Search

BackgroundThe presence of blebs increases the rupture risk of intracranial aneurysms (IAs).ObjectiveTo evaluate whether cross-sectional bleb formation models can identify aneurysms with focalized enlargement in longitudinal series.MethodsHemodynamic, geometric, and anatomical variables derived from computational fluid dynamics models of 2265 IAs from a cross-sectional dataset were used to train machine learning (ML) models for bleb development. ML algorithms, including logistic regression, random forest, bagging method, support vector machine, and K-nearest neighbors, were validated using an independent cross-sectional dataset of 266 IAs. The models' ability to identify aneurysms with focalized enlargement was evaluated using a separate longitudinal dataset of 174 IAs. Model performance was quantified by the area under the receiving operating characteristic curve (AUC), the sensitivity and specificity, positive predictive value, negative predictive value, F1 score, balanced accuracy, and misclassification error.ResultsThe final model, with three hemodynamic and four geometrical variables, along with aneurysm location and morphology, identified strong inflow jets, non-uniform wall shear stress with high peaks, larger sizes, and elongated shapes as indicators of a higher risk of focal growth over time. The logistic regression model demonstrated the best performance on the longitudinal series, achieving an AUC of 0.9, sensitivity of 85%, specificity of 75%, balanced accuracy of 80%, and a misclassification error of 21%.ConclusionsModels trained with cross-sectional data can identify aneurysms prone to future focalized growth with good accuracy. These models could potentially be used as early indicators of future risk in clinical practice.

BackgroundBleb presence in intracranial aneurysms (IAs) is a known indication of instability and vulnerability.ObjectiveTo develop and evaluate predictive models of bleb development in IAs based on hemodynamics, geometry, anatomical location, and patient population.MethodsCross-sectional data (one time point) of 2395 IAs were used for training bleb formation models using machine learning (random forest, support vector machine, logistic regression, k-nearest neighbor, and bagging). Aneurysm hemodynamics and geometry were characterized using image-based computational fluid dynamics. A separate dataset with 266 aneurysms was used for model evaluation. Model performance was quantified by the area under the receiving operating characteristic curve (AUC), true positive rate (TPR), false positive rate (FPR), precision, and balanced accuracy.ResultsThe final model retained 18 variables, including hemodynamic, geometrical, location, multiplicity, and morphology parameters, and patient population. Generally, strong and concentrated inflow jets, high speed, complex and unstable flow patterns, and concentrated, oscillatory, and heterogeneous wall shear stress patterns together with larger, more elongated, and more distorted shapes were associated with bleb formation. The best performance on the validation set was achieved by the random forest model (AUC=0.82, TPR=91%, FPR=36%, misclassification error=27%).ConclusionsBased on the premise that aneurysm characteristics prior to bleb formation resemble those derived from vascular reconstructions with their blebs virtually removed, machine learning models can identify aneurysms prone to bleb development with good accuracy. Pending further validation with longitudinal data, these models may prove valuable for assessing the propensity of IAs to progress to vulnerable states and potentially rupturing.

BackgroundBlebs are important secondary structures of intracranial aneurysms associated with increased rupture risk and can affect local wall stress and hemodynamics. Mechanisms of bleb development and evolution are not clearly understood. We investigate the relationship between blebs with different wall characteristics and local hemodynamics and rupture sites.MethodsBlebs with different wall appearances in intra-operative videos were analyzed with image-based computational fluid dynamics. Thin red blebs were compared against thick atherosclerotic/hyperplastic white/yellow blebs. Rupture points were identified in videos of ruptured aneurysms harboring blebs.ResultsThin blebs tended to be closer to the inflow than atherosclerotic blebs of the same aneurysm (P=0.0234). Blebs near the inflow had higher velocity (P=0.0213), vorticity (P=0.0057), shear strain rate (P=0.0084), wall shear stress (WSS) (P=0.0085), and WSS gradient (P=0.0151) than blebs far from the inflow. In a subset of 12 ruptured aneurysms harboring blebs, rupture points were associated with thin blebs in 42% of aneurysms, atherosclerotic blebs in 25%, and were away from blebs in the remaining 33%.ConclusionsNot all blebs are equal; some have thin translucent walls while others have thick atherosclerotic walls. Thin blebs tend to be located closer to the inflow than atherosclerotic blebs. Blebs near the inflow are exposed to stronger flows with higher and spatially variable WSS than blebs far from the inflow which tend to have uniformly lower WSS. Aneurysms can rupture at thin blebs, atherosclerotic blebs, and even away from blebs. Further study of wall failure in aneurysms with different bleb types is needed.

