On the prevalence of flow instabilities from high-fidelity computational fluid dynamics of intracranial bifurcation aneurysms.

High-fidelity computational fluid dynamics (HF-CFD) has revealed the potential for high-frequency flow instabilities (aka "turbulent-like" flow) in intracranial aneurysms, consistent with classic in vivo and in vitro reports of bruits and/or wall vibrations. However, HF-CFD has typically been performed on limited numbers of cases, often with unphysiological inflow conditions or focused on sidewall-type aneurysms where flow instabilities may be inherently less prevalent. Here we report HF-CFD of 50 bifurcation aneurysm cases from the open-source Aneurisk model repository. These were meshed using quadratic finite elements having an average effective spatial resolution of 0.065 mm, and solved under physiologically-pulsatile flow conditions using a well-validated, minimally-dissipative solver with 20,000 time-steps per cardiac cycle Flow instability was quantified using the recently introduced spectral power index (SPI), which quantifies, from 0 to 1, the power associated with velocity fluctuations above those of the driving inflow waveform. Of the 50 cases, nearly half showed regions within the sac having SPI up to 0.5, often with non-negligible power into the 100's of Hz, and roughly 1/3 had sac-averaged SPI > 0.1. High SPI did not significantly predict rupture status in this cohort. Proper orthogonal decomposition of cases with highest SPIavg revealed time-varying energetics consistent with transient turbulence. Our reported prevalence of high-frequency flow instabilities in HF-CFD modelling of aneurysms suggests that care must be taken to avoid routinely overlooking them if we are to understand the highly dynamic mechanical forces to which some aneurysm walls may be exposed, and their prevalence in vivo.

On delayed transition to turbulence in an eccentric stenosis model for clean vs. noisy high-fidelity CFD.

Recent comparisons between experiments and computational fluid dynamics (CFD) simulations of flow in the Food and Drug Administration (FDA) standardized nozzle geometry have highlighted the potential sensitivity of axisymmetric CFD models to small perturbations induced by mesh and inlet velocity, particularly for Reynolds numbers (Re) in the transitional regime. This evokes the classic experiment of Reynolds on transition to turbulence in a straight pipe, which can be delayed, apparently indefinitely, if special care is taken to control for external influences. Such idealized experiments are, however, extremely difficult to perform and, in the context of cardiovascular modeling, belie the "noise" inherent in typical experimental and physiological systems. Previous high-fidelity CFD of a canonical eccentric (i.e., non-axisymmetric) stenosis model showed transition occurring for steady flow at Re ~ 700-800, with modest delay caused by the introduction of shear-thinning rheology. On the other hand, recent experimental measurements of steady flowing blood and blood-mimicking fluids in this same stenosis model report transition for Re ~ 400-500. Taking a cue from the FDA nozzle controversy, the present study demonstrates that the addition of small-magnitude random noise at the inlet brings the eccentric-stenosis CFD results more in-line with experiments, and reveals a more gradual transition towards turbulence. This highlights that, even in non-axisymmetric idealized geometries, unnaturally "clean" high-fidelity CFD may impede not only good agreement with experiments, but also understanding of the onset and character of blood flow instabilities as they may exist, naturally, in the vasculature.

Errors in power-law estimations of inflow rates for intracranial aneurysm CFD.

Patient-specific inflow rates are rarely available for computational fluid dynamics (CFD) studies of intracranial aneurysms. Instead, inflow rates are often estimated from parent artery diameters via power laws, i.e. Q ∝ Dn, reflecting adaptation of conduit arteries to demanded flow. The present study aimed to validate the accuracy of these power laws. Internal carotid artery (ICA) flow rates were measured from 25 ICA aneurysm patients via 2D phase contrast MRI. ICA diameters, derived from 3D segmentation of rotational angiograms, were used to estimate inflow rates via power laws from the aneurysm CFD literature assuming the same inlet wall shear stress (WSS) (n = 3), velocity (n = 2) or flow rate (n = 0) for all cases. To illustrate the potential impact of errors in flow rate estimates, pulsatile CFD was carried out for four cases having large errors for at least one power law. Flow rates estimated by n = 3 and n = 0 power laws had significant (p < 0.01) mean biases of -22% to +32%, respectively, but with individual errors ranging from -78% to +120%. The n = 2 power law had no significant bias, but had non-negligible individual errors of -58% to +71%. CFD showed similarly large errors for time-averaged sac WSS; however, these were reduced after normalizing by parent artery WSS. High frequency WSS fluctuations, evident in 2/4 aneurysms, were also sensitive to inflow rate errors. Care should therefore be exercised in the interpretation of aneurysm CFD studies that rely on power law estimates of inflow rates, especially if absolute (vs. normalized) WSS, or WSS instabilities, are of interest.

