Reproducibility study of 3D geometrical reconstruction of the human carotid bifurcation from magnetic resonance images

Inter-operator Reliability of Magnetic Resonance Image-Based Computational Fluid Dynamics Prediction of Cerebrospinal Fluid Motion in the Cervical Spine

For the first time, inter-operator dependence of MRI based computational fluid dynamics (CFD) modeling of cerebrospinal fluid (CSF) in the cervical spinal subarachnoid space (SSS) is evaluated. In vivo MRI flow measurements and anatomy MRI images were obtained at the cervico-medullary junction of a healthy subject and a Chiari I malformation patient. 3D anatomies of the SSS were reconstructed by manual segmentation by four independent operators for both cases. CFD results were compared at nine axial locations along the SSS in terms of hydrodynamic and geometric parameters. Intraclass correlation (ICC) assessed the inter-operator agreement for each parameter over the axial locations and coefficient of variance (CV) compared the percentage of variance for each parameter between the operators. Greater operator dependence was found for the patient (0.19 < ICC < 0.99) near the craniovertebral junction compared to the healthy subject (ICC > 0.78). For the healthy subject, hydraulic diameter and Womersley number had the least variance (CV = ~2%). For the patient, peak diastolic velocity and Reynolds number had the smallest variance (CV = ~3%). These results show a high degree of inter-operator reliability for MRI-based CFD simulations of CSF flow in the cervical spine for healthy subjects and a lower degree of reliability for patients with Type I Chiari malformation.

Scan-Rescan reproducibility of carotid bifurcation geometry from routine contrast-enhanced MR angiography

PURPOSE

To demonstrate the feasibility of rapid and reliable geometric characterization of normal carotid bifurcation geometry from routine 3D contrast-enhanced magnetic resonance (MR) angiograms.

MATERIALS AND METHODS

Repeat scans of 61 participants, acquired as part of the Atherosclerosis Risk in Communities (ARIC) Carotid MRI substudy, were digitally segmented using automated 3D level set methods, relying on an operator only to select the branch endpoints and thresholds for the 3D lumen surface initialization. Geometric factors characterizing the 3D lumen geometry were then extracted automatically.

RESULTS

Of 122 scans, 117 could be segmented within 5 minutes each, with 40% being of sufficiently high quality to require less than 2 minutes each. Irrespective of scan quality, geometric factors were found to be highly reproducible, with intraclass correlation coefficients (ICCs) typically above 0.9. The reconstructed lumen surfaces were reproducible to <0.3 mm on average, comparable to previous MRI-based reproducibility studies. Owing to the automated nature of the analysis, operator reliability was near-perfect (ICC >0.99), with lumen surface differences <0.1 mm.

CONCLUSION

The 3D geometry of the carotid bifurcation can be characterized rapidly and with a high degree of consistency, even for suboptimal image qualities. This bodes well for large-scale retrospective or prospective studies aimed at teasing out the influence of local vs. systemic risk factors for early atherosclerosis.

Effect of posture change on the geometric features of the healthy carotid bifurcation

Segmented cross-sectional MRI images were used to construct 3-D virtual models of the carotid bifurcation in ten healthy volunteers. Geometric features, such as bifurcation angle, internal carotid artery (ICA) angle, planarity angle, asymmetry angle, tortuosity, curvature, bifurcation area ratio, ICA/common carotid artery (CCA), external carotid artery (ECA)/CCA, and ECA/ICA diameter ratios, were calculated for both carotids in two head postures: 1) the supine neutral position; and 2) the prone sleeping position with head rotation to the right ( ∼ 80°). The results obtained have shown that head rotation causes 1) significant variations in bifurcation angle [32% mean increase for the right carotid (RC) and 21% mean decrease for the left carotid (LC)] and internal carotid artery angle (97% mean increase for the RC, 43% mean decrease for the LC); 2) a slight increase in planarity and asymmetry angles for both RC and LC; 3) minor and variable curvature changes for the CCA and for the branches; 4) slight tortuosity changes for the braches but not for the CCA; and 5) unsubstantial alterations in area and diameter ratios (percentage changes %). The significant geometric changes observed in most subjects with head posture may also cause significant changes in bifurcation hemodynamics and warrant future investigation of the hemodynamic parameters related to the development of atherosclerotic disease such as low oscillating wall shear stress and particle residence times.

