Machine learning approaches to surrogate multifidelity Growth and Remodeling models for efficient abdominal aortic aneurysmal applications

Computational Growth and Remodeling (G&R) models have been widely used to capture the pathological development of arterial diseases and have shown promise for aiding clinical diagnosis, prognosis prediction, and staging classification. However, due to the high complexity of the arterial adaptation mechanism, high-fidelity arterial G&R simulation usually takes hours or even days, which hinders its application in clinical practice. To remedy this problem, we develop a computationally efficient arterial G&R simulation framework that comprehensively combines the physics-based G&R simulations and data-driven machine learning approaches. The proposed framework greatly enhances the computational efficiency of arterial G&R simulations, thereby enabling more time-consuming arterial applications, including personalized parameter estimation and arterial disease progression prediction. In particular, we achieve significant computational cost reduction mainly through two methods: (1) constructing a Multifidelity Surrogate (MFS) to approximate multifidelity G&R simulations by using a cokriging approach and (2) developing a novel iterative optimization algorithm for personalized parameter estimation. The proposed framework is demonstrated by estimating G&R model parameters and predicting individual aneurysm growth using follow-up CT images of Abdominal Aortic Aneurysms (AAAs) from 21 patients. Results show that the personalized parameters are satisfactorily estimated and the growth of AAAs is predicted within the clinically relevant time frame, i.e., less than 2 h, without a loss of accuracy.

Prediction of rupture risk in cerebral aneurysms by comparing clinical cases with fluid-structure interaction analyses

Cerebral aneurysms should be treated on the basis of accurate rupture risk prediction. Nowadays, the rupture risk in aneurysms has been estimated using hemodynamic parameters. In this paper, we suggest a new way to predict the rupture risks in cerebral aneurysms by using fluid–structure interaction (FSI) analysis for better decision-making regarding treatment. A patient-specific model was constructed using digital subtraction angiography of 51 cerebral aneurysms. For each model, a thin-walled area (TWA) was first predicted using computational fluid dynamics (CFD), and then the highest equivalent strain in the TWA was calculated with FSI by varying wall thicknesses and mechanical properties. A critical curve was made from 16 FSI results for each patient-specific model to estimate the rupture risk. On average, the equivalent strains of the ruptured aneurysms were higher than those of the unruptured aneurysms. Furthermore, the patterns of critical curves between unruptured and ruptured aneurysms were clearly distinguishable. From the rupture risk evaluation based on the cut-off value, 24 of the 27 unruptured aneurysms and 15 of the 24 ruptured aneurysms were matched with actual-clinical setting cases. The critical curve proposed in the present study could be an effective tool for the prediction of the rupture risk of aneurysm.

Aneurysms of the Lenticulostriate Artery: A Systematic Review

BACKGROUND

Lenticulostriate artery aneurysms (LSAs) are rare vascular aberrations. Despite the potentially catastrophic sequelae of aneurysmal rupture, the optimal management strategy for LSA aneurysms has not been determined. The aim of the present review is to provide an overview of the clinical presentation and treatment strategies for LSA aneurysms.

METHODS

Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, a literature search was conducted in the PubMed, Cochrane, EBSCOhost, Scopus, Web of Science, and ProQuest search engines to identify reported studies of LSA aneurysms until July 1, 2020. A descriptive analysis was performed.

RESULTS

A total of 71 studies with 112 cases of LSA aneurysms were included. Patient age ranged from 2 months to 83 years (median, 44.5 years). Male and female patients were affected similarly (49% and 51%, respectively). The most common presentation was aneurysmal rupture (78%), and headache was the most frequently reported symptom (36%). Overall, 48% of the patients had undergone underwent surgical treatment, 30% conservative management, 21% endovascular treatment, and 1% radiosurgery. Four patients died, all of whom had presented with aneurysmal rupture.

CONCLUSION

We have summarized the reported cases of LSA aneurysms, with their clinical presentation, management, and outcomes, for physicians who may be confronted with this diagnosis. Future studies that use available classification systems and include as much detail as possible should be encouraged to fully elucidate the optimal management strategy for these patients.

