Guiding the Prostatic Artery Embolization Procedure With Computational Fluid Dynamics.

Benign prostatic hyperplasia (BPH) is a common disease associated with lower urinary tract symptoms and is the most frequent benign tumor in men. To reduce BPH therapy complications, prostatic artery embolization (PAE) was developed to replace the surgical options. PAE is a minimally invasive technique in which emboli are injected into the prostate arteries (PA), obstructing the blood flow in the hypervascular nodules. In this work, a personalized PAE treatment strategy was proposed using patient-specific computational fluid dynamics (CFD). First, the hemodynamics environment in the iliac arterial tree considering a large network of bifurcations was studied. The results showed complex blood flow patterns in the iliac arterial network. Subsequently, the transport of embolic particulates during PAE for the standard horizontal and hypothetical vertical patient positioning was simulated using Lagrangian particle tracking. Emboli of different sizes were released at various locations across the iliac arterial tree. The emboli entering the PA were mapped back to their initial location to create emboli release maps (ERMs). The obtained ERMs during the standard patient positioning for smaller emboli at certain release locations showed distinct regions in which if the emboli were released within these regions, all of them would reach the PA without nontarget embolization. During the hypothetical vertical patient positioning, the larger emboli formed a larger coherent region in the ERMs. Our patient-specific model can be used to find the best spatial location for emboli injection and perform the embolization procedure with minimal off-target delivery.

Machine Learning for Cardiovascular Biomechanics Modeling: Challenges and Beyond.

Recent progress in machine learning (ML), together with advanced computational power, have provided new research opportunities in cardiovascular modeling. While classifying patient outcomes and medical image segmentation with ML have already shown significant promising results, ML for the prediction of biomechanics such as blood flow or tissue dynamics is in its infancy. This perspective article discusses some of the challenges in using ML for replacing well-established physics-based models in cardiovascular biomechanics. Specifically, we discuss the large landscape of input features in 3D patient-specific modeling as well as the high-dimensional output space of field variables that vary in space and time. We argue that the end purpose of such ML models needs to be clearly defined and the tradeoff between the loss in accuracy and the gained speedup carefully interpreted in the context of translational modeling. We also discuss several exciting venues where ML could be strategically used to augment traditional physics-based modeling in cardiovascular biomechanics. In these applications, ML is not replacing physics-based modeling, but providing opportunities to solve ill-defined problems, improve measurement data quality, enable a solution to computationally expensive problems, and interpret complex spatiotemporal data by extracting hidden patterns. In summary, we suggest a strategic integration of ML in cardiovascular biomechanics modeling where the ML model is not the end goal but rather a tool to facilitate enhanced modeling.

Local and global growth and remodeling in calcific aortic valve disease and aging.

Aging and calcific aortic valve disease (CAVD) are the main factors leading to aortic stenosis. Both processes are accompanied by growth and remodeling pathways that play a crucial role in aortic valve pathophysiology. Herein, a computational growth and remodeling (G&R) framework was developed to investigate the effects of aging and calcification on aortic valve dynamics. Particularly, an algorithm was developed to couple the global growth and stiffening of the aortic valve due to aging and the local growth and stiffening due to calcification with the aortic valve transient dynamics. The aortic valve dynamics during baseline were validated with available data in the literature. Subsequently, the changes in aortic valve dynamic patterns during aging and CAVD progression were studied. The results revealed the patterns in geometric orifice area reduction and an increase in the valve stress during local and global growth and remodeling of the aortic valve. The proposed algorithm provides a framework to couple mechanobiology models of disease growth with tissue-scale transient structural mechanics models to study the biomechanical changes during cardiovascular disease growth and aging.

Integrating multi-fidelity blood flow data with reduced-order data assimilation.