BackgroundAlthough it is generally believed that blebs represent weaker spots in the walls of intracranial aneurysms (IAs), it is largely unknown which aneurysm characteristics favor their development.ObjectiveTo investigate possible associations between aneurysm hemodynamic and geometric characteristics and the development of blebs in intracranial aneurysms.MethodsA total of 270 IAs in 199 patients selected for surgical clipping were studied. Blebs were visually identified and interactively marked on patient-specific vascular models constructed from presurgical images. Blebs were then deleted from the vascular reconstruction to approximate the aneurysm before bleb formation. Computational fluid dynamics studies were performed in these models and in cases without blebs. Hemodynamic and geometric characteristics of aneurysms with and without blebs were compared.ResultsA total of 173 aneurysms had no blebs, while 97 aneurysms had a total of 122 blebs. Aneurysms favoring bleb formation had stronger (p<0.0001) and more concentrated inflow jets (p<0.0001), higher flow velocity (p=0.0061), more complex (p<0.0001) and unstable (p=0.0157) flow patterns, larger maximum wall shear stress (WSS; p<0.0001), more concentrated (p=0.0005) and oscillatory (p=0.0004) WSS distribution, and a more heterogeneous WSS field (p<0.0001), than aneurysms without blebs. They were also larger (p<0.0001), more elongated (p<0.0001), had wider necks (p=0.0002), and more distorted and irregular shapes (p<0.0001).ConclusionsStrong and concentrated inflow jets, high-speed, complex, and unstable flow patterns, and concentrated, oscillatory, and heterogeneous WSS patterns favor the formation of blebs in IAs. Blebs are more likely to form in large, elongated, and irregularly shaped aneurysms. These adverse characteristics could be considered signs of aneurysm instability when evaluating aneurysms for conservative observation or treatment.

BackgroundBlebs are rupture risk factors in intracranial aneurysms (IAs), but their prevalence, distribution, and associations with clinical factors as well as their causes and effects on aneurysm vulnerability remain unclear.MethodsA total of 122 blebs in 270 IAs selected for surgery were studied using patient-specific vascular reconstructions from 3D angiographic images. Bleb geometry, location on the aneurysm, and frequency of occurrence in aneurysms at different locations were analyzed. Associations between gender, age, smoking, hypertension, hormone therapy, dental infection, and presence of blebs were investigated.ResultsOf all aneurysms with blebs, 77% had a single bleb and 23% had multiple blebs. Only 6% of blebs were at the neck, while 46% were in the body and 48% in the dome. Aneurysms with blebs were larger (p<0.0001), more elongated (p=0.0002), and with wider necks than aneurysms without blebs. Bleb presence was associated with dental infection (p=0.0426) and negatively associated with hormone therapy (p=0.0426) in women. Anterior and posterior communicating arteries had larger percentages of aneurysms with blebs than internal carotid arteries. Patients with a history of hypertension tended to have a larger percentage of aneurysms with blebs. However, these trends did not reach significance in this sample.ConclusionsBlebs are common in IAs, and most aneurysms harboring blebs have a single bleb. Blebs in the aneurysm neck are rare, but they are equally common in the body and dome. The presence of blebs in IAs was associated with dental infection, and negatively associated with hormone replacement therapy.

