Image-based Computational Models Research Articles

Image-based flow simulation of platelet aggregates under different shear rates.

Hemodynamics is crucial for the activation and aggregation of platelets in response to flow-induced shear. In this paper, a novel image-based computational model simulating blood flow through and around platelet aggregates is presented. The microstructure of aggregates was captured by two different modalities of microscopy images of in vitro whole blood perfusion experiments in microfluidic chambers coated with collagen. One set of images captured the geometry of the aggregate outline, while the other employed platelet labelling to infer the internal density. The platelet aggregates were modelled as a porous medium, the permeability of which was calculated with the Kozeny-Carman equation. The computational model was subsequently applied to study hemodynamics inside and around the platelet aggregates. The blood flow velocity, shear stress and kinetic force exerted on the aggregates were investigated and compared under 800 s-1, 1600 s-1 and 4000 s-1 wall shear rates. The advection-diffusion balance of agonist transport inside the platelet aggregates was also evaluated by local Péclet number. The findings show that the transport of agonists is not only affected by the shear rate but also significantly influenced by the microstructure of the aggregates. Moreover, large kinetic forces were found at the transition zone from shell to core of the aggregates, which could contribute to identifying the boundary between the shell and the core. The shear rate and the rate of elongation flow were investigated as well. The results imply that the emerging shapes of aggregates are highly correlated to the shear rate and the rate of elongation. The framework provides a way to incorporate the internal microstructure of the aggregates into the computational model and yields a better understanding of the hemodynamics and physiology of platelet aggregates, hence laying the foundation for predicting aggregation and deformation under different flow conditions.

Visual working memory in immersive visualization: a change detection experiment and an image-computable model

Visual working memory (VWM) is a cognitive mechanism essential for interacting with the environment and accomplishing ongoing tasks, as it allows fast processing of visual inputs at the expense of the amount of information that can be stored. A better understanding of its functioning would be beneficial to research fields such as simulation and training in immersive Virtual Reality or information visualization and computer graphics. The current work focuses on the design and implementation of a paradigm for evaluating VWM in immersive visualization and of a novel image-based computational model for mimicking the human behavioral data of VWM. We evaluated the VWM at the variation of four conditions: set size, spatial layout, visual angle (VA) subtending stimuli presentation space, and observation time. We adopted a full factorial design and analysed participants’ performances in the change detection experiment. The analysis of hit rates and false alarm rates confirms the existence of a limit of VWM capacity of around 7 ± 2 items, as found in the literature based on the use of 2D videos and images. Only VA and observation time influence performances (p<0.0001). Indeed, with VA enlargement, participants need more time to have a complete overview of the presented stimuli. Moreover, we show that our model has a high level of agreement with the human data, r>0.88 (p<0.05).

The effect of scar imaging quality on in-silico pace-mapping of post-infarction ventricular tachycardia

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): Wellcome Trust and British Heart Foundation. Background In-silico pacing mapping is a technique which involves using an anatomically-personalised image-based computational model to generate a ‘virtual’ clinical pace map for enhanced pre-procedural ablation planning of post-infarction ventricular tachycardia (VT). Utilising, as it does, a direct clinical ECG (or implanted device) recording of the clinical VT in order to construct the map, it has a number of advantages over other computational methods which attempt to simulate patient-specific VTs to identify ablation targets. However, accurate model construction relies on adequate scar imaging data, which may not be available with current clinical cardiac MR (CMR) sequences. Purpose To investigate the impact of scar imaging quality on the ability of in-silico pace-mapping to determine the VT origin. Methods VTs were simulated in three post-infarction porcine heart models reconstructed from high resolution (1mm isotropic) contrast-enhanced CMR data. Visual analysis of the VT circuits was performed to identify the VT site of origin (exit site) in each case (as the gold standard). The ECG of each simulated VT was computed and used as input for the construction of the in-silico virtual pace-mapping approach. Prior to pace map construction, the scar anatomy was modified to represent three different levels of imaging resolution: high resolution scar mapping (default); low resolution scar (representing standing clinical CMR sequences), obtained by performing dilation and erosion in the scar to close all inner pathways; and, finally, no scar (cardiac anatomy only, representing CT-based reconstructions). In-silico pace-mapping was performed by pacing the heart in N = 1000 sites from randomly chosen locations, on both surfaces and intramurally, within close proximity (&amp;lt;5mm) from the infarct. ECG signals computed following each paced activation wavefront were compared to the VT ECG through correlation analysis to construct the virtual pace map. The distance d between the VT exit site and the pacing location with the strongest correlation was also computed. Results As can be seen in Figure 1, correlation between VT and paced ECGs was found to be strongest at pacing locations near the VT exit site in all models. Correlation coefficients were similar in value and distribution in models with high- (min: -0.49±0.27 and max: 0.92±0.05) and low-resolution scars (min: -0.46±0.29 and max: 0.93±0.05). While the correlation map in the absence of scar had smaller values when compared to the other two models (min: -0.50±0.24 and max: 0.71±0.09), the VT exit was still located near the pacing location with the strongest correlation: d = 9.47±2.83 (high-res), d = 6.63±6.35mm (low-res) and d = 18.60±7.10mm (no scar). Conclusion In-silico pace-mapping successfully detected the VT exit sites regardless the resolution of the scar. This finding suggest that the level of infarct anatomical detail has a minor effect on simulated paced activation to localize VT ablation.