High-fidelity patient-specific modeling of cardiovascular flows and hemodynamics is challenging. Direct blood flow measurement inside the body with in-vivo measurement modalities such as 4D flow magnetic resonance imaging (4D flow MRI) suffer from low resolution and acquisition noise. In-vitro experimental modeling and patient-specific computational fluid dynamics (CFD) models are subject to uncertainty in patient-specific boundary conditions and model parameters. Furthermore, collecting blood flow data in the near-wall region (e.g., wall shear stress) with experimental measurement modalities poses additional challenges. In this study, a computationally efficient data assimilation method called reduced-order modeling Kalman filter (ROM-KF) was proposed, which combined a sequential Kalman filter with reduced-order modeling using a linear model provided by dynamic mode decomposition (DMD). The goal of ROM-KF was to overcome low resolution and noise in experimental and uncertainty in CFD modeling of cardiovascular flows. The accuracy of the method was assessed with 1D Womersley flow, 2D idealized aneurysm, and 3D patient-specific cerebral aneurysm models. Synthetic experimental data were used to enable direct quantification of errors using benchmark datasets. The accuracy of ROM-KF in reconstructing near-wall hemodynamics was assessed by applying the method to problems where near-wall blood flow data were missing in the experimental dataset. The ROM-KF method provided blood flow data that were more accurate than the computational and synthetic experimental datasets and improved near-wall hemodynamics quantification.

Flow topology and targeted drug delivery in cardiovascular disease.

Targeted drug delivery is a promising technique to direct the drug to the specific diseased region. Nanoparticles have provided an attractive approach for this purpose. In practice, the major focus of targeted delivery has been on targeting cell receptors. However, the complex fluid mechanics in diseased biomedical flows questions if a sufficient number of nanoparticles can reach the desired region. In this paper, we propose that hidden topological structures in cardiovascular flows identified with Lagrangian coherent structures (LCS) control drug transport and provide valuable information for optimizing targeted drug delivery efficiency. We couple image-based computational fluid dynamics (CFD) with continuum transport models to study nanoparticle transport in coronary artery disease. We simulate nanoparticle transport as well as the recently proposed shear targeted drug delivery system that couples micro-carriers with nanoparticle drugs. The role of the LCS formed near the stenosed artery in controlling drug transport is discussed. Our results motivate the design of smart micro-needles guided by flow topology, which could achieve optimal drug delivery efficiency.

Data-driven cardiovascular flow modelling: examples and opportunities.

High-fidelity blood flow modelling is crucial for enhancing our understanding of cardiovascular disease. Despite significant advances in computational and experimental characterization of blood flow, the knowledge that we can acquire from such investigations remains limited by the presence of uncertainty in parameters, low resolution, and measurement noise. Additionally, extracting useful information from these datasets is challenging. Data-driven modelling techniques have the potential to overcome these challenges and transform cardiovascular flow modelling. Here, we review several data-driven modelling techniques, highlight the common ideas and principles that emerge across numerous such techniques, and provide illustrative examples of how they could be used in the context of cardiovascular fluid mechanics. In particular, we discuss principal component analysis (PCA), robust PCA, compressed sensing, the Kalman filter for data assimilation, low-rank data recovery, and several additional methods for reduced-order modelling of cardiovascular flows, including the dynamic mode decomposition and the sparse identification of nonlinear dynamics. All techniques are presented in the context of cardiovascular flows with simple examples. These data-driven modelling techniques have the potential to transform computational and experimental cardiovascular research, and we discuss challenges and opportunities in applying these techniques in the field, looking ultimately towards data-driven patient-specific blood flow modelling.

Fluid-structure coupled biotransport processes in aortic valve disease.

Biological transport processes near the aortic valve play a crucial role in calcific aortic valve disease initiation and bioprosthetic aortic valve thrombosis. Hemodynamics coupled with the dynamics of the leaflets regulate these transport patterns. Herein, two-way coupled fluid-structure interaction (FSI) simulations of a 2D bicuspid aortic valve and a 3D mechanical heart valve were performed and coupled with various convective mass transport models that represent some of the transport processes in calcification and thrombosis. Namely, five different continuum transport models were developed to study biochemicals that originate from the blood and the leaflets, as well as residence-time and flow stagnation. Low-density lipoprotein (LDL) and platelet activation were studied for their role in calcification and thrombosis, respectively. Coherent structures were identified using vorticity and Lagrangian coherent structures (LCS) for the 2D and 3D models, respectively. A very close connection between vortex structures and biochemical concentration patterns was shown where different vortices controlled the concentration patterns depending on the transport mechanism. Additionally, the relationship between leaflet concentration and wall shear stress was revealed. Our work shows that blood flow physics and coherent structures regulate the flow-mediated biological processes that are involved in aortic valve calcification and thrombosis, and therefore could be used in the design process to optimize heart valve replacement durability.

