-arm study conducted at six centres in China. Eligible patients had native coronary artery target lesions with visually estimated diameter stenosis of 30–90% and diagnosis of stable or unstable angina pectoris. Using computational pressure-fluid dynamics, in conjunction with thrombolysis in myocardial infarction (TIMI) frame count, applied to coronary angiography, caFFR was measured online in real-time and compared blind to conventional invasive FFR by an independent core laboratory. The primary endpoint was the agreement between caFFR and FFR, with a pre-specified performance goal of 84%. Between June and December 2018, matched caFFR and FFR measurements were performed in 328 coronary arteries. Total operational time for caFFR was 4.54 ± 1.48 min. caFFR was highly correlated to FFR (R = 0.89, P = 0.76) with a mean bias of −0.002 ± 0.049 (95% limits of agreement −0.098 to 0.093). The diagnostic performance of caFFR vs. FFR was diagnostic accuracy 95.7%, sensitivity 90.4%, specificity 98.6%, positive predictive value 97.2%, negative predictive value 95.0%, and area under the receiver operating characteristic curve of 0.979.

Conclusions

Using wire-based FFR as the reference, caFFR has high accuracy, sensitivity, and specificity. caFFR could eliminate the need of a pressure wire, technical error and potentially increase adoption of physiological assessment of coronary artery stenosis severity.

Clinical Trial Registration

URL: http://www.chictr.org.cn Unique Identifier: ChiCTR1800019522.

Computational fluid dynamics in coronary artery disease

Computational fluid dynamics (CFD) is a widely used method in mechanical engineering to solve complex problems by analysing fluid flow, heat transfer, and associated phenomena by using computer simulations. In recent years, CFD has been increasingly used in biomedical research of coronary artery disease because of its high performance hardware and software. CFD techniques have been applied to study cardiovascular haemodynamics through simulation tools to predict the behaviour of circulatory blood flow in the human body. CFD simulation based on 3D luminal reconstructions can be used to analyse the local flow fields and flow profiling due to changes of coronary artery geometry, thus, identifying risk factors for development and progression of coronary artery disease. This review aims to provide an overview of the CFD applications in coronary artery disease, including biomechanics of atherosclerotic plaques, plaque progression and rupture; regional haemodynamics relative to plaque location and composition. A critical appraisal is given to a more recently developed application, fractional flow reserve based on CFD computation with regard to its diagnostic accuracy in the detection of haemodynamically significant coronary artery disease.

Computational fluid dynamics in cardiovascular disease

Computational fluid dynamics (CFD) is a mechanical engineering field for analyzing fluid flow, heat transfer, and associated phenomena, using computer-based simulation. CFD is a widely adopted methodology for solving complex problems in many modern engineering fields. The merit of CFD is developing new and improved devices and system designs, and optimization is conducted on existing equipment through computational simulations, resulting in enhanced efficiency and lower operating costs. However, in the biomedical field, CFD is still emerging. The main reason why CFD in the biomedical field has lagged behind is the tremendous complexity of human body fluid behavior. Recently, CFD biomedical research is more accessible, because high performance hardware and software are easily available with advances in computer science. All CFD processes contain three main components to provide useful information, such as pre-processing, solving mathematical equations, and post-processing. Initial accurate geometric modeling and boundary conditions are essential to achieve adequate results. Medical imaging, such as ultrasound imaging, computed tomography, and magnetic resonance imaging can be used for modeling, and Doppler ultrasound, pressure wire, and non-invasive pressure measurements are used for flow velocity and pressure as a boundary condition. Many simulations and clinical results have been used to study congenital heart disease, heart failure, ventricle function, aortic disease, and carotid and intra-cranial cerebrovascular diseases. With decreasing hardware costs and rapid computing times, researchers and medical scientists may increasingly use this reliable CFD tool to deliver accurate results. A realistic, multidisciplinary approach is essential to accomplish these tasks. Indefinite collaborations between mechanical engineers and clinical and medical scientists are essential. CFD may be an important methodology to understand the pathophysiology of the development and progression of disease and for establishing and creating treatment modalities in the cardiovascular field.

Computational study of pulsatile blood flow in prototype vessel geometries of coronary segments

The spatial and temporal distributions of wall shear stress (WSS) in prototype vessel geometries of coronary segments are investigated via numerical simulation, and the potential association with vascular disease and specifically atherosclerosis and plaque rupture is discussed. In particular, simulation results of WSS spatio-temporal distributions are presented for pulsatile, non-Newtonian blood flow conditions for: (a) curved pipes with different curvatures, and (b) bifurcating pipes with different branching angles and flow division. The effects of non-Newtonian flow on WSS (compared to Newtonian flow) are found to be small at Reynolds numbers representative of blood flow in coronary arteries. Specific preferential sites of average low WSS (and likely atherogenesis) were found at the outer regions of the bifurcating branches just after the bifurcation, and at the outer-entry and inner-exit flow regions of the curved vessel segment. The drop in WSS was more dramatic at the bifurcating vessel sites (less than 5% of the pre-bifurcation value). These sites were also near rapid gradients of WSS changes in space and time - a fact that increases the risk of rupture of plaque likely to develop at these sites. The time variation of the WSS spatial distributions was very rapid around the start and end of the systolic phase of the cardiac cycle, when strong fluctuations of intravascular pressure were also observed. These rapid and strong changes of WSS and pressure coincide temporally with the greatest flexion and mechanical stresses induced in the vessel wall by myocardial motion (ventricular contraction). The combination of these factors may increase the risk of plaque rupture and thrombus formation at these sites.