Better Than Nothing: A Rational Approach for Minimizing the Impact of Outflow Strategy on Cerebrovascular Simulations.

BACKGROUND AND PURPOSE: Computational fluid dynamics simulations of neurovascular diseases are impacted by various modeling assumptions and uncertainties, including outlet boundary conditions. Many studies of intracranial aneurysms, for example, assume zero pressure at all outlets, often the default ("do-nothing") strategy, with no physiological basis. Others divide outflow according to the outlet diameters cubed, nominally based on the more physiological Murray's law but still susceptible to subjective choices about the segmented model extent. Here we demonstrate the limitations and impact of these outflow strategies, against a novel "splitting" method introduced here. MATERIALS AND METHODS: With our method, the segmented lumen is split into its constituent bifurcations, where flow divisions are estimated locally using a power law. Together these provide the global outflow rate boundary conditions. The impact of outflow strategy on flow rates was tested for 70 cases of MCA aneurysm with 0D simulations. The impact on hemodynamic indices used for rupture status assessment was tested for 10 cases with 3D simulations. RESULTS: Differences in flow rates among the various strategies were up to 70%, with a non-negligible impact on average and oscillatory wall shear stresses in some cases. Murray-law and splitting methods gave flow rates closest to physiological values reported in the literature; however, only the splitting method was insensitive to arbitrary truncation of the model extent. CONCLUSIONS: Cerebrovascular simulations can depend strongly on the outflow strategy. The default zero-pressure method should be avoided in favor of Murray-law or splitting methods, the latter being released as an open-source tool to encourage the standardization of outflow strategies.

Vessel calibre and flow splitting relationships at the internal carotid artery terminal bifurcation.

OBJECTIVE: Vessel lumen calibres and flow rates are thought to be related by mathematical power laws, reflecting the optimization of cardiac versus metabolic work. While these laws have been confirmed indirectly via measurement of branch calibres, there is little data confirming power law relationships of flow distribution to branch calibres at individual bifurcations. APPROACH: Flow rates and diameters of parent and daughter vessels of the internal carotid artery terminal bifurcation were determined, via robust and automated methods, from 4D phase-contrast magnetic resonance imaging and 3D rotational angiography of 31 patients. MAIN RESULTS: Junction exponents were 2.06 ± 0.44 for relating parent to daughter branch diameters (geometrical exponent), and 2.45 ± 0.75 for relating daughter branch diameters to their flow division (flow split exponent). These exponents were not significantly different, but showed large inter- and intra-individual variations, and with confidence intervals excluding the theoretical optimum of 3. Power law fits of flow split versus diameter ratio and pooled flow rates versus diameters showed exponents of 2.17 and 1.96, respectively. A significant negative correlation was found between age and the geometrical exponent (r = -0.55, p = 0.003) but not the flow split exponent. We also found a dependence of our results on how lumen diameter is measured, possibly explaining some of the variability in the literature. SIGNIFICANCE: Our study confirms that, on average, division of flow to the middle and anterior cerebral arteries is related to these vessels' relative calibres via a power law, but it is closer to a square law than a cube law as commonly assumed.

Non-Newtonian versus numerical rheology: Practical impact of shear-thinning on the prediction of stable and unstable flows in intracranial aneurysms.