The Story of Wall Shear Stress in Coronary Artery Atherosclerosis: Biochemical Transport and Mechanotransduction.

Coronary artery atherosclerosis is a local, multifactorial, complex disease, and the leading cause of death in the US. Complex interactions between biochemical transport and biomechanical forces influence disease growth. Wall shear stress (WSS) affects coronary artery atherosclerosis by inducing endothelial cell mechanotransduction and by controlling the near-wall transport processes involved in atherosclerosis. Each of these processes is controlled by WSS differently and therefore has complicated the interpretation of WSS in atherosclerosis. In this paper, we present a comprehensive theory for WSS in atherosclerosis. First, a short review of shear stress-mediated mechanotransduction in atherosclerosis was presented. Next, subject-specific computational fluid dynamics (CFD) simulations were performed in ten coronary artery models of diseased and healthy subjects. Biochemical-specific mass transport models were developed to study low-density lipoprotein, nitric oxide, adenosine triphosphate, oxygen, monocyte chemoattractant protein-1, and monocyte transport. The transport results were compared with WSS vectors and WSS Lagrangian coherent structures (WSS LCS). High WSS magnitude protected against atherosclerosis by increasing the production or flux of atheroprotective biochemicals and decreasing the near-wall localization of atherogenic biochemicals. Low WSS magnitude promoted atherosclerosis by increasing atherogenic biochemical localization. Finally, the attracting WSS LCS's role was more complex where it promoted or prevented atherosclerosis based on different biochemicals. We present a summary of the different pathways by which WSS influences coronary artery atherosclerosis and compare different mechanotransduction and biotransport mechanisms.

Super-resolution and denoising of 4D-Flow MRI using physics-Informed deep neural nets.

BACKGROUND AND OBJECTIVE: Time resolved three-dimensional phase contrast magnetic resonance imaging (4D-Flow MRI) has been used to non-invasively measure blood velocities in the human vascular system. However, issues such as low spatio-temporal resolution, acquisition noise, velocity aliasing, and phase-offset artifacts have hampered its clinical application. In this research, we developed a purely data-driven method for super-resolution and denoising of 4D-Flow MRI. METHODS: The flow velocities, pressure, and the MRI image magnitude are modeled as a patient-specific deep neural net (DNN). For training, 4D-Flow MRI images in the complex Cartesian space are used to impose data-fidelity. Physics of fluid flow is imposed through regularization. Creative loss function terms have been introduced to handle noise and super-resolution. The trained patient-specific DNN can be sampled to generate noise-free high-resolution flow images. The proposed method has been implemented using the TensorFlow DNN library and tested on numerical phantoms and validated in-vitro using high-resolution particle image velocitmetry (PIV) and 4D-Flow MRI experiments on transparent models subjected to pulsatile flow conditions. RESULTS: In case of numerical phantoms, we were able to increase spatial resolution by a factor of 100 and temporal resolution by a factor of 5 compared to the simulated 4D-Flow MRI. There is an order of magnitude reduction of velocity normalized root mean square error (vNRMSE). In case of the in-vitro validation tests with PIV as reference, there is similar improvement in spatio-temporal resolution. Although the vNRMSE is reduced by 50%, the method is unable to negate a systematic bias with respect to the reference PIV that is introduced by the 4D-Flow MRI measurement. CONCLUSIONS: This work has demonstrated the feasibility of using the readily available machinery of deep learning to enhance 4D-Flow MRI using a purely data-driven method. Unlike current state-of-the-art methods, the proposed method is agnostic to geometry and boundary conditions and therefore eliminates the need for tedious tasks such as accurate image segmentation for geometry, image registration, and estimation of boundary flow conditions. Arbitrary regions of interest can be selected for processing. This work will lead to user-friendly analysis tools that will enable quantitative hemodynamic analysis of vascular diseases in a clinical setting.