Flow studies in three-dimensional aorto-right coronary bypass graft system

This paper presents the fluid dynamics of blood flow in a coronary bypass model of the aorto-right coronary bypass system. Three-dimensional computational fluid dynamic simulations are developed of the blood flow in coronary artery-bypass systems, using the computational fluid dynamics software (FLUENT 6.0.1). These blood flow simulations are performed within small intervals of the cardiac cycle, using input data consisting of physiological measurements of flow rates in the aorta, obtained from earlier studies. We have calculated the flow-field distributions of the velocity and the wall shear stress at four typical instants of the cardiac cycle, two during systole and two during the diastole phase. Plots of velocity vector and the wall shear stress are displayed in the aorto-graft-coronary arterial flow-field domain, providing an insight into the link between fluid dynamics and arterial diseases. The prime regions of disturbed flow patterns are at the entrance into the graft from the aorta and at the exit from the graft into the right coronary artery. Our objective is to obtain an understanding of how the coronary artery is perfused by the graft, and thereby into the factors affecting graft patency.

Effects of the shape of coronary arteries on the presence, extent, and severity of their disease

It is now known that a C-shaped right coronary artery (RCA) (C-RCA) is associated with more atherosclerotic disease than a sigma-shaped RCA (S-RCA). This study was designed to investigate the relationship between the shape of the RCA and the presence, extent, and severity of coronary artery disease (CAD) in all coronary arteries. Two hundred and forty-five patients were divided into C-RCA (group 1, n = 161) and S-RCA (group 2, n = 84) groups. The vessel score, extent score, and Gensini score were higher in group 1 than in group 2 (P < 0.001 for all comparisons). In multivariate analyses, age (P = 0.001), male sex (P < 0.001), smoking (P = 0.002), and C-RCA (P < 0.001) were independent predictors of significant CAD (presence of at least one lesion causing at least >50% stenosis). Multiple linear regression analysis to predict vessel score identified C-RCA (P < 0.001), age (P = 0.004), and male sex (P = 0.020), to predict extent score identified C-RCA (P < 0.001), age (P = 0.005), and male sex (P = 0.003), and to predict Gensini score identified C-RCA (P < 0.001), male sex (P = 0.009), and dyslipidemia (P = 0.049) as independently associated variables. Sensitivity and specificity of C-RCA for detecting significant CAD were 79% and 46%, respectively. This study showed that C-RCA was an independent predictor of significant CAD, which was independently associated with vessel score, extent score, and Gensini score. However, it was not clinically useful, because it was not specific or sensitive enough to identify patients with significant CAD.

Computational model of blood flow in the aorto-coronary bypass graft

Background

Coronary artery bypass grafting surgery is an effective treatment modality for patients with severe coronary artery disease. The conduits used during the surgery include both the arterial and venous conduits. Long- term graft patency rate for the internal mammary arterial graft is superior, but the same is not true for the saphenous vein grafts. At 10 years, more than 50% of the vein grafts would have occluded and many of them are diseased. Why do the saphenous vein grafts fail the test of time? Many causes have been proposed for saphenous graft failure. Some are non-modifiable and the rest are modifiable. Non-modifiable causes include different histological structure of the vein compared to artery, size disparity between coronary artery and saphenous vein. However, researches are more interested in the modifiable causes, such as graft flow dynamics and wall shear stress distribution at the anastomotic sites. Formation of intimal hyperplasia at the anastomotic junction has been implicated as the root cause of long- term graft failure.

Many researchers have analyzed the complex flow patterns in the distal sapheno-coronary anastomotic region, using various simulated model in an attempt to explain the site of preferential intimal hyperplasia based on the flow disturbances and differential wall stress distribution. In this paper, the geometrical bypass models (aorto-left coronary bypass graft model and aorto-right coronary bypass graft model) are based on real-life situations. In our models, the dimensions of the aorta, saphenous vein and the coronary artery simulate the actual dimensions at surgery. Both the proximal and distal anastomoses are considered at the same time, and we also take into the consideration the cross-sectional shape change of the venous conduit from circular to elliptical. Contrary to previous works, we have carried out computational fluid dynamics (CFD) study in the entire aorta-graft-perfused artery domain. The results reported here focus on (i) the complex flow patterns both at the proximal and distal anastomotic sites, and (ii) the wall shear stress distribution, which is an important factor that contributes to graft patency.

Methods

The three-dimensional coronary bypass models of the aorto-right coronary bypass and the aorto-left coronary bypass systems are constructed using computational fluid-dynamics software (Fluent 6.0.1). To have a better understanding of the flow dynamics at specific time instants of the cardiac cycle, quasi-steady flow simulations are performed, using a finite-volume approach. The data input to the models are the physiological measurements of flow-rates at (i) the aortic entrance, (ii) the ascending aorta, (iii) the left coronary artery, and (iv) the right coronary artery.

Results

The flow field and the wall shear stress are calculated throughout the cycle, but reported in this paper at two different instants of the cardiac cycle, one at the onset of ejection and the other during mid-diastole for both the right and left aorto-coronary bypass graft models. Plots of velocity-vector and the wall shear stress distributions are displayed in the aorto-graft-coronary arterial flow-field domain. We have shown (i) how the blocked coronary artery is being perfused in systole and diastole, (ii) the flow patterns at the two anastomotic junctions, proximal and distal anastomotic sites, and (iii) the shear stress distributions and their associations with arterial disease.

Conclusion

The computed results have revealed that (i) maximum perfusion of the occluded artery occurs during mid-diastole, and (ii) the maximum wall shear-stress variation is observed around the distal anastomotic region. These results can enable the clinicians to have a better understanding of vein graft disease, and hopefully we can offer a solution to alleviate or delay the occurrence of vein graft disease.