Computational fluid dynamics (CFD) shows promise for informing treatment planning and rupture risk assessment for intracranial aneurysms. Much attention has been paid to the impact on predicted hemodynamics of various modelling assumptions and uncertainties, including the need for modelling the non-Newtonian, shear-thinning rheology of blood, with equivocal results. Our study clarifies this issue by contextualizing the impact of rheology model against the recently demonstrated impact of CFD solution strategy on the prediction of aneurysm flow instabilities. Three aneurysm cases were considered, spanning a range of stable to unstable flows. Simulations were performed using a high-resolution/accuracy solution strategy with Newtonian and modified-Cross rheology models and compared against results from a so-called normal-resolution strategy. Time-averaged and instantaneous wall shear stress (WSS) distributions, as well as frequency content of flow instabilities and dome-averaged WSS metrics, were minimally affected by the rheology model, whereas numerical solution strategy had a demonstrably more marked impact when the rheology model was fixed. We show that point-wise normalization of non-Newtonian by Newtonian WSS values tended to artificially amplify small differences in WSS of questionable physiological relevance in already-low WSS regions, which might help to explain the disparity of opinions in the aneurysm CFD literature regarding the impact of non-Newtonian rheology. Toward the goal of more patient-specific aneurysm CFD, we conclude that attention seems better spent on solution strategy and other likely "first-order" effects (eg, lumen segmentation and choice of flow rates), as opposed to "second-order" effects such as rheology.

Improved reduced-order modelling of cerebrovascular flow distribution by accounting for arterial bifurcation pressure drops.

Reduced-order modelling offers the possibility to study global flow features in cardiovascular networks. In order to validate these models, previous studies have been conducted in which they compared 3D computational fluid dynamics simulations with reduced-order simulations. Discrepancies have been reported between the two methods. The loss of energy at the bifurcations is usually neglected and has been pointed out as a possible explanation for these discrepancies. We present distributed lumped models of cerebrovasculatures created automatically from 70 cerebrovascular networks segmented from 3D angiograms. The outflow rate repartitions predicted with and without modelling the energy loss at the bifurcations are compared against 3D simulations. When neglecting the energy loss at the bifurcations, the flow rates though the anterior cerebral arteries are overestimated by 4.7±6.8% (error relative to the inlet flow rate, mean ± standard deviation), impacting the remaining volume of flow going to the other vessels. When the energy loss is modelled, this error is dropping to 0.1±3.2%. Overall, over the total of 337 outlet vessels, when the energy losses at the bifurcations are not modelled the 95% of agreement is in the range of ±13.5% and is down to ±6.5% when the energy losses are considered. With minimal input and computational resources, the presented method can estimate the outflow rates reliably. This study constitutes the largest validation of a reduced-order flow model against 3D simulations. The impact of the energy loss at the bifurcations is here demonstrated for cerebrovasculatures but can be applied to other physiological networks.

Narrowing the Expertise Gap for Predicting Intracranial Aneurysm Hemodynamics: Impact of Solver Numerics versus Mesh and Time-Step Resolution.

BACKGROUND AND PURPOSE: Recent high-resolution computational fluid dynamics studies have uncovered the presence of laminar flow instabilities and possible transitional or turbulent flow in some intracranial aneurysms. The purpose of this study was to elucidate requirements for computational fluid dynamics to detect these complex flows, and, in particular, to discriminate the impact of solver numerics versus mesh and time-step resolution. MATERIALS AND METHODS: We focused on 3 MCA aneurysms, exemplifying highly unstable, mildly unstable, or stable flow phenotypes, respectively. For each, the number of mesh elements was varied by 320× and the number of time-steps by 25×. Computational fluid dynamics simulations were performed by using an optimized second-order, minimally dissipative solver, and a more typical first-order, stabilized solver. RESULTS: With the optimized solver and settings, qualitative differences in flow and wall shear stress patterns were negligible for models down to ∼800,000 tetrahedra and ∼5000 time-steps per cardiac cycle and could be solved within clinically acceptable timeframes. At the same model resolutions, however, the stabilized solver had poorer accuracy and completely suppressed flow instabilities for the 2 unstable flow cases. These findings were verified by using the popular commercial computational fluid dynamics solver, Fluent. CONCLUSIONS: Solver numerics must be considered at least as important as mesh and time-step resolution in determining the quality of aneurysm computational fluid dynamics simulations. Proper computational fluid dynamics verification studies, and not just superficial grid refinements, are therefore required to avoid overlooking potentially clinically and biologically relevant flow features.