Calibration of parameters for cardiovascular models with application to arterial growth

We present a computational framework for the calibration of parameters describing cardiovascular models with a focus on the application of growth of abdominal aortic aneurysms (AAA). The growth rate in this sort of pathology is considered a critical parameter in the risk management and is an essential indicator for the assessment of surveillance intervals. Parameters describing growth of AAAs are not measurable directly and need to be estimated from available data often given by medical imaging technologies. Registration procedures often applied in standard workflows of parameter identification to extract the image encoded information are a source of significant systematic error. The concept of surface currents provides means to effectively avoid this source of errors by establishing a mathematical framework to compare surface information, directly accessible from image data. By utilizing this concept it is possible to inversely estimate growth parameters using sophisticated numerical models of AAAs from measurements available as surface information. In this work we present a framework to obtain spatial distributions of parameters governing growth of arterial tissue, and we show how the use of surface currents can significantly improve the results. We further present the application to patient specific follow-up data resulting in a spatial map of volumetric growth rates enabling, for the first time, prediction of further AAA expansion.

New small-intestine modeling method for surface-based computational human phantoms

When converting voxel phantoms to a surface format, the small intestine (SI), which is usually not accurately represented in a voxel phantom due to its complex and irregular shape on one hand and the limited voxel resolutions on the other, cannot be directly converted to a high-quality surface model. Currently, stylized pipe models are used instead, but they are strongly influenced by developer's subjectivity, resulting in unacceptable geometric and dosimetric inconsistencies. In this paper, we propose a new method for the construction of SI models based on the Monte Carlo approach. In the present study, the proposed method was tested by constructing the SI model for the polygon-mesh version of the ICRP reference male phantom currently under development. We believe that the new SI model is anatomically more realistic than the stylized SI models. Furthermore, our simulation results show that the new SI model, for both external and internal photon exposures, leads to dose values that are more similar to those of the original ICRP male voxel phantom than does the previously constructed stylized SI model.

Intracranial aneurysmal pulsatility as a new individual criterion for rupture risk evaluation: biomechanical and numeric approach (IRRAs Project)

BACKGROUND AND PURPOSE

The present study follows an experimental work based on the characterization of the biomechanical behavior of the aneurysmal wall and a numerical study where a significant difference in term of volume variation between ruptured and unruptured aneurysm was observed in a specific case. Our study was designed to highlight by means of numeric simulations the correlation between aneurysm sac pulsatility and the risk of rupture through the mechanical properties of the wall.

MATERIALS AND METHODS

In accordance with previous work suggesting a correlation between the risk of rupture and the material properties of cerebral aneurysms, 12 fluid-structure interaction computations were performed on 12 "patient-specific" cases, corresponding to typical shapes and locations of cerebral aneurysms. The variations of the aneurysmal volume during the cardiac cycle (ΔV) are compared by using wall material characteristics of either degraded or nondegraded tissues.

RESULTS

Aneurysms were located on 6 different arteries: middle cerebral artery (4), anterior cerebral artery (3), internal carotid artery (1), vertebral artery (1), ophthalmic artery (1), and basilar artery (1). Aneurysms presented different shapes (uniform or multilobulated) and diastolic volumes (from 18 to 392 mm3). The pulsatility (ΔV/V) was significantly larger for a soft aneurysmal material (average of 26%) than for a stiff material (average of 4%). The difference between ΔV, for each condition, was statistically significant: P=.005.

CONCLUSIONS

The difference in aneurysmal pulsatility as highlighted in this work might be a relevant patient-specific predictor of aneurysm risk of rupture.

Understanding the role of hemodynamics in the initiation, progression, rupture, and treatment outcome of cerebral aneurysm from medical image-based computational studies

About a decade ago, the first image-based computational hemodynamic studies of cerebral aneurysms were presented. Their potential for clinical applications was the result of a right combination of medical image processing, vascular reconstruction, and grid generation techniques used to reconstruct personalized domains for computational fluid and solid dynamics solvers and data analysis and visualization techniques. A considerable number of studies have captivated the attention of clinicians, neurosurgeons, and neuroradiologists, who realized the ability of those tools to help in understanding the role played by hemodynamics in the natural history and management of intracranial aneurysms. This paper intends to summarize the most relevant results in the field reported during the last years.