Evaluation of a deep learning-enabled automated computational heart modelling workflow for personalized assessment of ventricular arrhythmias.

Personalized, image-based computational heart modelling is a powerful technology that can be used to improve patient-specific arrhythmia risk stratification and ventricular tachycardia (VT) ablation targeting. However, most state-of-the-art methods still require manual interactions by expert users. The goal of this study is to evaluate the feasibility of an automated, deep learning-based workflow for reconstructing personalized computational electrophysiological heart models to guide patient-specific treatment of VT. Contrast-enhanced computed tomography (CE-CT) images with expert ventricular myocardium segmentations were acquired from 111 patients across five cohorts from three different institutions. A deep convolutional neural network (CNN) for segmenting left ventricular myocardium from CE-CT was developed, trained and evaluated. From both CNN-based and expert segmentations in a subset of patients, personalized electrophysiological heart models were reconstructed and rapid pacing was used to induce VTs. CNN-based and expert segmentations were more concordant in the middle myocardium than in the heart's base or apex. Wavefront propagation during pacing was similar between CNN-based and original heart models. Between most sets of heart models, VT inducibility was the same, the number of induced VTs was strongly correlated, and VT circuits co-localized. Our results demonstrate that personalized computational heart models reconstructed from deep learning-based segmentations even with a small training set size can predict similar VT inducibility and circuit locations as those from expertly-derived heart models. Hence, a user-independent, automated framework for simulating arrhythmias in personalized heart models could feasibly be used in clinical settings to aid VT risk stratification and guide VT ablation therapy. KEY POINTS: Personalized electrophysiological heart modelling can aid in patient-specific ventricular tachycardia (VT) risk stratification and VT ablation targeting. Current state-of-the-art, image-based heart models for VT prediction require expert-dependent, manual interactions that may not be accessible across clinical settings. In this study, we develop an automated, deep learning-based workflow for reconstructing personalized heart models capable of simulating arrhythmias and compare its predictions with that of expert-generated heart models. The number and location of VTs was similar between heart models generated from the deep learning-based workflow and expert-generated heart models. These results demonstrate the feasibility of using an automated computational heart modelling workflow to aid in VT therapeutics and has implications for generalizing personalized computational heart technology to a broad range of clinical centres.

Non-invasive estimation of the parameters of a three-element windkessel model of aortic arch arteries in patients undergoing thoracic endovascular aortic repair.