The Computational Fluid Dynamics Rupture Challenge 2013­Phase I: prediction of rupture status in intracranial aneurysms.

BACKGROUND AND PURPOSE: Rupture risk assessment for intracranial aneurysms remains challenging, and risk factors, including wall shear stress, are discussed controversially. The primary purpose of the presented challenge was to determine how consistently aneurysm rupture status and rupture site could be identified on the basis of computational fluid dynamics. MATERIALS AND METHODS: Two geometrically similar MCA aneurysms were selected, 1 ruptured, 1 unruptured. Participating computational fluid dynamics groups were blinded as to which case was ruptured. Participants were provided with digitally segmented lumen geometries and, for this phase of the challenge, were free to choose their own flow rates, blood rheologies, and so forth. Participants were asked to report which case had ruptured and the likely site of rupture. In parallel, lumen geometries were provided to a group of neurosurgeons for their predictions of rupture status and site. RESULTS: Of 26 participating computational fluid dynamics groups, 21 (81%) correctly identified the ruptured case. Although the known rupture site was associated with low and oscillatory wall shear stress, most groups identified other sites, some of which also experienced low and oscillatory shear. Of the 43 participating neurosurgeons, 39 (91%) identified the ruptured case. None correctly identified the rupture site. CONCLUSIONS: Geometric or hemodynamic considerations favor identification of rupture status; however, retrospective identification of the rupture site remains a challenge for both engineers and clinicians. A more precise understanding of the hemodynamic factors involved in aneurysm wall pathology is likely required for computational fluid dynamics to add value to current clinical decision-making regarding rupture risk.

Mind the gap: impact of computational fluid dynamics solution strategy on prediction of intracranial aneurysm hemodynamics and rupture status indicators.

BACKGROUND AND PURPOSE: Computational fluid dynamics has become a popular tool for studying intracranial aneurysm hemodynamics, demonstrating success for retrospectively discriminating rupture status; however, recent highly refined simulations suggest potential deficiencies in solution strategies normally used in the aneurysm computational fluid dynamics literature. The purpose of the present study was to determine the impact of this gap. MATERIALS AND METHODS: Pulsatile flow in 12 realistic MCA aneurysms was simulated by using both high-resolution and normal-resolution strategies. Velocity fields were compared at selected instants via domain-averaged error. We also compared wall shear stress fields and various reduced hemodynamic indices: cycle-averaged mean and maximum wall shear stress, oscillatory shear index, low shear area, viscous dissipation ratio, and kinetic energy ratio. RESULTS: Instantaneous differences in flow and wall shear stress patterns were appreciable, especially for bifurcation aneurysms. Linear regressions revealed strong correlations (R(2) > 0.9) between high-resolution and normal-resolution solutions for all indices except kinetic energy ratio (R(2) = 0.25) and oscillatory shear index (R(2) = 0.23); however, for most indices, the slopes were significantly <1, reflecting a pronounced underestimation by the normal-resolution simulations. Some high-resolution simulations were highly unstable, with fluctuating wall shear stresses reflected by the poor oscillatory shear index correlation. CONCLUSIONS: Typical computational fluid dynamics solution strategies may ultimately be adequate for augmenting rupture risk assessment on the basis of certain highly reduced indices; however, they cannot be relied on for predicting the magnitude and character of the complex biomechanical stimuli to which the aneurysm wall may be exposed. This impact of the computational fluid dynamics solution strategy is likely greater than that for other modeling assumptions or uncertainties.

Comparison of carotid plaque ulcer detection using contrast-enhanced and time-of-flight MRA techniques.