Purpose: Incidental aneurysms pose a challenge to physicians who need to decide whether or not to treat them. A statistical model could potentially support such treatment decisions. The aim of this study was to compare a previously developed aneurysm rupture logistic regression probability model (LRM) to other machine learning (ML) classifiers for discrimination of aneurysm rupture status. Methods: Hemodynamic, morphological, and patient-related information of 1631 cerebral aneurysms characterized by computational fluid dynamics simulations were used to train support vector machines (SVMs) with linear and RBF kernel (RBF-SVM), k-nearest neighbors (kNN), decision tree, random forest, and multilayer perceptron (MLP) neural network classifiers for predicting the aneurysm rupture status. The classifiers’ accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were evaluated and compared to the LRM using 249 test cases obtained from two external cohorts. Additionally, important variables were determined based on the random forest and weights of the linear SVM. Results: The AUCs of the MLP, LRM, linear SVM, RBF-SVM, kNN, decision tree, and random forest were 0.83, 0.82, 0.80, 0.81, 0.76, 0.70, and 0.79, respectively. The accuracy ranged between 0.76 (decision tree,) and 0.79 (linear SVM, RBF-SVM, and MLP). Important variables for predicting the aneurysm rupture status included aneurysm location, the mean surface curvature, and maximum flow velocity. Conclusion: The performance of the LRM was overall comparable to that of the other ML classifiers, confirming its potential for aneurysm rupture assessment. To further improve the predictions, additional information, e.g., related to the aneurysm wall, might be needed.

Purpose: Assessing the rupture probability of intracranial aneurysms (IAs) remains challenging. Therefore, hemodynamic simulations are increasingly applied toward supporting physicians during treatment planning. However, due to several assumptions, the clinical acceptance of these methods remains limited. Methods: To provide an overview of state-of-the-art blood flow simulation capabilities, the Multiple Aneurysms AnaTomy CHallenge 2018 (MATCH) was conducted. Seventeen research groups from all over the world performed segmentations and hemodynamic simulations to identify the ruptured aneurysm in a patient harboring five IAs. Although simulation setups revealed good similarity, clear differences exist with respect to the analysis of aneurysm shape and blood flow results. Most groups (12/71%) included morphological and hemodynamic parameters in their analysis, with aspect ratio and wall shear stress as the most popular candidates, respectively. Results: The majority of groups (7/41%) selected the largest aneurysm as being the ruptured one. Four (24%) of the participating groups were able to correctly select the ruptured aneurysm, while three groups (18%) ranked the ruptured aneurysm as the second most probable. Successful selections were based on the integration of clinically relevant information such as the aneurysm site, as well as advanced rupture probability models considering multiple parameters. Additionally, flow characteristics such as the quantification of inflow jets and the identification of multiple vortices led to correct predictions. Conclusions: MATCH compares state-of-the-art image-based blood flow simulation approaches to assess the rupture risk of IAs. Furthermore, this challenge highlights the importance of multivariate analyses by combining clinically relevant metadata with advanced morphological and hemodynamic quantification.

Supplemental Digital Content is available in the text.

BackgroundFlow diverters (FDs) are increasingly used in the treatment of intracranial aneurysms, and carry the risk of thromboembolic complications, even in patients treated with dual antiplatelet therapy. The effect of FDs on the downstream vascular is unknown. The aim of the study was to investigate vascular wall pulse wave velocity (PWV) and contractility changes following FD treatment in a rabbit model.MethodsFDs (Pipeline Embolic Device, Medtronic Inc., Irvine, California, USA) were implanted in the aorta of normal rabbits and sham-operated aorta were used as controls (n=6 per group). Pulse wave imaging with ultra-fast ultrasound at 1600 frames per second (Vantage, Verasonics, Inc., Kirkland, WA) was performed in the vessel wall distal to FD prior to device implantation and at 8- week follow-up to measure the PWV. Force contraction vascular reactivity studies were conducted in the aortic rings using an organ bath.ResultsThe difference in mean PWV in the follow-up compared with pre-implantation was significantly higher in the distal vessels compared with sham controls (1.18 m/s [SD=0.54] vs. 0.37 m/s [SD=1.09], P=0.03). Conversely, the aortic segments distal to the FD exhibited a 55% increase in vascular contractility compared with proximal segments (P=0.002). We observed a significant positive correlation between mean PWV and mean vascular contractility.ConclusionImplantation of FD was associated with increased PWV and vascular contractility, suggesting that FD implantation causes changes to the vascular wall. Further studies are needed to understand the clinical implication of changes in vascular PWV and contractility.