Background: Image-based computational hemodynamic modeling and simulations are important for personalized diagnosis and treatment of cardiovascular diseases. However, the required patient-specific boundary conditions are often not available and need to be estimated. Methods: We propose a pipeline for estimating the parameters of the popular three-element Windkessel (WK3) models (a proximal resistor in series with a parallel combination of a distal resistor and a capacitor) of the aortic arch arteries in patients receiving thoracic endovascular aortic repair of aneurysms. Pre-operative and post-operative 1-week duplex ultrasound scans were performed to obtain blood flow rates, and intra-operative pressure measurements were also performed invasively using a pressure transducer pre- and post-stent graft deployment in arch arteries. The patient-specific WK3 model parameters were derived from the flow rate and pressure waveforms using an optimization algorithm reducing the error between simulated and measured pressure data. The resistors were normalized by total resistance, and the capacitor was normalized by total resistance and heart rate. The normalized WK3 parameters can be combined with readily available vessel diameter, brachial blood pressure, and heart rate data to estimate WK3 parameters of other patients non-invasively. Results: Ten patients were studied. The medians (interquartile range) of the normalized proximal resistor, distal resistor, and capacitor parameters are 0.10 (0.07-0.15), 0.90 (0.84-0.93), and 0.46 (0.33-0.58), respectively, for common carotid artery; 0.03 (0.02-0.04), 0.97 (0.96-0.98), and 1.91 (1.63-2.26) for subclavian artery; 0.18 (0.08-0.41), 0.82 (0.59-0.92), and 0.47 (0.32-0.85) for vertebral artery. The estimated pressure showed fairly high tolerance to patient-specific inlet flow rate waveforms using the WK3 parameters estimated from the medians of the normalized parameters. Conclusion: When patient-specific outflow boundary conditions are not available, our proposed pipeline can be used to estimate the WK3 parameters of arch arteries.

Image-based computer modeling assessment of microwave ablation for treatment of adrenal tumors

Purpose To assess the feasibility of delivering microwave ablation for targeted treatment of aldosterone producing adenomas using image-based computational models. Methods We curated an anonymized dataset of diagnostic 11C-metomidate PET/CT images of 14 patients with aldosterone producing adenomas (APA). A semi-automated approach was developed to segment the APA, adrenal gland, and adjacent organs within 2 cm of the APA boundary. The segmented volumes were used to implement patient-specific 3D electromagnetic-bioheat transfer models of microwave ablation with a 2.45 GHz directional microwave ablation applicator. Ablation profiles were quantitatively assessed based on the extent of the APA target encompassed by an ablative thermal dose, while limiting thermal damage to the adjacent normal adrenal tissue and sensitive critical structures. Results Across the 14 patients, adrenal tumor volumes ranged between 393 mm3 and 2,395 mm3. On average, 70% of the adrenal tumor volumes received an ablative thermal dose of 240CEM43, while limiting thermal damage to non-target structures, and thermally sparing 83.5–96.4% of normal adrenal gland. Average ablation duration was 293 s (range: 60–600 s). Simulations indicated coverage of the APA with an ablative dose was limited when the axis of the ablation applicator was not well aligned with the major axis of the targeted APA. Conclusions Image-based computational models demonstrate the potential for delivering microwave ablation to APA targets within the adrenal gland, while limiting thermal damage to surrounding non-target structures.

Non-invasive delineation of ventricular tachycardia substrates for cardiac stereotactic body radiotherapy: utility of in-silico pace-mapping

Abstract Funding Acknowledgements Type of funding sources: Public Institution(s). Main funding source(s): EPSRC Background Cardiac stereotactive body radiotherapy (CSBRT) is an emerging, non-invasive ablation modality that targets ventricular tachycardia (VT) substrates in patients with limited conventional treatment options. Success of CSBRT hinges primarily on the correct identification of VT targets, which requires non-invasive planning. Current non-invasive, pre-procedure strategies employ multi-electrode electrocardiographic imaging (ECGi). Given its significant cost and potential challenges in detecting endocardial, intramural and/or septal VT sites, there is a need to optimise VT delineation strategies for CSBRT; patient-specific simulations may show promise at guiding such planning non-invasively. Purpose We aim to perform non-invasive, in-silico pace-mapping on an image-based computational model to identify VT substrates for CSBRT. We intend to show the utility of our fast computational pipeline - relying on CT imaging data only - to provide further insights on inaccessible, scar-related VT episodes. Methods A detailed computational torso model of a CSBRT candidate with incessant VT was generated from CT imaging data. Extracellular content volumes (ECVs) were used to identify different tissue types (healthy, border zone and non-conducting), and scale model tissue conductivities accordingly. In-silico pace-mapping was performed by simulating ~360 paced beats across the LV, and computing corresponding 12-lead ECGs within a fast electrophysiological (EP) simulation environment combining reaction-eikonal and lead field methods. QRS complexes from simulated paced beats were used to construct the virtual correlation pace-map against the measured QRS of the clinically-induced VT, along with a ‘reference-less’ virtual pace-map constructed from neighbouring paced-beat QRSs (within a 20 mm radius). An epicardial activation map of the clinically-induced VT was reconstructed from ECGi measurement, and used for comparison against our virtual pace-maps. Results Correlations between simulated paced-beat QRS complexes and the clinically-induced VT QRS were higher in mid-apical, infero-septal segments - segment 9 (85.71%), 10 (87.95%) and 15 (89.58%) - identifying septal origin and pathway of the induced re-entrant circuit. A possible septal VT isthmus was also identified by a high gradient in the virtual reference-less pace-map in segment 9 (&amp;gt; 2.5%/mm). Our in-silico predictions were in agreement with the clinical regions identified for CSBRT (segment 9 and 15), and provided additional information on the 3D and septal dynamics of the VT episode. Conclusions Our in-silico pace-mapping study successfully localised VT substrates in a patient unable to receive standard ablative procedures, and provided further clinical insight on the induced VT dynamics. Our rapid in-silico pace-mapping approach may be utilised to support optimal identification of VT target volumes for CSBRT.