BACKGROUND AND PURPOSE: Ulceration in carotid plaque is a risk indicator for ischemic stroke. Our aim was to compare plaque ulcer detection by standard TOF and CE-MRA techniques and to identify factors that influence its detection. MATERIALS AND METHODS: Carotid MR imaging scans were acquired on 2066 participants in the ARIC study. We studied the 600 thickest plaques. TOF-MRA, CE-MRA, and black-blood MR images were analyzed together to define ulcer presence (plaque surface niche ≥2 mm in depth). Sixty ulcerated arteries were detected. These arteries were randomly assigned, along with 40 nonulcerated plaques from the remaining 540, for evaluation of ulcer presence by 2 neuroradiologists. Associations between ulcer detection and ulcer characteristics, including orientation, location, and size, were determined and explored by CFD modeling. RESULTS: One CE-MRA and 3 TOF-MRAs were noninterpretable and excluded. Of 71 ulcers in 56 arteries, readers detected an average of 39 (55%) on both TOF-MRA and CE-MRA, 26.5 (37.5%) only on CE-MRA, and 1 (1.5%) only on TOF-MRA, missing 4.5 (6%) ulcers by both methods. Ulcer detection by TOF-MRA was associated with its orientation (distally pointing versus perpendicular: OR = 5.57 [95% CI, 1.08-28.65]; proximally pointing versus perpendicular: OR = 0.21 [95% CI, 0.14-0.29]); location relative to point of maximum stenosis (distal versus isolevel: OR = 5.17 [95% CI, 2.10-12.70]); and neck-to-depth ratio (OR = 1.96 [95% CI, 1.11-3.45]) after controlling for stenosis and ulcer volume. CONCLUSIONS: CE-MRA detects more ulcers than TOF-MRA in carotid plaques. Missed ulcers on TOF-MRA are influenced by ulcer orientation, location relative to point of maximum stenosis, and neck-to-depth ratio.

An objective approach to digital removal of saccular aneurysms: technique and applications.

Human studies of haemodynamic factors in the pathogenesis of cerebral aneurysms require knowledge of the pre-aneurysmal vasculature. This paper presents an objective and automated technique to digitally remove an aneurysm and reconstruct the parent artery, based on lumen geometries segmented from angiographic images. Relying on robust computational geometry concepts, notably Voronoi diagrams of the digitised lumen surface, the aneurysm attachment region is first defined objectively using lumen centrelines. Centrelines within this region are replaced by smooth interpolations, which then guide the interpolation of Voronoi points within the attachment region. Combined with Voronoi points from outside the attachment region, the parent artery lumen, without the aneurysm, can be reconstructed. Plausible reconstructions were obtained, automatically, for a set of 10 side-wall or terminal aneurysms, of various sizes and shapes, from the ANEURISK project data set. Application of image-based computational fluid dynamics analysis to a five side-wall aneurysm cases data set revealed an association between the recently proposed gradient oscillatory number (GON) and the site of aneurysm formation in four of five cases; however, elevated GON was also evident at non-aneurysmal sites. A potential application to the automated delineation of aneurysms for morphological characterisations is also suggested. The proposed approach may serve as a broad platform for investigating haemodynamic and morphological factors in aneurysm initiation, rupture and therapy in a way amenable to large-scale clinical studies or routine clinical use. Nevertheless, while the parent artery reconstructions are plausible, it remains to be proven that they are faithful representations of the pre-aneurysmal artery.

On the overestimation of early wall thickening at the carotid bulb by black blood MRI, with implications for coronary and vulnerable plaque imaging.

Black blood MRI is an attractive tool for monitoring normal and pathological wall thickening; however, limited spatial resolutions can conspire with complex vascular geometries to distort the appearance of the wall in ways hitherto unclear. To elucidate this, a thin-walled cylinder model was developed to predict the composite effects of obliqueness, in-plane resolution and voxel anisotropy on the accuracy of MRI-derived wall thickness measurements. These predictions were validated by means of imaging of a thin-walled carotid bifurcation phantom. Typical thick-slice axial acquisitions were found to result in artifactual wall thickening at the carotid bulb, owing to its obliqueness to the nominal imaging plane. Obliqueness was less problematic for near-isotropic resolutions; however, the obligatory reduction of in-plane resolution served to inflate wall thicknesses uniformly by up to 50%. Moreover, the nonlinear relationship between wall thickness and its overestimation served to mask genuine differences in wall thickness, an effect predicted to be worse for thinner coronary artery walls and plaque caps. Therefore, care must be taken when interpreting black blood MRI wall thickness measurements in the presence-or absence-of observed differences within or between individuals.