Particle transport and deposition correlation with near-wall flow characteristic under inspiratory airflow in lung airways.

Exposure of lung airways to detrimental suspended aerosols in the environment increases the vulnerability of the respiratory and cardiovascular systems. In addition, recent developments in therapeutic inhalation devices magnify the importance of particle transport. In this manuscript, particle transport and deposition patterns in the upper tracheobronchial (TB) tree were studied where the inertial forces are considerable for microparticles. Wall shear stress divergence (WSSdiv) is proposed as a wall-based parameter that can predict particle deposition patterns. WSSdiv is proportional to near-wall normal velocity and can quantify the strength of flow towards and away from the wall. Computational fluid dynamics (CFD) simulations were performed to quantify airflow velocity and WSS vectors for steady inhalation in one case-control and unsteady inhalation in six subject-specific airway trees. Turbulent flow simulation was performed for the steady case using large eddy simulation to study the effect of turbulence. Magnetic resonance velocimetry (MRV) measurements were used to validate the case-control CFD simulation. Inertial particle transport was modeled by solving the Maxey-Riley equation in a Lagrangian framework. Deposition percentage (DP) was quantified for the case-control model over five particle sizes. DP was found to be proportional to particle size in agreement with previous studies in the literature. A normalized deposition concentration (DC) was defined to characterize localized deposition. A relatively strong correlation (Pearson value > 0.7) was found between DC and positive WSSdiv for physiologically relevant Stokes (St) numbers. Additionally, a regional analysis was performed after dividing the lungs into smaller areas. A spatial integral of positive WSSdiv over each division was shown to maintain a very strong correlation (Pearson value > 0.9) with cumulative spatial DC or regional dosimetry. The conclusions were generalized to a larger population in which two healthy and four asthmatic patients were investigated. This study shows that WSSdiv could be used to predict the qualitative surface deposition and relative regional dosimetry without the need to solve a particle transport problem.

Coronary artery plaque growth: A two-way coupled shear stress-driven model.

Atherosclerosis in coronary arteries can lead to plaque growth, stenosis formation, and blockage of the blood flow supplying the heart tissue. Several studies have shown that hemodynamics play an important role in the growth of coronary artery plaques. Specifically, low wall shear stress (WSS) appears to be the leading hemodynamic parameter promoting atherosclerotic plaque growth, which in turn influences the blood flow and WSS distribution. Therefore, a two-way coupled interaction exists between WSS and atherosclerosis growth. In this work, a computational framework was developed to study the coupling between WSS and plaque growth in coronary arteries. Computational fluid dynamics (CFD) was used to quantify WSS distribution. Surface mesh nodes were moved in the inward normal direction according to a growth model based on WSS. After each growth stage, the geometry was updated and the CFD simulation repeated to find updated WSS values for the next growth stage. One hundred twenty growth stages were simulated in an idealized tube and an image-based left anterior descending artery. An automated framework was developed using open-source software to couple CFD simulations with growth. Changes in plaque morphology and hemodynamic patterns during different growth stages are presented. The results show larger plaque growth towards the downstream segment of the plaque, agreeing with the reported clinical observations. The developed framework could be used to establish hemodynamic-driven growth models and study the interaction between these processes.

Topological analysis of particle transport in lung airways: Predicting particle source and destination.

Particle transport in lung airways can induce respiratory disease and play a vital role in aerosol drug delivery. Herein, we present dynamical systems features that influence airflow and particle transport in the tracheobronchial trees. Computational fluid dynamics (CFD) was used to solve for unsteady airflow in a patient-specific model. Particle tracking simulations were performed for micron-size particles. The destination map that connects the particle final location to the initial location and injection time was constructed. Finite-time Lyapunov exponent (FTLE) fields were calculated to identify inertial Lagrangian coherent structures (ILCS), topological features that act as separatrices. Our results demonstrated that these topological features control the destination map at the trachea. The temporal evolution of ILCS influenced the sensitivity of particle transport fate to injection time, whereas the emergence of new ILCS with an increased integration time controlled transport to different generations of airways. Additionally, particles starting at the ILCS were shown to mostly deposit at the airway walls. Finally, an innovative source inversion strategy was introduced to integrate the Maxey-Riley equation backward in time and identify the origin of dispersed particles. Our study explores novel dynamical systems tools that improve our understanding of particle transport and deposition in the airways and could be used to guide future targeted drug delivery studies.