Robust and efficient fixed-point algorithm for the inverse elastostatic problem to identify myocardial passive material parameters and the unloaded reference configuration

Image-based computational models of the heart represent a powerful tool to shed new light on the mechanisms underlying physiological and pathological conditions in cardiac function and to improve diagnosis and therapy planning. However, in order to enable the clinical translation of such models, it is crucial to develop personalized models that are able to reproduce the physiological reality of a given patient. There have been numerous contributions in experimental and computational biomechanics to characterize the passive behavior of the myocardium. However, most of these studies suffer from severe limitations and are not applicable to high-resolution geometries. In this work, we present a novel methodology to perform an automated identification of in vivo properties of passive cardiac biomechanics. The highly-efficient algorithm fits material parameters against the shape of a patient-specific approximation of the end-diastolic pressure-volume relation (EDPVR). Simultaneously, an unloaded reference configuration is generated, where a novel line search strategy to improve convergence and robustness is implemented. Only clinical image data or previously generated meshes at one time point during diastole and one measured data point of the EDPVR are required as an input. The proposed method can be straightforwardly coupled to existing finite element (FE) software packages and is applicable to different constitutive laws and FE formulations. Sensitivity analysis demonstrates that the algorithm is robust with respect to initial input parameters.

Identification of Small, Regularly Shaped Cerebral Aneurysms Prone to Rupture.

Many small, regularly shaped cerebral aneurysms rupture; however, they usually receive a low score based on current risk-assessment methods. Our goal was to identify patient and aneurysm characteristics associated with rupture of small, regularly shaped aneurysms and to develop and validate predictive models of rupture in this aneurysm subpopulation. Cross-sectional data from 1079 aneurysms smaller than 7 mm with regular shapes (without blebs) were used to train predictive models for aneurysm rupture using machine learning methods. These models were based on the patient population, aneurysm location, and hemodynamic and geometric characteristics derived from image-based computational fluid dynamics models. An independent data set with 102 small, regularly shaped aneurysms was used for validation. Adverse hemodynamic environments characterized by strong, concentrated inflow jets, high speed, complex and unstable flow patterns, and concentrated, oscillatory, and heterogeneous wall shear stress patterns were associated with rupture in small, regularly shaped aneurysms. Additionally, ruptured aneurysms were larger and more elongated than unruptured aneurysms in this subset. A total of 5 hemodynamic and 6 geometric parameters along with aneurysm location, multiplicity, and morphology, were used as predictive variables. The best machine learning rupture prediction-model achieved a good performance with an area under the curve of 0.84 on the external validation data set. This study demonstrated the potential of using predictive machine learning models based on aneurysm-specific hemodynamic, geometric, and anatomic characteristics for identifying small, regularly shaped aneurysms prone to rupture.

