Article Text

Original research
How to perform intra-aneurysmal coil embolization after Pipeline deployment: a study from a hemodynamic viewpoint
  1. Mingqi Zhang1,
  2. Zhongbin Tian2,
  3. Yisen Zhang1,
  4. Ying Zhang1,
  5. Kun Wang1,
  6. Xiaochang Leng3,
  7. Xinjian Yang1,
  8. Jianping Xiang3,
  9. Jian Liu1
  1. 1 Department of Interventional Neuroradiology, Beijing Neurosurgical Institute and Beijing Tiantan Hospital, Capital Medical University, Beijing, China
  2. 2 Department of Neurosurgery, Zhujiang Hospital, Southern Medical University, Guangzhou, Guangdong, China
  3. 3 ArteryFlow Technology, Hangzhou, China
  1. Correspondence to Dr Jian Liu, Department of Interventional Neuroradiology, Beijing Neurosurgical Institute and Beijing Tiantan Hospital, Capital Medical University, Beijing 100070, China; jianliu_ns{at}; Dr Jianping Xiang; jianping.xiang{at}


Background Pipeline embolization device (PED) deployment combined with coil therapy for large complex intracranial aneurysms is effective and considered superior to PED deployment alone. However, the optimal strategy for use of coils remains unclear. We used patient-specific aneurysm models and finite element analysis to determine the ideal packing density of coils after PED placement.

Methods Finite element analysis was used to provide a higher-fidelity model for accurate post-treatment computational fluid dynamics analysis to simulate the real therapeutic process of PED and all coils. We then calculated and analyzed the reduction ratio of velocity to identify the hemodynamic change during PED deployment and each coil embolization.

Results Sixteen consecutive patients underwent PED plus coil procedures to treat internal carotid artery intracranial aneurysms. After PED deployment, the intra-aneurysmal flow velocity significantly decreased (15.3 vs 10.0 cm/s; p<0.001). When the first coil was inserted, the flow velocity in the aneurysm further decreased and the reduction was significant (10.0 vs 5.3 cm/s; p<0.001). Analysis of covariance showed that the effect of the reduction ratio of velocity of the second coil was significantly lower than that of the first coil (p<0.001)—that is, when the packing density increased to 7.06%, the addition of coils produced no further hemodynamic effect.

Conclusion Adjunct coiling could improve the post-PED hemodynamic environment in treated intracranial aneurysms. However, dense packing is not necessary because the intra-aneurysmal hemodynamics tend to stabilize as the packing density reaches an average of 7.06% or after insertion of the second coil.

  • aneurysm
  • coil
  • flow diverter

Data availability statement

Data are available upon reasonable request. Not applicable.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The Pipeline embolization device (PED), a dedicated flow diverter (FD), is a braided stent device that redirects the blood flow from the intracranial aneurysm (IA) sac into the parent artery. PED deployment is a new IA treatment strategy that mainly focuses on parent artery reconstruction by hemodynamic modification. The concept of aneurysm intervention has changed from endosac coil embolization to parent artery flow diversion. The IA occlusion rate at follow-up can reportedly exceed 90%. In the treatment of large and giant aneurysms, FDs show significant advantages over conventional coil embolization.1 2

Although coil embolization is not mandatory when a PED is used, increasingly more neurointerventionalists prefer PED deployment combined with some coils. PED deployment combined with coils has been shown to be more effective in aneurysm occlusion than PED deployment alone.3 4 Dense packing is needed when an aneurysm is treated with coiling or stent-assisted coiling. However, there is no guideline for the packing density (PD) when an aneurysm is treated with a PED, and loose packing is usually performed. From a hemodynamic viewpoint, more coils are unnecessary when there are no significant flow changes in the sac of an unruptured aneurysm. More importantly, more manipulation in the aneurysm sac leads to a higher risk of intraoperative rupture and higher treatment costs.

Therefore, we explored the hemodynamic changes of the coils after PED deployment to identify a balanced point of loose packing for this situation. We speculated that coils provide no further hemodynamic benefit when the PD achieves a threshold. In this study, we aimed to investigate the coil embolization strategy after using an FD.