Time-efficient patient-specific quantification of regional carotid artery fluid dynamics and spatial correlation with plaque burden

PURPOSE

Low wall shear stress (WSS) and high oscillatory shear index (OSI) influence plaque formation, yet little is known about their role in progression/regression of established plaques because of lack of practical means to calculate them in individual patients. Our aim was to use computational fluid dynamics (CFD) models of patients with carotid plaque undergoing statin treatment to calculate WSS and OSI in a time-efficient manner, and determine their relationship to plaque thickness (PT), plaque composition (PC), and regression.

METHODS

Eight patients (68 +/- 9 yr, one female) underwent multicontrast 3 T MRI at baseline and six-month post statin treatment. PT and PC were measured in carotid segments (common-CC, bifurcation-B, internal-IC) and circumferentially in nonoverlapping 600 angles and correlated with CFD models created from MRI, ultrasound, and blood pressure.

RESULTS

PT was highest in B (2.42 +/- 0.98 versus CC: 1.60 +/- 0.47, IC: 1.62 +/- 0.52 mm, p < 0.01). Circumferentially, plaque was greatest opposite the flow divider (p < 0.01), where the lowest WSS and highest OSI were observed. In B and IC, PT was inversely related to WSS (R = -0.28 and -0.37, p < 0.01) and directly related to OSI (R = 0.22 and 0.52, p < 0.05). The total plaque volume changed from 1140 +/- 437 to 974 +/- 587 mm3 at six months (p = 0.1). Baseline WSS, but not OSI, correlated with changes in PT, necrotic tissue, and hemorrhage in B and IC, but not CC. CFD modeling took 49 +/- 18 h per patient.

CONCLUSIONS

PT and PC correspond to adverse WSS and OSI in B and IC, and WSS is modestly but significantly related to changes in PT after short-term statin treatment. Regional hemodynamics from CFD can feasibly augment routine clinical imaging for comprehensive plaque evaluation.

Study of reproducibility of human arterial plaque reconstruction and its effects on stress analysis based on multispectral in vivo magnetic resonance imaging

PURPOSE

To quantify the uncertainties of carotid plaque morphology reconstruction based on patient-specific multispectral in vivo magnetic resonance imaging (MRI) and their impacts on the plaque stress analysis.

MATERIALS AND METHODS

In this study, three independent investigators were invited to reconstruct the carotid bifurcation with plaque based on MR images from two subjects to study the geometry reconstruction reproducibility. Finite element stress analyses were performed on the carotid bifurcations, as well as the models with artificially modified plaque geometries to mimic the image segmentation uncertainties, to study the impacts of the uncertainties to the stress prediction.

RESULTS

Plaque reconstruction reproducibility was generally high in the study. The uncertainties among interobservers are around one or the subpixel level. It also shows that the predicted stress is relatively less sensitive to the arterial wall segmentation uncertainties, and more affected by the accuracy of lipid region definition. For a model with lipid core region artificially increased by adding one pixel on the lipid region boundary, it will significantly increase the maximum Von Mises Stress in fibrous cap (>100%) compared with the baseline model for all subjects.

CONCLUSION

The current in vivo MRI in the carotid plaque could provide useful and reliable information for plaque morphology. The accuracy of stress analysis based on plaque geometry is subject to MRI quality. The improved resolution/quality in plaque imaging with newly developed MRI protocols would generate more realistic stress predictions.