Image-based biomechanical modeling for coronary atherosclerotic plaque progression and vulnerability prediction

Atherosclerotic plaque progression and rupture play an important role in cardiovascular disease development and the final drastic events such as heart attack and stroke. Medical imaging and image-based computational modeling methods advanced considerably in recent years to quantify plaque morphology and biomechanical conditions and gain a better understanding of plaque evolution and rupture process. This article first briefly reviewed clinical imaging techniques for coronary thin-cap fibroatheroma (TCFA) plaques used in image-based computational modeling. This was followed by a summary of different types of biomechanical models for coronary plaques. Plaque progression and vulnerability prediction studies based on image-based computational modeling were reviewed and compared. Much progress has been made and a reasonable high prediction accuracy has been achieved. However, there are still some inconsistencies in existing literature on the impact of biomechanical and morphological factors on future plaque behavior, and it is very difficult to perform direct comparison analysis as differences like image modality, biomechanical factors selection, predictive models, and progression/vulnerability measures exist among these studies. Encouraging data and model sharing across the research community would partially resolve these differences, and possibly lead to clearer assertive conclusions. In vivo image-based computational modeling could be used as a powerful tool for quantitative assessment of coronary plaque vulnerability for potential clinical applications.

Challenges in Modeling Hemodynamics in Cerebral Aneurysms Related to Arteriovenous Malformations.

The significantly higher incidence of aneurysms in patients with arteriovenous malformations (AVMs) suggests a strong hemodynamic relationship between these lesions. The presence of an AVM alters hemodynamics in proximal vessels by drastically changing the distal resistance, thus affecting intra-aneurysmal flow. This study discusses the challenges associated with patient-specific modeling of aneurysms in the presence of AVMs. We explore how the presence of a generic distal AVM affects upstream aneurysms by examining the relationship between distal resistance and aneurysmal wall shear stress using physiologically realistic estimates for the influence of the AVM on hemodynamics. Using image-based computational models of aneurysms and surrounding vasculature, aneurysmal wall-shear stress is calculated for a range of distal resistances corresponding to the presence of AVMs of various sizes and compared with a control case representing the absence of an AVM. In the patient cases considered, the alteration in aneurysmal wall shear stress due to the presence of an AVM is considerable, as much as 19 times the base case wall shear stress. Furthermore, the relationship between aneurysmal wall shear stress and distal resistance is shown to be highly geometry-dependent and nonlinear. In most cases, the range of physiologically realistic possibilities for AVM-related distal resistance are so large that patient-specific flow measurements are necessary for meaningful predictions of wall shear stress. The presented work offers insight on the impact of distal AVMs on aneurysmal wall shear stress using physiologically realistic computational models. Patient-specific modeling of hemodynamics in aneurysms and associated AVMs has great potential for understanding lesion pathogenesis, surgical planning, and assessing the effect of treatment of one lesion relative to another. However, we show that modeling approaches cannot usually meaningfully quantify the impact of AVMs if based solely on imaging data from CT and X-ray angiography, currently used in clinical practice. Based on recent studies, it appears that 4D flow MRI is one promising approach to obtaining meaningful patient-specific flow boundary conditions that improve modeling fidelity.

Impaired pulmonary ventilation beyond pneumonia in COVID-19: A preliminary observation.

Coronavirus disease 2019 (COVID-19) may severely impair pulmonary function and cause hypoxia. However, the association of COVID-19 pneumonia on CT with impaired ventilation remains unexplained. This pilot study aims to demonstrate the relationship between the radiological findings on COVID-19 CT images and ventilation abnormalities simulated in a computational model linked to the patients' symptoms. Twenty-five patients with COVID-19 and four test-negative healthy controls who underwent a baseline non-enhanced CT scan: 7 dyspneic patients, 9 symptomatic patients without dyspnea, and 9 asymptomatic patients were included. A 2D U-Net-based CT segmentation software was used to quantify radiological futures of COVID-19 pneumonia. The CT image-based full-scale airway network (FAN) flow model was employed to assess regional lung ventilation. Functional and radiological features were compared across groups and correlated with the clinical symptoms. Heterogeneity in ventilation distribution and ventilation defects associated with the pneumonia and the patients' symptoms were assessed. Median percentage ventilation defects were 0.2% for healthy controls, 0.7% for asymptomatic patients, 1.2% for symptomatic patients without dyspnea, and 11.3% for dyspneic patients. The median of percentage pneumonia was 13.2% for dyspneic patients and 0% for the other groups. Ventilation defects preferentially affected the posterior lung and worsened with increasing pneumonia linearly (y = 0.91x + 0.99, R2 = 0.73) except for one of the nine dyspneic patients who had disproportionally large ventilation defects (7.8% of the entire lung) despite mild pneumonia (1.2%). The symptomatic and dyspneic patients showed significantly right-skewed ventilation distributions (symptomatic without dyspnea: 0.86 ± 0.61, dyspnea 0.91 ± 0.79) compared to the patients without symptom (0.45 ± 0.35). The ventilation defect analysis with the FAN model provided a comparable diagnostic accuracy to the percentage pneumonia in identifying dyspneic patients (area under the receiver operating characteristic curve, 0.94 versus 0.96). COVID-19 pneumonia segmentations from CT scans are accompanied by impaired pulmonary ventilation preferentially in dyspneic patients. Ventilation analysis with CT image-based computational modelling shows it is able to assess functional impairment in COVID-19 and potentially identify one of the aetiologies of hypoxia in patients with COVID-19 pneumonia.

