The flow of our study is shown in Fig. 1.
Patient-specific aneurysm phantom
A patient-specific aneurysm that we had previously developed and used for validating an Enterprise stent via an FVST was selected for use in the present study [11]. Digital images (Fig. 1a) acquired from three-dimensional (3D) rotational angiography were collected to create 3D reconstruction of the aneurysm and then saved to a stereolithography format file (Fig. 1b). Some inessential small arteries which were reconstructed relatively difficult and inaccurate, such as ophthalmic artery and posterior communicating artery, were removed to ensure the computational efficiency of CFD, following which a patient-specific hollow phantom was fabricated using a 3D printer (Materialise, Lueven, Belgium) (Fig. 1g). The aneurysm was wide-necked (width 12.27 mm, height 9.35 mm) and located at the C6 segment of the internal carotid artery. Of note, this phantom model is consistent with the vascular anatomy of the patient, not only regarding aneurysmal morphology but also vascular curvature.
Deployment of stent: Real and simulated LVIS respectively
After fabrication of the specific model, a 4.5 × 23 mm LVIS stent, whose dimensions were dictated by the vascular vessel and aneurysm size, was ready for endovascular placement. The stent was released successfully by an experienced neurosurgeon and his assistant according to operative standards (Fig. 1h). The phantom model deployed with LVIS was then scanned by a micro cone-beam computed tomography (micro-CT) system with high resolution (8.8 μm). The implanted stent was reconstructed manually to generate a real LVIS (R-LVIS) stereolithography file using Mimics 10.01 (Materialise, Leuven, Belgium) (Fig. 1i).
The F-LVIS constructed is similar to that of the Enterprise stent that we previously reported [11, 12]. The flow of work to construct the F-LVIS is based on the concept of simplex meshes generating a deformable structure that is finally swept into 3D stent wires. The work flow occurs mainly in three steps.
-
1.
Preprocessing: The aneurysm’s geometry is isolated from the parent artery using Geomagic Studio (Raindrop Geomagic, Research Triangle Park, NC, USA). Subsequently, a centerline suitable to the parent artery is extracted using Mimics 10.01 (Fig. 1c).
-
2.
Simplex mesh initiation and expansion: Simplex mesh representing the surface of LVIS based on the centerline is initiated along the parent vessel. The simplex mesh then undergoes radial expansion via MATLAB 2013 (MathWorks, Natick, MA, USA). The expansion stops when the deformable mesh is in optimal apposition with the vessel wall (Fig. 1d).
-
3.
LVIS pattern mapping: The expanded simplex mesh is then input into Abaqus/Explicit 6.12 (Simulia, Providence, RI, USA) to obtain wire curves using an in-house LVIS python code. Next, the wire curves are swept into 3D structures in the computer-aided designed program Creo Parametric 2.0 (PTC, Needham, MA, USA). Finally, the LVIS modeled by 16 overlaying wires is completed (Fig. 1e).
F-LVIS and R-LVIS were then merged with the patient’s vascular system to create computational models (Fig. 1f, g). CFD calculations were performed to analyze the hemodynamic variations.
CFD calculation and hemodynamic analysis
We performed CFD simulations in the F-LVIS and R-LVIS models in a manner similar to those described previously [13, 14]. ICEM software (version 14.5; ANSYS Inc., Canonsburg, PA, USA) was used to create finite-volume tetrahedral elements. After meshing, the hemodynamics was simulated with ANSYS CFX software (ANSYS CFX 14.0; ANSYS, Inc.). The Navier–Stokes formulation was solved to simulate fluid flow, with blood assumed to be a homogenous, laminar, incompressible Newtonian fluid. A measured viscosity (4 cPa, specific density 1060 kg/m3) was applied under rigid-walled, no-slip boundary conditions. Zero pressure was set at the outlet, and a representative pulsatile period velocity profile obtained by transcranial Doppler imaging was implemented as the inflow boundary condition. To confirm the numerical stability, we performed two cardiac cycle simulations. The results of the final cardiac cycle was output to underlie hemodynamic analyses.
When CFD calculations were completed, the following meaningful parameters were collected using ANSYS CFD-post software (ANSYS CFX 14.0; ANSYS, Inc.): velocity streamlines, vectors of the flow velocity, wall shear stress (WSS). Qualitative and quantitative results of the hemodynamics were compared between F-LVIS and R-LVIS.
Statistical analysis
Data analysis was performed with statistical software (SPSS V.19.0; IBM, Chicago, IL, USA). The one-sample Kolmogorov–Smirnov test was used to test normal-distribution data for continuous parameters, and the paired-samples t-test was used when the data were approximately normally distributed. Continuous variables were presented as medians (interquartile range). Two-tailed p < 0.05 was considered to indicate statistical significance.