Region-based geometric modelling of human airways and arterial vessels

Anatomically precise geometric models of human airways and arterial vessels play a critical role in the analysis of air and blood flows in human bodies. The established geometric modelling methods become invalid when the model consists of bronchioles or small vessels. This paper presents a new method for reconstructing the entire airway tree and carotid vessels from point clouds obtained from CT or MR images. A novel layer-by-layer searching algorithm has been developed to recognize branches of the airway tree and arterial vessels from the point clouds. Instead of applying uniform accuracy to all branches regardless of the number of available points, the surface patches on each branch are constructed adaptively based on the number of available elemental points, which leads to the elimination of distortions occurring at small bronchi and vessels.

Patient-specific modeling of cardiovascular mechanics

Advances in numerical methods and three-dimensional imaging techniques have enabled the quantification of cardiovascular mechanics in subject-specific anatomic and physiologic models. Patient-specific models are being used to guide cell culture and animal experiments and test hypotheses related to the role of biomechanical factors in vascular diseases. Furthermore, biomechanical models based on noninvasive medical imaging could provide invaluable data on the in vivo service environment where cardiovascular devices are employed and on the effect of the devices on physiologic function. Finally, patient-specific modeling has enabled an entirely new application of cardiovascular mechanics, namely predicting outcomes of alternate therapeutic interventions for individual patients. We review methods to create anatomic and physiologic models, obtain properties, assign boundary conditions, and solve the equations governing blood flow and vessel wall dynamics. Applications of patient-specific models of cardiovascular mechanics are presented, followed by a discussion of the challenges and opportunities that lie ahead.

Generating an ex vivo vascular model

Realistic ex vivo anthropometric vascular environments are required for endovascular device optimization and for preclinical evaluation of interventional procedures. The objective of this research is to build an anthropomorphic model of the human carotid artery. The combination of magnetic resonance angiography image processing and computer-aided design and manufacturing techniques allowed fabrication of multicomponent morphologically precise casts of the carotid artery. The lost core technique was used to produce a hollow vessel prototype incorporating polyvinyl alcohol cryogel (PVA-C) as a tissue-mimicking vessel wall material. PVA-C was mechanically characterized by uniaxial tensile testing after different numbers of freeze/thaw cycles. The novel model construction approach outlined in this study accounts for the morphologic complexities of the human vasculature, and proved successful for the production of realistic compliant ex vivo arterial model.

Effect of vessel curvature on Doppler derived velocity profiles and fluid flow

Side-branches and curvatures in the arterial tree yield deviations from the axial oriented velocity. Velocity or volume flow estimates based on the assumption that flow is axially oriented are of limited value at these sites. This article evaluates information obtainable by using a multigate Doppler ultrasound (US) instrument used with curved phantoms, which resemble the human coronary arteries. The comparison of experimental velocity data with data provided by an accurate computational fluid dynamics (CFD) method shows differences in the range of 4 to 11% for four curvatures with different radii. Multigate data are also used to estimate the volume flow in the curved segments at different experimental conditions. An error lower than 15% is obtained, to be compared with a 24% error obtained by assuming a parabolic velocity profile. In particular, it is shown that the residual error is not related to the small deviation of the velocity vectors from the axial direction due to the presence of secondary velocity components, which are found to be of magnitude less than 10% with respect to the axial velocity component.

Operator dependence of 3-D ultrasound-based computational fluid dynamics for the carotid bifurcation

The association between vascular wall shear stress (WSS) and the local development of atherosclerotic plaque makes estimation of in vivo WSS of considerable interest. Three-dimensional ultrasound (3DUS) combined with computational fluid dynamics (CFD) provides a potentially valuable tool for acquiring subject-specific WSS, but the interoperator and intraoperator variability associated with WSS calculations using this method is not known. Here, the accuracy, reproducibility and operator dependence of 3DUS-based computational fluid dynamics were examined through a phantom and in vivo studies. A carotid phantom was scanned and reconstructed by two operators. In the in vivo study, four operators scanned a healthy subject a total of 11 times, and their scan data were processed by three individuals. The study showed that with some basic training, operators could acquire accurate carotid geometry for flow reconstructions. The variability of measured cross-sectional area and predicted shear stress was 8.17% and 0.193 N/m2 respectively for the in vivo study. It was shown that the variability of the examined parameters was more dependent on the scan operators than the image processing operator. The range of variability of geometrical and flow parameters reported here can be used as a reference for future in vivo studies using the 3DUS-based CFD approach.