Quantifying Patient-Specific in vivo Coronary Plaque Material Properties for Accurate Stress/Strain Calculations: An IVUS-Based Multi-Patient Study.

Introduction: Mechanical forces are closely associated with plaque progression and rupture. Precise quantifications of biomechanical conditions using in vivo image-based computational models depend heavily on the accurate estimation of patient-specific plaque mechanical properties. Currently, mechanical experiments are commonly performed on ex vivo cardiovascular tissues to determine plaque material properties. Patient-specific in vivo coronary material properties are scarce in the existing literature.Methods: In vivo Cine intravascular ultrasound and virtual histology intravascular ultrasound (IVUS) slices were acquired at 20 plaque sites from 13 patients. A three-dimensional thin-slice structure-only model was constructed for each slice to obtain patient-specific in vivo material parameter values following an iterative scheme. Effective Young's modulus (YM) was calculated to indicate plaque stiffness for easy comparison purposes. IVUS-based 3D thin-slice models using in vivo and ex vivo material properties were constructed to investigate their impacts on plaque wall stress/strain (PWS/PWSn) calculations.Results: The average YM values in the axial and circumferential directions for the 20 plaque slices were 599.5 and 1,042.8 kPa, respectively, 36.1% lower than those from published ex vivo data. The YM values in the circumferential direction of the softest and stiffest plaques were 103.4 and 2,317.3 kPa, respectively. The relative difference of mean PWSn on lumen using the in vivo and ex vivo material properties could be as high as 431%, while the relative difference of mean PWS was much lower, about 3.07% on average.Conclusion: There is a large inter-patient and intra-patient variability in the in vivo plaque material properties. In vivo material properties have a great impact on plaque stress/strain calculations. In vivo plaque material properties have a greater impact on strain calculations. Large-scale-patient studies are needed to further verify our findings.

High Wall Shear Stress can Predict Wall Degradation in Ascending Aortic Aneurysms: An Integrated Biomechanics Study.

Background: Blood flow patterns can alter material properties of ascending thoracic aortic aneurysms (ATAA) via vascular wall remodeling. This study examines the relationship between wall shear stress (WSS) obtained from image-based computational modelling with tissue-derived mechanical and microstructural properties of the ATAA wall using segmental analysis. Methods: Ten patients undergoing surgery for ATAA were recruited. Exclusions: bicuspid aortopathy, connective tissue disease. All patients had pre-operative 4-dimensional flow magnetic resonance imaging (4D-MRI), allowing for patient-specific computational fluid dynamics (CFD) analysis and anatomically precise WSS mapping of ATAA regions (6–12 segments per patient). ATAA samples were obtained from surgery and subjected to region-specific tensile and peel testing (matched to WSS segments). Computational pathology was used to characterize elastin/collagen abundance and smooth muscle cell (SMC) count. Results: Elevated values of WSS were predictive of: reduced wall thickness [coef −0.0489, 95% CI (−0.0905, −0.00727), p = 0.022] and dissection energy function (longitudinal) [−15,0, 95% CI (−33.00, −2.98), p = 0.048]. High WSS values also predicted higher ultimate tensile strength [coef 0.136, 95% CI (0 0.001, 0.270), p = 0.048]. Additionally, elevated WSS also predicted a reduction in elastin levels [coef −0.276, 95% (CI −0.531, −0.020), p = 0.035] and lower SMC count ([oef −6.19, 95% CI (−11.41, −0.98), p = 0.021]. WSS was found to have no effect on collagen abundance or circumferential mechanical properties. Conclusions: Our study suggests an association between elevated WSS values and aortic wall degradation in ATAA disease. Further studies might help identify threshold values to predict acute aortic events.