A critical comparison of different residence time measures in aneurysms.

Flow stagnation and residence time (RT) are important features of diseased arterial flows that influence biochemical transport processes and thrombosis. RT calculation methods are classified into Eulerian and Lagrangian approaches where several measures have been proposed to quantify RT. Each of these methods has a different definition of RT, and it is not clear how they are related. In this study, image-based computational models of blood flow in an abdominal aortic aneurysm and a cerebral aneurysm were considered and RT was calculated using different methods. In the Lagrangian methods, discrete particle tracking of massless tracers was used to calculate particle residence time and mean exposure time. In the Eulerian methods, continuum transport models were used to quantify RT using Eulerian RT and virtual ink approaches. Point-wise RT and Eulerian indicator RT were also computed based on measures derived from velocity. A comparison of these methods is presented and the implications of each method are discussed. Our results highlight that most RT methods have a conceptually distinct definition of RT and therefore should be utilized depending on the specific application of interest.

Finite element modeling of near-wall mass transport in cardiovascular flows.

Many cardiovascular processes involve mass transport between blood and the vessel wall. Finite element methods are commonly used to numerically simulate these processes. Cardiovascular mass transport problems are typically characterized by high Péclet numbers, requiring fine near-wall mesh resolution as well as the use of stabilization techniques to avoid numerical instabilities. In this work, we develop a set of guidelines for solving high-Péclet-number near-wall mass transport problems using the finite element method. We use a steady, idealized test case to investigate the required mesh resolution and finite element basis order to accurately capture near-wall concentration boundary layers, as well as the performance of several commonly used stabilization techniques. Linear tetrahedral meshes were found to outperform quadratic tetrahedral meshes of equivalent degrees of freedom, and the commonly used discontinuity-capturing stabilization technique was found to be overly diffusive for these types of problems. Best practices derived from the idealized test case were then applied to a typical patient-specific vascular blood flow modeling application, where it was found that the commonly applied technique of avoiding numerical difficulties by artificially increasing mass diffusivity provides qualitatively similar but quantitatively erroneous results.

Accounting for residence-time in blood rheology models: do we really need non-Newtonian blood flow modelling in large arteries?

Patient-specific computational fluid dynamics (CFD) is a promising tool that provides highly resolved haemodynamics information. The choice of blood rheology is an assumption in CFD models that has been subject to extensive debate. Blood is known to exhibit shear-thinning behaviour, and non-Newtonian modelling has been recommended for aneurysmal flows. Current non-Newtonian models ignore rouleaux formation, which is the key player in blood's shear-thinning behaviour. Experimental data suggest that red blood cell aggregation and rouleaux formation require notable red blood cell residence-time (RT) in a low shear rate regime. This study proposes a novel hybrid Newtonian and non-Newtonian rheology model where the shear-thinning behaviour is activated in high RT regions based on experimental data. Image-based abdominal aortic and cerebral aneurysm models are considered and highly resolved CFD simulations are performed using a minimally dissipative solver. Lagrangian particle tracking is used to define a backward particle RT measure and detect stagnant regions with increased rouleaux formation likelihood. Our novel RT-based non-Newtonian model shows a significant reduction in shear-thinning effects and provides haemodynamic results qualitatively identical and quantitatively close to the Newtonian model. Our results have important implications in patient-specific CFD modelling and suggest that non-Newtonian models should be revisited in large artery flows.

Wall shear stress fixed points in cardiovascular fluid mechanics.