Background and purposeVertebrobasilar dolichoectatic and fusiform aneurysms (VBDAs) are known to have a poor natural history, with high rates of growth, rupture, and stroke. The purpose of this study was to identify hemodynamic characteristics that differ between VBDAs associated with growth, rupture, and stroke.Materials and methodsVBDAs with CT angiography or MR angiography followed longitudinally without treatment were studied. Unstable aneurysms were defined as those that grew or ruptured during follow-up. Aneurysms associated with stroke were defined as those associated with posterior circulation infarct at follow-up. Baseline data, including demographics, comorbidities, and aneurysm morphology and size were collected. Image based computational fluid dynamics models were created and run under pulsatile flow conditions. Relevant hemodynamic and geometric variables were calculated and compared between groups (stable vs unstable and no stroke vs stroke) using the Wilcoxon test.ResultsA total of 37 VBDAs were included (24 stable, 13 unstable; 30 no stroke, 7 stroke). Unstable aneurysms had lower shear rates (P=0.05), blood flow velocity (P=0.03), and lower vorticity (P=0.049) than stable aneurysms. In addition, unstable aneurysms had higher mean oscillatory shear indices (P=0.001). There were no differences in the hemodynamic characteristics of aneurysms in the stroke group compared with the non-stroke group.ConclusionThis small study suggests there may be hemodynamic differences between unstable and stable VBDAs. Unstable VBDAs appear to be under lower flow conditions with lower velocity, vorticity, and shear rates, and have more oscillatory flow. There was no difference in the hemodynamic characteristics of aneurysms in the stroke and no stroke group.

BackgroundWhile clinical and angiographic risk factors for intracranial aneurysm instability are well established, it is reasonable to postulate that intra-aneurysmal hemodynamics also have a role in aneurysm instability.ObjectiveTo identify hemodynamic characteristics that differ between radiologically unstable and stable unruptured intracranial aneurysms.Materials and methods12 pairs of unruptured intracranial aneurysms with a 3D rotational angiographic set of images and followed up longitudinally without treatment were studied. Each pair consisted of one stable aneurysm (no change on serial imaging) and one unstable aneurysm (demonstrated growth of at least 1 mm diameter or ruptured during follow-up) of matching size (within 10%) and locations. Patient-specific computational fluid dynamics models were created and run under pulsatile flow conditions. Relevant hemodynamic and geometric variables were calculated and compared between groups using the paired Wilcoxon test.ResultsThe area of the aneurysm under low wall shear stress (low shear stress area (LSA)) was 2.26 times larger in unstable aneurysms than in stable aneurysms (p=0.0499). The mean aneurysm vorticity was smaller by a factor of 0.57 in unstable aneurysms compared with stable aneurysms (p=0.0499). No statistically significant differences in geometric variables or shape indices were found.ConclusionsThis pilot study suggests there may be hemodynamic differences between unstable and stable unruptured cerebral aneurysms. In particular, the area under low wall shear stress was larger in unstable aneurysms. These findings should be considered tentative until confirmed by future larger studies.