Physical factors effecting cerebral aneurysm pathophysiology

Many factors that are either blood-, wall-, or hemodynamics-borne have been associated with the initiation, growth, and rupture of intracranial aneurysms. The distribution of cerebral aneurysms around the bifurcations of the circle of Willis has provided the impetus for numerous studies trying to link hemodynamic factors (flow impingement, pressure, and/or wall shear stress) to aneurysm pathophysiology. The focus of this review is to provide a broad overview of such hemodynamic associations as well as the subsumed aspects of vascular anatomy and wall structure. Hemodynamic factors seem to be correlated to the distribution of aneurysms on the intracranial arterial tree and complex, slow flow patterns seem to be associated with aneurysm growth and rupture. However, both the prevalence of aneurysms in the general population and the incidence of ruptures in the aneurysm population are extremely low. This suggests that hemodynamic factors and purely mechanical explanations by themselves may serve as necessary, but never as necessary and sufficient conditions of this disease's causation. The ultimate cause is not yet known, but it is likely an additive or multiplicative effect of a handful of biochemical and biomechanical factors.

A shell-based inverse approach of stress analysis in intracranial aneurysms

Predicting pressure induced wall stress in intracranial aneurysms continues to be of interest for aneurysm safety assessment. In quasi-static analysis, there are two distinct approaches that one may take, the forward approach and the inverse approach. The inverse approach starts from a deformed configuration and thus is naturally suited to image-based, patient-specific analysis. Early studies by the authors' team suggested that the inverse approach, in the context of estimating the wall stress in cerebral aneurysms, depends weakly on the material description. In this article, we present a population study to further demonstrate the inverse method, in particular, the remarkable feature of insensitivity to material properties. Twenty-six aneurysm models derived from patient-specific images were employed in the study. Wall stresses were predicted in both the inverse and forward approaches using three material models. Results showed that, while forward computation yielded up to ~100% stress difference between some materials, the inverse solutions stayed close across materials. The inverse method, in addition to being methodologically accurate in dealing with pre-deformations, has the added convenience of insensitivity to uncertainties in wall tissue properties. New insight into the stress-geometry relation was also discussed.

A Quasi-Nonlinear Analysis of the Anisotropic Behaviour of Human Gallbladder Wall

Estimation of biomechanical parameters of soft tissues from noninvasive measurements has clinical significance in patient-specific modeling and disease diagnosis. In this work, we present a quasi-nonlinear method that is used to estimate the elastic moduli of the human gallbladder wall. A forward approach based on a transversely isotropic membrane material model is used, and an inverse iteration is carried out to determine the elastic moduli in the circumferential and longitudinal directions between two successive ultrasound images of gallbladder. The results demonstrate that the human gallbladder behaves in an anisotropic manner, and constitutive models need to incorporate this. The estimated moduli are also nonlinear and patient dependent. Importantly, the peak stress predicted here differs from the earlier estimate from linear membrane theory. As the peak stress inside the gallbladder wall has been found to strongly correlate with acalculous gallbladder pain, reliable mechanical modeling for gallbladder tissue is crucial if this information is to be used in clinical diagnosis.

Comprehensive validation of computational fluid dynamics simulations of in-vivo blood flow in patient-specific cerebral aneurysms

PURPOSE

Recently, image-based computational fluid dynamic (CFD) simulations have been proposed to investigate the local hemodynamics inside human cerebral aneurysms. It was suggested that the knowledge of the computed three-dimensional flow fields can be used to assist clinical risk assessment and treatment decision making. Therefore, it was desired to know the reliability of CFD for cerebral blood flow simulation, and be able to provide clinical feedback. However, the validations were not yet comprehensive as they lack either patient-specific boundary conditions (BCs) required for CFD simulations or quantitative comparison methods.

METHODS

In this study, based on a recently proposed in-vitro quantitative CFD evaluation approach via virtual angiography, the CFD evaluation was extended from phantom to patient studies. In contrast to previous work, patient-specific blood flow rates obtained by transcranial color coded Doppler ultrasound measurements were used to impose CFD BCs. Virtual angiograms (VAs) were constructed which resemble clinically acquired angiograms (AAs). Quantitative measures were defined to thoroughly evaluate the correspondence of the detailed flow features between the AAs and the VAs, and thus, the reliability of CFD simulations.

RESULTS

The proposed simulation pipeline provided a comprehensive validation method of CFD simulation for reproducing cerebral blood flow, with a focus on the aneurysm region. Six patient cases were tested and close similarities were found in terms of spatial and temporal variations of contrast agent (CA) distribution between AAs and VAs. For patient #1 to #5, discrepancies of less than 11% were found for the relative root mean square errors in time intensity curve comparisons from characteristic vasculature positions. For patient #6, where the CA concentration curve at vessel inlet cannot be directly extracted from the AAs and given as a BC, deviations about 20% were found.