The Emergence of Contemplative Neuroscience

Image-based computational models have been used by many researchers to quantify mechanical stress/strain conditions and flow wall shear stress and try to use them to understand mechanisms governing vulnerable plaque progression and rupture processes. In this chapter, we will focus on human coronary plaque modeling efforts based on in vivo imaging data, including image preparation, 3D reconstruction, residual stress, vessel zero-load geometry, in vivo vessel material properties, plaque progression prediction, and plaque vulnerability change prediction based on patient follow-up data. Imaging and modeling limitations and challenges will also be discussed.

Toward Patient-Specific Prediction of Ablation Strategies for Atrial Fibrillation Using Deep Learning.

Atrial fibrillation (AF) is a common cardiac arrhythmia that affects 1% of the population worldwide and is associated with high levels of morbidity and mortality. Catheter ablation (CA) has become one of the first line treatments for AF, but its success rates are suboptimal, especially in the case of persistent AF. Computational approaches have shown promise in predicting the CA strategy using simulations of atrial models, as well as applying deep learning to atrial images. We propose a novel approach that combines image-based computational modelling of the atria with deep learning classifiers trained on patient-specific atrial models, which can be used to assist in CA therapy selection. Therefore, we trained a deep convolutional neural network (CNN) using a combination of (i) 122 atrial tissue images obtained by unfolding patient LGE-MRI datasets, (ii) 157 additional synthetic images derived from the patient data to enhance the training dataset, and (iii) the outcomes of 558 CA simulations to terminate several AF scenarios in the corresponding image-based atrial models. Four CNN classifiers were trained on this patient-specific dataset balanced using several techniques to predict three common CA strategies from the patient atrial images: pulmonary vein isolation (PVI), rotor-based ablation (Rotor) and fibrosis-based ablation (Fibro). The training accuracy for these classifiers ranged from 96.22 to 97.69%, while the validation accuracy was from 78.68 to 86.50%. After training, the classifiers were applied to predict CA strategies for an unseen holdout test set of atrial images, and the results were compared to outcomes of the respective image-based simulations. The highest success rate was observed in the correct prediction of the Rotor and Fibro strategies (100%), whereas the PVI class was predicted in 33.33% of the cases. In conclusion, this study provides a proof-of-concept that deep neural networks can learn from patient-specific MRI datasets and image-derived models of AF, providing a novel technology to assist in tailoring CA therapy to a patient.

Multi-patient study for coronary vulnerable plaque model comparisons: 2D/3D and fluid\u2013structure interaction simulations

Several image-based computational models have been used to perform mechanical analysis for atherosclerotic plaque progression and vulnerability investigations. However, differences of computational predictions from those models have not been quantified at multi-patient level. In vivo intravascular ultrasound (IVUS) coronary plaque data were acquired from seven patients. Seven 2D/3D models with/without circumferential shrink, cyclic bending and fluid–structure interactions (FSI) were constructed for the seven patients to perform model comparisons and quantify impact of 2D simplification, circumferential shrink, FSI and cyclic bending plaque wall stress/strain (PWS/PWSn) and flow shear stress (FSS) calculations. PWS/PWSn and FSS averages from seven patients (388 slices for 2D and 3D thin-layer models) were used for comparison. Compared to 2D models with shrink process, 2D models without shrink process overestimated PWS by 17.26%. PWS change at location with greatest curvature change from 3D FSI models with/without cyclic bending varied from 15.07% to 49.52% for the seven patients (average = 30.13%). Mean Max-FSS, Min-FSS and Ave-FSS from the flow-only models under maximum pressure condition were 4.02%, 11.29% and 5.45% higher than those from full FSI models with cycle bending, respectively. Mean PWS and PWSn differences between FSI and structure-only models were only 4.38% and 1.78%. Model differences had noticeable patient variations. FSI and flow-only model differences were greater for minimum FSS predictions, notable since low FSS is known to be related to plaque progression. Structure-only models could provide PWS/PWSn calculations as good approximations to FSI models for simplicity and time savings in calculation.