For this purpose, we used patient-specific aneurysm models before and after treatment from 16 consecutive patients in our center. We then used finite element analysis (FEA) to provide a higher-fidelity model for accurate post-treatment computational fluid dynamics (CFD) analysis to simulate the real therapeutic process of the PED and all coils.5 The hemodynamic effect of each coil was further analyzed based on the premise that the PED had been implanted.


Patient population

This study was approved by the institutional review board of our hospital. Informed consent was obtained from each study patient or his or her relatives, and all data were collected anonymously. We prospectively included 45 representative patients consecutively treated at Beijing Tiantan Hospital from February 2019 to November 2019. The patient selection process is shown in the online supplemental material. Sixteen patients were diagnosed with internal carotid artery aneurysms by digital subtraction angiography, and all aneurysms were treated by a single PED combined with coil embolization (table 1). Deployment and apposition of the PED were satisfactory in all patients. The coils were embolized after PED deployment and the number of coils was determined by neurointerventionalists with >10 years of extensive treatment experience. Each coil was based on the size of the aneurysm to ensure selection of the appropriate coil type during the operation. The details of the endovascular procedure are shown in the online supplemental material.

Table 1

Clinical characteristics and demographics of study patients

Aneurysm, stents, and coil modeling

Patient-specific aneurysm morphologies were reconstructed and obtained from three-dimensional (3D) rotational angiography images. The 3D geometry surface was displayed, segmented, and smoothed using software (Geomagic Studio v.12.0; Geomagic, Research Triangle Park, North Carolina, USA), and the geometries were saved in standard tessellation language (stl) format.6

The PED and coils were modeled and simulated as previously reported.7 In this study, the finite element method (FEM) technique was applied to simulate the PED and coils. The stent deployment modeling was processed with FEM-based workflow using ABAQUS/Explicit v.6.14 (SIMULIA, Providence, Rhode Island, USA). The three steps in PED simulation were crimping, delivery, and deployment. First, the PED was virtually crimped and fit into a microcatheter mode, which was modeled by rebar-reinforced shell elements with polymer-like elastic material. A delivery pathway was then made for stent system deployment with several central points of the cross sections of the blood vessel. The PED within the microcatheter was then delivered along the pathway to the target region. The distal and proximal landing zones for the PED were determined according to the location at which the PED landed in the angiographic images during clinical treatment. After the PED system was in position, the PED began to be released virtually. The deployed PED was then swept into 3D solid geometry for CFD analysis.

After the PED deployment, according to the real coils used in the procedure, coils were also implemented using the general purpose FEM software ABAQUS 6.14 in Abaqus/Explicit mode.7 Two types of coils were used in the procedure: frame coils and helical coils. Briefly, the frame and helical coils were first created in MATLAB (MathWorks, Natick, Massachusetts, USA) and then imported into the FEM code ABAQUS to generate different types of coils. Next, the coils were packaged into a virtual microcatheter to be deployed into the aneurysm sac. The coils were swept to a 3D solid model using Abaqus/CAE with the real diameter of the coils. Finally, the surface-based aneurysm and vessel model with the 3D representation of the coils and the PED was subsequently used for the CFD analysis.

CFD simulation and hemodynamic analysis

In this study, according to the process of procedure, several CFD models were simulated for each patient-specific aneurysm model: a pretreatment model, a PED deployment model, and each step coil models after PED placement. CFD simulations were performed as described previously.8 9 The inflow boundary condition we used was a pulsatile period velocity profile obtained from transcranial Doppler imaging of a healthy adult. The outlet pressure conditions at outlet arteries were imposed to p=0 Pa. The surface geometry models (aneurysms merged with the PED and coils) were imported into mesh generation software (ICEM CFD v.14.0; ANSYS, Canonsburg, Pennsylvania, USA) to create finite-volume tetrahedral elements. The largest element was 0.2 mm, and the element size of the PED and coils was set to 0.02 mm for adequate representation of the geometry. After meshing, CFX v.14.0 software (ANSYS) was used to simulate the blood hemodynamics. The governing equations underlying the calculation were the Navier–Stokes equations under homogenous, incompressible, laminar, and Newtonian fluid conditions with a density of 1056 kg/m3 and a viscosity of 0.0035 n×s/m2. The blood vessel wall was assumed to be rigid. We mainly calculated the reduction ratio of velocity (RRV) to identify the hemodynamic changes during PED deployment and each coil embolization:

Embedded Image

Embedded Image

Embedded Image

In addition, we also calculated the inflow rate and energy loss for each PED and coil model. Details of the calculation methods are shown in the online supplemental material. The metal coverage (MC) of the PED at aneurysmal ostium was also calculated. MC was determined as the coverage area of metal at the aneurysmal ostium divided by the whole aneurysmal ostium area simulated by the finite element model, which is affected by vascular morphology and manipulation. The normal view refers to a view perpendicular to the aneurysmal ostium of the measured part and passing through the center line of the parent artery. From the normal view of the ostium, the MC area and the ostium area were measured. To avoid an error caused by the bending surface, we rotated the model around the vessel center line and measured the covered area. Details of the process have been given in a previous study.10 The PD was also calculated and determined as the ratio of the coil volume to the aneurysm sac volume.

Statistical analysis

Statistical analysis was performed using the SPSS 24.0 package (IBM Corp, Chicago, Illinois, USA) and the GraphPad Prism (v.8). We used PD as a correlated variable and performed analysis of covariance to examine any measured change in the RRV (coil) for each coil embolization. We also used a regression model to fit the relationship between MC and RRV (PED). The mean differences between continuous variables were analyzed using the paired sample t-test, and p values <0.05 were considered significant. Before applying the tests, normal distribution and homogeneity of variance were confirmed for each sample.


Patient characteristics

The mean age of the patients was 57.94 years (range 14–74 years) and 14 of the 16 patients (87.5%) were women. The mean aneurysm size was 14.6 mm (range 7–29 mm). All aneurysms were located at the internal carotid artery and were unruptured. Five patients experienced headaches, two patients experienced dizziness, and the remaining patients were asymptomatic. After the procedure, the mean PD was 12.6% and the mean MC was 21.6%. The number of coils and the PD for each case are listed in table 1.

Hemodynamics of PED

Before PED deployment, the mean intra-aneurysmal flow velocity was 15.3±7 cm/s. Fourteen of the 16 cases (87.5%) showed a concentrated inflow impingement into the aneurysm sac. After PED deployment, the intra-aneurysmal flow velocity was significantly reduced (15.3±7 vs 10.0±5 cm/s; p<0.001). The mean RRV (PED) after PED deployment was 34.63%±13.69% (see online supplemental table 1). Meanwhile, the impingement flow impact was obviously redirected in 12 cases (75%). A representative case is shown in figure 1. A correlative analysis was performed between RRV (PED) and MC. There was no correlation between RRV (PED) and MC (r=0.18, p=0.51) (figure 2A).

Figure 1

A representative case of the hemodynamic changes in an aneurysm after treatment with a PED and loosely packed coils. (A) Pretreatment 3D angiographic image. (B) Position and configuration of the PED in the parent artery. (C) Digital subtraction angiogram showing the shape of the coils. (D) Shape of the coils after simulation. (E) Velocity distribution in the aneurysm sac before treatment (E1) and after PED placement (E2). The velocity in the aneurysm gradually decreased during the process of packing the first to fifth coils (E3–E5). (F) Impingement of flow in the aneurysm before treatment (F1) and after PED placement (F2). The impingement of flow in the aneurysm sac obviously decreased during the process of packing the first to fifth coils (F3–F5). C1, after the first coil embolization, C2, after the second coil embolization, etc. PED, Pipeline embolization device.

Figure 2

(A) Correlative analysis between RRV (PED) and MC. (B) Mean PD of first to fifth coils. (C) Analysis of covariance between the two groups. There were no significant changes in RRV after the second coil (PD >7.06%). A p value of <0.05 was considered statistically significant. **p<0.001. (D) In the whole cohort, the flow velocity in the aneurysm decreased as the PD increased. PD, packing density; PED, Pipeline embolization device; RRV, reduction ratio of velocity.

Hemodynamics of first coil after PED deployment

After PED deployment, the coils were inserted into the aneurysm sac. After the first coil embolization, the mean PD was 3.7%±1.84% (figure 2B). Compared with the flow status after PED deployment, the flow velocity in the aneurysm was further decreased with the first coil embolization and the reduction was significant (10.0 vs 5.3 cm/s; p<0.001). The mean RRV (first coil) was 44.82%±12.11%, which was not significantly different from the RRV (PED) (44.82% vs 34.63%; p=0.414) (the RRV of every coil is shown in online supplemental table 1).