Two-equation turbulence modeling of pulsatile flow in a stenosed tube.

The study of pulsatile flow in stenosed vessels is of particular importance because of its significance in relation to blood flow in human pathophysiology. To date, however, there have been few comprehensive publications detailing systematic numerical simulations of turbulent pulsatile flow through stenotic tubes evaluated against comparable experiments. In this paper, two-equation turbulence modeling has been explored for sinusoidally pulsatile flow in 75% and 90% area reduction stenosed vessels, which undergoes a transition from laminar to turbulent flow as well as relaminarization. Wilcox's standard k-omega model and a transitional variant of the same model are employed for the numerical simulations. Steady flow through the stenosed tubes was considered first to establish the grid resolution and the correct inlet conditions on the basis of comprehensive comparisons of the detailed velocity and turbulence fields to experimental data. Inlet conditions based on Womersley flow were imposed at the inlet for all pulsatile cases and the results were compared to experimental data from the literature. In general, the transitional version of the k-omega model is shown to give a better overall representation of both steady and pulsatile flow. The standard model consistently over predicts turbulence at and downstream of the stenosis, which leads to premature recovery of the flow. While the transitional model often under-predicts the magnitude of the turbulence, the trends are well-described and the velocity field is superior to that predicted using the standard model. On the basis of this study, there appears to be some promise for simulating physiological pulsatile flows using a relatively simple two-equation turbulence model.

Computational modeling of arterial biomechanics: insights into pathogenesis and treatment of vascular disease.

We review how advances in computational techniques are improving our understanding of the biomechanical behavior of the healthy and diseased cardiovascular system. Numerical modeling of biomechanics is being used in a wide variety of ways, including assessment of effects of mural and hemodynamically induced stresses on atherogenesis, development of risk measures for aneurysm rupture, improvement in interpretation of medical images, and quantification of oxygen transport in diseased and healthy arteries. Although not amenable to routine clinical use, numerical modeling of cardiovascular biomechanics is a powerful research tool.

Effect of distal graft anastomosis site on retrograde perfusion and flow patterns of native coronary vasculature.

BACKGROUND: To select the site of a target vessel for distal anastomosis surgeons use different approaches. Some try to place the graft as close to the stenosis as possible, whereas others routinely anastomose the graft onto the distal portion. In this latter case the proximal portion and its tributaries are perfused from the graft in a retrograde rather than an antegrade fashion. The aim of this study was to investigate the effect of local hemodynamics associated with the different location of distal anastomoses on flow patterns in the proximal native artery and its branches. METHODS: Computational fluid dynamic and in vitro model studies were carried out in a control model composed of a straight tube (host) with a 45E side branch and models in which the proximal end of the host had various degrees of stenosis; a 45E end-to-side "graft" anastomosis was introduced either proximal (upstream) or distal (downstream) to the branch. RESULTS: Placing the graft proximal to the branch largely preserved the flow patterns that were seen in the control model. Placing the graft distal to the branch, however, introduced an extensive region of relatively stagnant flow in the native vessel near the branch. Such regions are known to promote thrombus formation that could ultimately lead to occlusion of the retrograde portion of the host vessel. CONCLUSIONS: This study suggests that, although often less convenient surgically, long-term outcome of coronary artery bypass grafting may be improved by placing grafts in the most proximal portion of the native vessel, as close to the occlusion or stenosis as possible for better preservation of a proximal artery and its branches.

Effect of black blood MR image quality on vessel wall segmentation.