Intracranial aneurysms and artery stenosis are vascular diseases with different pathophysiological characteristics. However, although unusual, aneurysms may coexist in up to 5% of patients with stenotic plaque, according to a previous study. Another study showed that incidental detection of cerebral aneurysm in the same cerebral circulation as the stenotic plaque was less than 2%. Patients with concomitant carotid artery stenosis and unruptured intracranial aneurysms pose a difficult management decision for the physician. Case reports showed patients who died due to aneurysm rupture months after endarterectomy but before aneurysm clipping, while others did not show any change in the aneurysm after plaque removal, having optimum outcome after aneurysm coiling. The purpose of this study is to investigate the intraaneurysmal hemodynamic changes before and after treatment of stenotic plaque. Idealized models were constructed with different stenotic grade, distance and relative position to the aneurysm. Digital removal of the stenotic plaque was performed in the reconstructed model of a patient with both pathologies. Computational fluid dynamic simulations were performed using a finite element method approach. Blood velocity field and hemodynamic forces were recorded and analyzed. Changes in the flow patterns and wall shear stress values and distributions were observed in both ideal and image-based models. Detailed investigation of wall shear stress distributions in patients with both pathologies is required to make the best management decision.

It is widely accepted that complexity in the flow pattern at the anterior communicating artery (AComA) is associated with the high rate of aneurysm formation in that location observed in large studies. A previous computational hemodynamic study showed a possible association between high maximum intraaneurysmal wall shear stress (WSS) at the systolic peak with rupture in a cohort of AComA aneurysms. In another study it was observed a connection between location of aneurysm blebs and regions of high WSS in models where blebs were virtually removed. However, others reported associations between low WSS and either rupture or blister formation. The purpose of this work is to study associations between hemodynamic patterns and AComA aneurysm initiation by comparing hemodynamics in the aneurysm and the normal model where the aneurysm was computationally removed. Vascular models of both right and left circulation were independently reconstructed from three-dimensional rotational angiography images using deformable models, and fused using a surface merging algorithm. The geometric models were then used to generate high-quality volumetric finite element grids of tetrahedra with an advancing front technique. For each patient, the second anatomical model was created by digitally removing the aneurysm. It was iteratively achieved by applying a Laplacian smoothing filter and remeshing the surface. Finite element blood flow numerical simulations were performed. It was observed that aneurysms initiated in regions of high and moderate WSS in the counterpart normal models. Adjacent or close to those regions, low WSS portions of the arterial wall were not affected by the disease.

The study of hemodynamics in arterial models constructed from patient‐specific medical images requires the solution of the incompressible flow equations in geometries characterized by complex branching tubular structures. The main challenge with this kind of geometries is that the convergence rate of the pressure Poisson solver is dominated by the graph depth of the computational grid. This paper presents a deflated preconditioned conjugate gradients (DPCG) algorithm for accelerating the pressure Poisson solver. A subspace deflation technique is used to approximate the lowest eigenvalues along the tubular domains. This methodology was tested with an idealized cylindrical model and three patient‐specific models of cerebral arteries and aneurysms constructed from medical images. For these cases, the number of iterations decreased by up to a factor of 16, while the total CPU time was reduced by up to 4 times when compared with the standard PCG solver. Copyright © 2009 John Wiley & Sons, Ltd.

Several extensions and improvements to surface merging procedures based on the extraction of iso-surfaces from a distance map defined on an adaptive background grid are presented. The main objective is to extend the application of these algorithms to surfaces with sharp edges and corners. In order to deal with objects of different length scales, the initial background grids are created using a Delaunay triangulation method and local voxelizations. A point enrichment technique that introduces points into the background grid along detected surface features such as ridges is used to ensure that these features are preserved in the final merged surface. The surface merging methodology is extended to include other Boolean operations between surface triangulations. The iso-surface extraction algorithms are modified to obtain the correct iso-surface for multi-component objects. The procedures are demonstrated with various examples, ranging from simple geometrical entities to complex engineering applications. The present algorithms allow realistic modelling of a large number of complex engineering geometries using overlapping components defined discretely, i.e. via surface triangulations. This capability is very useful for grid generation starting from data originated in measurements or images. Copyright © 2002 John Wiley & Sons, Ltd.