CONCLUSIONS

As a conclusion, the reliability of the CFD simulations was well confirmed. Besides, it was shown that the accuracy of CFD simulations was closely related to the input BCs.

Analysis of several biomechanical models for the simulation of lamb liver behaviour using similarity coefficients from medical image

In this study, six biomechanical models for simulating lamb liver behaviour are presented. They are validated using similarity coefficients from Medical Image on reconstructed volumes from computerised tomography images. In particular, the Jaccard and Hausdorff coefficients are used. Loads of 20 and 40 g are applied to the livers and their deformation is simulated by means of the finite element method. The models used are a linear elastic model, a neo-Hookean model, a Mooney-Rivlin model, an Ogden model, a linear viscoelastic model and a viscohyperelastic model. The model that provided a behaviour that is closest to reality was the viscohyperelastic model, where the hyperelastic part was modelled with an Ogden model.

Biomechanical wall properties of human intracranial aneurysms resected following surgical clipping (IRRAs Project)

BACKGROUND AND PURPOSE

Individual rupture risk assessment of intracranial aneurysms is a major issue in the clinical management of asymptomatic aneurysms. Aneurysm rupture occurs when wall tension exceeds the strength limit of the wall tissue. At present, aneurysmal wall mechanics are poorly understood and thus, risk assessment involving mechanical properties is inexistent. Aneurysm computational hemodynamics studies make the assumption of rigid walls, an arguable simplification. We therefore aim to assess mechanical properties of ruptured and unruptured intracranial aneurysms in order to provide the foundation for future patient-specific aneurysmal risk assessment. This work also challenges some of the currently held hypotheses in computational flow hemodynamics research.

METHODS

A specific conservation protocol was applied to aneurysmal tissues following clipping and resection in order to preserve their mechanical properties. Sixteen intracranial aneurysms (11 female, 5 male) underwent mechanical uniaxial stress tests under physiological conditions, temperature, and saline isotonic solution. These represented 11 unruptured and 5 ruptured aneurysms. Stress/strain curves were then obtained for each sample, and a fitting algorithm was applied following a 3-parameter (C(10), C(01), C(11)) Mooney-Rivlin hyperelastic model. Each aneurysm was classified according to its biomechanical properties and (un)rupture status.

RESULTS

Tissue testing demonstrated three main tissue classes: Soft, Rigid, and Intermediate. All unruptured aneurysms presented a more Rigid tissue than ruptured or pre-ruptured aneurysms within each gender subgroup. Wall thickness was not correlated to aneurysmal status (ruptured/unruptured). An Intermediate subgroup of unruptured aneurysms with softer tissue characteristic was identified and correlated with multiple documented risk factors of rupture.

CONCLUSION

There is a significant modification in biomechanical properties between ruptured aneurysm, presenting a soft tissue and unruptured aneurysms, presenting a rigid material. This finding strongly supports the idea that a biomechanical risk factor based assessment should be utilized in the to improve the therapeutic decision making.

Characterizing heterogeneous properties of cerebral aneurysms with unknown stress-free geometry: a precursor to in vivo identification

Knowledge of elastic properties of cerebral aneurysms is crucial for understanding the biomechanical behavior of the lesion. However, characterizing tissue properties using in vivo motion data presents a tremendous challenge. Aside from the limitation of data accuracy, a pressing issue is that the in vivo motion does not expose the stress-free geometry. This is compounded by the nonlinearity, anisotropy, and heterogeneity of the tissue behavior. This article introduces a method for identifying the heterogeneous properties of aneurysm wall tissue under unknown stress-free configuration. In the proposed approach, an accessible configuration is taken as the reference; the unknown stress-free configuration is represented locally by a metric tensor describing the prestrain from the stress-free configuration to the reference configuration. Material parameters are identified together with the metric tensor pointwisely. The paradigm is tested numerically using a forward-inverse analysis loop. An image-derived sac is considered. The aneurysm tissue is modeled as an eightply laminate whose constitutive behavior is described by an anisotropic hyperelastic strain-energy function containing four material parameters. The parameters are assumed to vary continuously in two assigned patterns to represent two types of material heterogeneity. Nine configurations between the diastolic and systolic pressures are generated by forward quasi-static finite element analyses. These configurations are fed to the inverse analysis to delineate the material parameters and the metric tensor. The recovered and the assigned distributions are in good agreement. A forward verification is conducted by comparing the displacement solutions obtained from the recovered and the assigned material parameters at a different pressure. The nodal displacements are found in excellent agreement.