Black blood MRI has become a popular technique for measuring arterial wall area as an indicator of plaque size. Computer-assisted techniques for segmenting vessel boundaries have been developed to increase measurement precision. In this study, the carotid arteries of four normal subjects were imaged at seven different fields of view (FOVs), keeping all other imaging parameters fixed, to determine whether spatial resolution could be increased at the expense of image quality without sacrificing precision. Wall areas were measured via computer-assisted segmentation of the vessel boundaries performed repeatedly by two operators. Analysis of variance (ANOVA) demonstrated that the variability of wall area measurements was below 1.5 mm(2) for in-plane spatial resolutions between 0.22 mm and 0.37 mm. An inverse relationship between operator variability and the signal difference-to-noise ratio (SDNR) demonstrated that semi-automatic segmentation of the wall boundaries was robust for SDNR >3, defining a criterion above which subjective image quality can be degraded without an appreciable loss of information content. Our study also suggested that spatial resolutions higher than 0.3 mm may be required to quantify normal wall areas to within 10% accuracy, but that the reduced SNR associated with the higher resolution may be tolerated by semi-automated wall segmentation without an appreciable loss of precision.

A semi-automatic technique for measurement of arterial wall from black blood MRI.

Black blood magnetic resonance imaging (MRI) has become a popular technique for imaging the artery wall in vivo. Its noninvasiveness and high resolution make it ideal for studying the progression of early atherosclerosis in normal volunteers or asymptomatic patients with mild disease. However, the operator variability inherent in the manual measurement of vessel wall area from MR images hinders the reliable detection of relatively small changes in the artery wall over time. In this paper we present a semi-automatic method for segmenting the inner and outer boundary of the artery wall, and evaluate its operator variability using analysis of variance (ANOVA). In our approach, a discrete dynamic contour is approximately initialized by an operator, deformed to the inner boundary, dilated, and then deformed to the outer boundary. A group of four operators performed repeated measurements on 12 images from normal human subjects using both our semiautomatic technique and a manual approach. Results from the ANOVA indicate that the inter-operator standard error of measurement (SEM) of total wall area decreased from 3.254 mm2 (manual) to 1.293 mm2 (semi-automatic), and the intra-operator SEM decreased from 3.005 mm2 to 0.958 mm2. Operator reliability coefficients increased from less than 69% to more than 91% (inter-operator) and 95% (intra-operator). The minimum detectable change in wall area improved from more than 8.32 mm2 (intra-operator, manual) to less than 3.59 mm2 (inter-operator, semi-automatic), suggesting that it is better to have multiple operators measure wall area with our semi-automatic technique than to have a single operator make repeated measurements manually. Similar improvements in wall thickness and lumen radius measurements were also recorded. Since the semi-automatic technique has effectively ruled out the effect of the operator on these measurements, it may be possible to use such techniques to expand prospective studies of atherogenesis to multiple centers so as to increase access to real patient data without sacrificing reliability.

Effect of contralateral carotid artery stenosis on carotid ultrasound velocity measurements.

BACKGROUND AND PURPOSE: Carotid ultrasonography is being increasingly performed as the sole investigation to assess internal carotid artery (ICA) stenosis. A potential source of error in using ultrasound peak systolic velocity (PSV) measurements is that the redistribution of blood flow due to severe stenosis in a contralateral carotid artery may lead to artificially elevated values. METHODS: Ultrasonography was performed before and after carotid endarterectomy in symptomatic patients who participated in the North American Symptomatic Carotid Endarterectomy Trial (NASCET). The mean change in PSV in the unoperated artery was assessed across all degrees of angiographically defined stenosis. A simple theoretical resistance model of the cerebral circulation was also derived. RESULTS: Complete bilateral ultrasound examinations were performed within 90 days of the initial scan in 386 patients. In the presence of a contralateral severe (70% to 99%) ICA stenosis, the PSV in the unoperated artery was artificially elevated by a mean of 84 cm/s (P:=0.03; 95% CI, 10 to 159 cm/s). The mean elevation was less pronounced for lesser degrees of stenosis (11 to 21 cm/s). Small elevations (3 to 12 cm/s) were observed when the contralateral artery had <70% stenosis. The patterns of observed results were congruent with those from the theoretical model. CONCLUSIONS: The present study showed that a severely stenosed contralateral ICA can artificially elevate ultrasound PSV. Since the effect was greatest when bilaterally severe stenoses were present, caution must be exercised when assessing the degree of ICA stenosis on the basis of ultrasonography PSV measurements alone.