Hemodynamics of subsequent coils

The PD correspondingly increased with further coil embolization. The relationship between PD and RRV (coil) in the aneurysm sac is shown in figure 2C. The effect of RRV (coil) of the second coil was significantly lower than that of the first coil (p<0.001)—that is, when the PD increased to 7.06%±3.69%, the hemodynamic effect of subsequent coils was not statistically significant. The velocity in the sac decreased as additional coils were inserted (figure 2D). The RRV (coil) after the third coil was accordingly lower, but no significant changes in velocity occurred (p=0.49).

The reduction ratio of the inflow rate was decreased along with the coil insertion, but no significant differences were found. No obvious changes in the reduction ratios of energy loss during the process of coil insertion were seen (online supplemental figure 1A).


The goal of the present study was to investigate the coil embolization strategy after FD deployment from a hemodynamic viewpoint. We believe that the results truly reflect the hemodynamic changes during the process of coil packing and may help to identify the appropriate number of coils or the PD when using an FD. To the best of our knowledge, this study is the first to use high-fidelity finite element simulation technology for both an FD and coils to analyze the hemodynamic changes in a cohort of real cases. This is also the first hemodynamic study to focus on the necessity of coil usage and how to perform coil embolization after FD deployment. We found that, although the use of coils was beneficial for intra-aneurysmal flow reduction, it may be unnecessary to continue packing after the second coil has been packed or after the PD has reached an average of 7.06%.

Hemodynamics and FD treatment

The concept of treatment using FDs is closely related to hemodynamics theory. It is more valuable and meaningful to analyze and discuss the outcome of an aneurysm treated by an FD from a hemodynamic viewpoint. Some clinical problems of FD treatment have been reported in the literature using hemodynamic methods, including aneurysm occlusion and delayed hemorrhage after FD placement. Chen et al 11 carried out CFD simulation for 94 aneurysms treated with PED. They showed that the lower RRVs in the whole aneurysm and the neck plane could lead to incomplete occlusion of post-embolization. Mut et al 12 constructed CFD models and used them to characterize the hemodynamic environment immediately before and after PED treatment. They found that rapidly occluded aneurysms had a lower post-treatment mean velocity, inflow rate, and shear rate than slowly occluded aneurysms. Additionally, complete stent expansion is another critical point for FD treatment. Zhang et al 13 examined the stent wire configurations in 14 different scenarios through CFD. They found that the effect of aneurysmal flow reduction will be degradation when incomplete stent expansion occurred at the aneurysmal orifice and distal of the stent. These studies show that many researchers have begun to study the role of FDs using hemodynamic methods and that hemodynamic mechanisms play an important role in the clinical events associated with FDs.

Various complications have also been described when using FDs.14 15 Postoperative delayed rupture of IAs is a fatal complication. Li et al 16 showed that an unstable flow pattern and higher energy loss after PED placement for treatment of IAs may be important hemodynamic risk factors related to delayed aneurysm rupture. Cebral et al also analyzed the mechanism of delayed rupture of aneurysms. The results of CFD research have suggested that higher pressure and continued inflow into the aneurysm after FD placement might serve as mechanisms for delayed rupture.17 Thus, supplementary loose coil packing is sometimes considered to help create stasis in the aneurysm and may constitute a safe and effective treatment for aneurysms at high risk of rupture. However, no hemodynamic studies have indicated how many coils are needed after FD placement.