Influence of head position on carotid hemodynamics in young adults

Studies in adults have shown marked changes in geometry and relative positions of the carotid arteries when rotating the head. The aim of this study was to quantify the change in geometry and analyze its effect on carotid hemodynamics as a result of head rotation. The right carotid arteries of nine young adult subjects were investigated in supine position with straight and left turned head positions, respectively. The three-dimensional (3D) carotid geometry was reconstructed by using 3D ultrasound (3D US), and the carotid hemodynamics were calculated by combining 3D US with computational fluid dynamics. It was observed that cross-sectional areas and shapes did not change markedly with head rotation, but carotid vessel center lines altered with planarification of the common carotid artery as a main feature ( P < 0.05). Measured common carotid flow rates changed significantly at the individual level when the head was turned, but on the average, the change in mean common carotid flow rate was relatively small (0.37 ± 1.11 ml/s). The effect of the altered center lines and flow rates on the atherogenic nature of the carotid bifurcation was evaluated by using calculated hemodynamic wall parameters, such as wall shear stress (WSS) and oscillatory shear index (OSI). It was found that WSS and OSI patterns changed significantly with head rotation, but the variations were very subject dependent and could not have been predicted without assessing the altered geometry and flow of the carotid bifurcation for individual cases. This study suggests that there is a need for standardization of the choice of head position in the 3D US scan protocol, and that carotid stents and emboli diverters should be studied in different head positions.

Robust and objective decomposition and mapping of bifurcating vessels

Computational modeling of human arteries has been broadly employed to investigate the relationships between geometry, hemodynamics and vascular disease. Recent developments in modeling techniques have made it possible to perform such analyses on realistic geometries acquired noninvasively and, thus, have opened up the possibility to extend the investigation to populations of subjects. However, for this to be feasible, novel methods for the comparison of the data obtained from large numbers of realistic models in the presence of anatomic variability must be developed. In this paper, we present an automatic technique for the objective comparison of distributions of geometric and hemodynamic quantities over the surface of bifurcating vessels. The method is based on centerlines and consists of robustly decomposing the surface into its constituent branches and mapping each branch onto a template parametric plane. The application of the technique to realistic data demonstrates how similar results are obtained over similar geometries, allowing for proper model-to-model comparison. Thanks to the computational and differential geometry criteria adopted, the method does not depend on user-defined parameters or user interaction, it is flexible with respect to the bifurcation geometry and it is readily extendible to more complex configurations of interconnecting vessels.

Carotid geometry reconstruction: a comparison between MRI and ultrasound

Image-based Computational Fluid Dynamics (CFD) has become a popular tool for the prediction of in vivo flow profiles and hemodynamic wall parameters. Currently, Magnetic Resonance Imaging (MRI) is most widely used for in vivo geometry acquisition. For superficial arteries such as the carotids and the femoral artery, three-dimensional (3-D) extravascular ultrasound (3-DUS) could be a cost-effective alternative to MRI. In this study, nine healthy subjects were scanned both with MRI and 3-DUS. The reconstructed carotid artery geometries for each subject were compared by evaluating cross-sectional areas, centerlines, and carotid nonplanarity. Lumen areas agreed very well between the two different acquisition techniques, whereas centerlines and nonplanarity parameters showed measurable disagreement, possibly due to the different neck and head positions adopted for 3-DUS versus MRI. With the current level of agreement achieved, 3-DUS has the potential to become an inexpensive and fast alternative to MRI for image-based CFD modeling of superficial arteries.