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

AimsTo 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 resultsVirtual 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.ConclusionVirtual 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.

Image-Based Computational Model Predicts Dobutamine-Induced Hemodynamic Changes in Patients With Aortic Coarctation.

Pharmacological stress testing can help to uncover pathological hemodynamic conditions and is, therefore, used in the clinical routine to assess patients with structural heart diseases such as aortic coarctation with borderline indication for treatment. The aim of this study was to develop and test a reduced-order model predicting dobutamine stress induced pressure gradients across the coarctation. The reduced-order model was developed based on n=21 imaging data sets of patients with aortic coarctation and a meta-analysis of subjects undergoing dobutamine stress testing. Within an independent test cohort of n=21 patients with aortic coarctation, the results of the model were compared with dobutamine stress testing during catheterization. In n=19 patients responding to dobutamine stress testing, pressure gradients across the coarctation during dobutamine stress increased from 15.7±5.1 to 33.6±10.3 mm Hg (paired t test, P<0.001). The model-predicted pressure gradients agreed with catheter measurements with a mean difference of -2.2 mm Hg and a limit of agreement of ±11.16 mm Hg according to Bland-Altman analysis. Significant equivalence between catheter-measured and simulated pressure gradients during stress was found within the study cohort (two 1-sided tests of equivalence with a noninferiority margin of 5.0 mm Hg, 33.6±10.33 versus 31.5±11.15 mm Hg, P=0.021). The developed reduced-order model can instantly predict dobutamine-induced hemodynamic changes with accuracy equivalent to heart catheterization in patients with aortic coarctation. The method is easy to use, available as a web-based calculator, and provides a promising alternative to conventional stress testing in the clinical routine. Registration: URL: https://www.clinicaltrials.gov. Unique identifier: NCT02591940.

Abstract 14985: Presence of Repolarization Gradients Reverses Post-infarct Ventricular Tachycardia Exit and Entrance Sites in Personalized Digital Hearts

Introduction: Post-infarct ventricular tachycardias (VT) arise due to structural remodeling (scarring). Physiological repolarization gradients (apicobasal and transmural) exist in the human heart, but the effects on post-infarct VT dynamics are unknown. Hypothesis: We hypothesized that incorporation of repolarization gradients in personalized digital hearts of post-infarct patients impacts VT exit sites without altering the location of the VTs. Methods: 3D late-gadolinium enhanced CMR images were acquired from 7 post-infarct patients. Personalized image-based computational heart models (digital hearts) representing scar and infarct border zone distributions were constructed. Apicobasal (AB) and transmural (TM) repolarization gradients were incorporated using a validated method (Fig A). VTs were induced at baseline (no repolarization gradient) via rapid pacing in the right ventricular apex, using two pacing cycle lengths, mimicking non-invasive programmed stimulation. Pacing protocols that induced baseline VTs were repeated under AB and TM conditions. Results: Ten VTs were induced in baseline digital hearts. 8 AB VTs and 8 TM VTs were induced; the remaining 2 VTs for both AB and TM digital hearts could not be induced. 5/8 induced AB VTs had VT exit sites matching baseline VT exit sites; the remaining 3/8 AB VTs had reversed VT exit and entrance sites from the corresponding baseline VTs (Fig B, VT 1 &amp; 2). 4/8 induced TM VTs had exit sites that matched those at baseline; the remaining TM VTs had exit and entrance sites reversed from those of baseline VTs (Fig B, VT 1, 2 &amp; 3). All 8 AB VTs and 8 TM VTs had the same location as corresponding baseline VTs. Conclusion: AB and TM repolarization gradients can act to reverse VT entrance and exit sites without changing VT location. Thus, incorporation of physiological repolarization gradients into personalized digital hearts may not impact VT ablation targeting but may affect accurate prediction of VT exit or entrance sites.