How to perform coil embolization after FD placement

With the widespread use of FDs in clinical practice, the use of combined coils is becoming increasingly more common. This method is considered to increase the complete occlusion rate of the aneurysm and reduce the risk of postoperative delayed aneurysm rupture.18 19 A clinical multicenter retrospective study in China showed that the occlusion rate of PED placement plus coiling (about 85.9%) was significantly higher than that of PED placement alone (about 77.1%) (p<0.001).20 A study by Lin et al also showed that PED deployment combined with coils can provide a higher occlusion rate than PED deployment alone (93.1% vs 74.7%, p=0.03).21 Many studies have shown that PED deployment combined with coils is more effective than PED deployment alone. Loose packing is most commonly used, and no standard for the degree of embolization has been established. Chen et al 11 investigated the factors influencing the outcome of aneurysms by the CFD method. Their results showed that, in the group of patients treated with a PED with coils (mean PD 9.84%), the velocity in the whole aneurysm decreased by 76.3%, which was a greater decrease than in the group treated by a PED alone (55.8%). However, the porous media simulation method that the authors used cannot meet the simulation accuracy of coils. In the present study, FEA provided a highly accurate model for post-treatment CFD analysis to simulate the real therapeutic process of the PED and every coil. Moreover, a virtual coiling technique was used by Morales et al to treat the aneurysm geometries with coil models.22 They showed that the influence of the coil configuration on intra-aneurysmal hemodynamics was negligible for elevated PDs (near 30%). However, the hemodynamic results obtained in their study were not based on treatment with FDs but on treatment with coils only. These findings cannot answer the question of the most effective PD when FDs are combined with packing coils. Damiano et al used the FEM technique to simulate the optimal PD of coils after FD deployment. They proved that the addition of coils produced no further hemodynamic effect until the PD exceeded 11%.23 However, they only examined virtual coil packing into a single aneurysm, and the model was too idealistic. In the present study, we used a series of simulations based on clinical therapy rather than an idealized model. Our study shows that the packing of the first coil can further reduce the intra-aneurysmal velocity. According to this result, the use of coils may improve the post-PED hemodynamic effect in treated aneurysms.

As the PD increases, the intra-aneurysmal blood flow velocity gradually decreases, indicating that the impact of subsequent coils is gradually enhanced. However, for unruptured aneurysms treated by an FD, excessive packing of the aneurysm sac may be futile and may increase the risk of rupture.24 This may be an excessive operation for aneurysms with a higher PD because of the excellent blood flow diversion ability of FDs. Therefore, this study was performed to identify the critical value of effective hemodynamic benefits in the process of coil packing of aneurysms through high-fidelity finite element simulation methods, thus avoiding unnecessary packing after the optimal PD is reached from the perspective of hemodynamics. We observed that the intra-aneurysmal hemodynamics tended to stabilize as the PD approached an average of 7.06% or after the second coil based on the PED had been used. Our results may help physicians treating IAs with PED implantation to determine how many coils can achieve the best treatment effect from a hemodynamic viewpoint.

Limitations of the study

Despite these advantages and our interesting findings, this work has some limitations. The small sample size is the main limitation of this study and might have influenced the accuracy of the results. Thus, larger prospective studies with more comprehensive assessments are necessary in the future. In addition, the inflow boundary condition we used was uniformly pulsatile and this assumption may affect the hemodynamic results. We previously studied the effect of the inflow boundary conditions on the hemodynamic results between uniformly and patient-specific pulsatile.8 We found that the patient-specific inflow boundary conditions were not necessary for conclusion. Moreover, in the present study we used the reduction ratio of the parameter as the normalized one in the comparison. The bias from the inflow boundary conditions should be diminished. Furthermore, this was a single-center study and was therefore subject to inherent bias in patient selection. Additionally, the number of coils was determined by the neurointerventionalists of our hospital, so the coils for each patient were inconsistent. Finally, the sizes of the aneurysms differed among the patients enrolled during the study period, which resulted in differences in the PD.


From a hemodynamic viewpoint, adjunct coiling could improve the post-PED hemodynamic environment of treated intracranial aneurysms. However, dense packing is not necessary because the intra-aneurysmal hemodynamics tend to stabilize as the PD reaches an average of 7.06% or after insertion of the second coil.

Data availability statement

Data are available upon reasonable request. Not applicable.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by the Ethics Committee of Beijing Tiantan Hospital (KY2017-017-01). Participants gave informed consent to participate in the study before taking part.


We would like to thank the patients who participated in this study.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @Ying Zhang

  • Contributors Study concept and design: JL, MZ, XY. Acquisition of data and technique support: MZ, ZT, XL, XY, JX. Analysis and interpretation of data: YZ, KW, YZ. Drafting or revising the manuscript: MZ, JL, XL. Final approval of the version to be published: JL, JX. Agreement to be accountable for all aspects of the work: MZ. Responsible for the overall content: JL.

  • Funding This work was supported by Beijing Municipal Administration of Hospitals Incubating Program (PX2022022), Beijing Tiantan Hospitals Authority Youth Programme (code: QML20190503), and National Natural Science Foundation of China (grant numbers: 81801156, 82072036, 81901178 and 81801158).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.