Complex blood flow in large arteries creates rich wall shear stress (WSS) vectorial features. WSS acts as a link between blood flow dynamics and the biology of various cardiovascular diseases. WSS has been of great interest in a wide range of studies and has been the most popular measure to correlate blood flow to cardiovascular disease. Recent studies have emphasized different vectorial features of WSS. However, fixed points in the WSS vector field have not received much attention. A WSS fixed point is a point on the vessel wall where the WSS vector vanishes. In this article, WSS fixed points are classified and the aspects by which they could influence cardiovascular disease are reviewed. First, the connection between WSS fixed points and the flow topology away from the vessel wall is discussed. Second, the potential role of time-averaged WSS fixed points in biochemical mass transport is demonstrated using the recent concept of Lagrangian WSS structures. Finally, simple measures are proposed to quantify the exposure of the endothelial cells to WSS fixed points. Examples from various arterial flow applications are demonstrated.

A strain-based finite element model for calcification progression in aortic valves.

Calcific aortic valve disease (CAVD) is a serious disease affecting the aging population. A complex interaction between biochemicals, cells, and mechanical cues affects CAVD initiation and progression. In this study, motivated by the progression of calcification in regions of high strain, we developed a finite element method (FEM) based spatial calcification progression model. Several cardiac cycles of transient structural FEM simulations were simulated. After each simulation cycle, calcium deposition was placed in regions of high circumferential strain. Our results show the radial expansion of calcification as spokes starting from the attachment region, agreeing very well with the reported clinical data.

Multiscale Systems Biology Model of Calcific Aortic Valve Disease Progression.

Calcific aortic valve disease is a common cause of aortic stenosis, a life threatening condition. In this study, a mathematical model is developed to simulate the cascade of mechanosensitive biochemical events that occur upon damage to the endothelial layer, leading to calcification. The model contains two phases. In the initiation phase, the model accounts for low-density lipoprotein (LDL) penetration into the subendothelial space, oxidation of LDL, and monocyte penetration and differentiation to activated macrophages. In the calcification phase, transforming growth factor beta is secreted from macrophages, inducing differentiation of valvular interstitial cells into activated myofibroblasts that can enable calcium deposition. Wall shear stress and mechanical strain are taken into account with simplified models updated based on calcification progression. The model parameters are estimated based on experimental data. Next, a statin therapy simulation is performed to evaluate the effect of lipid lowering therapy on calcification progression, demonstrating an age-dependent effectiveness in statin therapy. A new potential therapy targeting transforming growth factor-ß activation is proposed and simulated. The long-term evolution of calcification is compared to two sets of published longitudinal clinical data, showing promising agreement. The proposed model can provide clinically valuable data, potentially guiding surgeons in valve replacement decision makings.

Wall shear stress exposure time: a Lagrangian measure of near-wall stagnation and concentration in cardiovascular flows.

Near-wall transport is of utmost importance in connecting blood flow mechanics with cardiovascular disease progression. The near-wall region is the interface for biologic and pathophysiologic processes such as thrombosis and atherosclerosis. Most computational and experimental investigations of blood flow implicitly or explicitly seek to quantify hemodynamics at the vessel wall (or lumen surface), with wall shear stress (WSS) quantities being the most common descriptors. Most WSS measures are meant to quantify the frictional force of blood flow on the vessel lumen. However, WSS also provides an approximation to the near-wall blood flow velocity. We herein leverage this fact to compute a wall shear stress exposure time (WSSET) measure that is derived from Lagrangian processing of the WSS vector field. We compare WSSET against the more common relative residence time (RRT) measure, as well as a WSS divergence measure, in several applications where hemodynamics are known to be important to disease progression. Because these measures seek to quantify near-wall transport and because near-wall transport is important in several cardiovascular pathologies, surface concentration computed from a continuum transport model is used as a reference. The results show that compared to RRT, WSSET is able to better approximate the locations of near-wall stagnation and concentration build-up of chemical species, particularly in complex flows. For example, the correlation to surface concentration increased on average from 0.51 (RRT) to 0.79 (WSSET) in abdominal aortic aneurysm flow. Because WSSET considers integrated transport behavior, it can be more suitable in regions of complex hemodynamics that are traditionally difficult to quantify, yet encountered in many disease scenarios.