Bio-Medical Materials and Engineering 24 (2014) 491–499 491

DOI 10.3233/BME-130835
IOS Press

Modeling the circle of Willis to assess the

effect of anatomical variations on the
development of unilateral internal carotid
artery stenosis1
Chi Zhanga, Ling Wanga, Xiaoyun Lia, Shuyu Lia, Fang Pua,b, Yubo Fana,b, and Deyu Lia,b,∗
a
Key Laboratory for Biomechanics and Mechanobiology of Ministry of Education, School of
Biological Science and Medical Engineering, Beihang University, No. 37 Xueyuan Road, Haidian
District, Beijing, 100191, China
b
State Key Lab of Virtual Reality of Technology and Systems, Beihang University, No. 37 Xueyuan
Road, Haidian District, Beijing, 100191, China

Abstract. Circle of Willis (CoW) plays a significant role in maintaining the blood supply for the brain. Specifically, when the
stenosis occurs in the internal carotid artery (ICA), abnormal structures of CoW would decrease the compensatory capacity,
leading to the local insufficiency of cerebral blood supply. The present paper built a series of lumped parameter models for
CoW, and simulated the blood redistribution caused by the unilateral ICA stenosis with different severities in cerebral arteries
in the normal and abnormal CoW respectively. The results showed that when unilateral ICA stenosis occurred, the collateral
circulation was built through the anterior communicating artery and the ipsilateral posterior communicating artery,
maintaining the flow in cerebral arteries. The absence of the two communicating arteries would cause an obvious decrease of
flow in local cerebral arteries in the anterior circulation. In conclusion, the two arteries play a significant role in maintaining
the balance of cerebral blood supply in the development of ICA stenosis.

Keywords: Circle of Willis, internal carotid artery, stenosis, anatomical variation, lumped parameter model

1. Introduction

Circle of Willis (CoW) plays a pivotal role in preventing the stroke caused by cerebral ischemia,
because the collateral circulation of CoW can maintain sufficient and stable blood supply for the brain.
CoW consists of a network of cerebral arteries, with the function of autoregulation when the inflow of
CoW fluctuates or decreases. The blood is delivered to the brain through the two internal carotid
arteries (ICAs) and the basilar artery (BA). Each ICA bifurcates into a middle cerebral artery (MCA)

1
This study was supported by the National Key Technology R & D Program (2012BAI19B04), National Natural Science
Foundation of China (No. 61190122) and the grant from the State Key Laboratory of Virtual Reality Technology and
Systems, Beihang University.
∗
Corresponding author: Deyu LI, Email: deyuli@buaa.edu.cn

0959-2989/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved
492 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations

and an anterior cerebral artery (ACA) to perfuse the temporal, parietal and frontal regions of the brain
(anterior circulation). BA is divided into the left and right posterior cerebral arteries (PCAs), to deliver
blood to the regions of the brain stem (posterior circulation). Two ACAs are interconnected by the
anterior communicating artery (ACoA). The anterior and posterior circulations are linked by two
posterior communicating arteries (PCoAs). The ring-like structure is considered as a safeguard
mechanism against cerebral arterial stenosis or occlusion.
Unfortunately, the normal structure of CoW is relatively uncommon. Plenty of clinical studies found
that generally only a half of human beings have the normal CoW structures [2, 15]. Though there is no
evidence that the variation of CoW structure is correlated with the race and region, it has been
proposed that the variation is associated with some vascular diseases [4, 16]. The anatomical
variations reduce the collateral availability and increase the risk for stroke and transient ischemia
attack (TIA) [12]. The investigation on the morphological variations and the change of the
compensatory capacity of CoW has significant implications for understanding the mechanism of
stroke and for planning the intracranial neurosurgery or vascular surgery [6, 13].
ICA, as an important pathway for cerebral blood supply, is found a preferred site for stenosis in
Chinese people [14, 24]. The ICA stenosis is proved to be closely associated with stroke and TIA
[7,20]. Beyond all doubt, the anatomical variation of CoW plays a significant role in the mechanism of
stroke induced by the ICA stenosis. Several mathematical studies have been conducted to investigate
the change of compensatory capability of CoW caused by the anatomical variations [5, 11, 21, 23].
They showed that the compensatory ability of the normal CoW could maintain the required efferent
flow rates in case of an occlusion of an ICA. Although they also pointed out the worst variation of
CoW when an ICA was occluded, the process of blood redistribution in CoW was not fully revealed in
the development of ICA stenosis.
Accordingly, the present study focuses on the establishment of collateral circulation in the
development of unilateral stenosis based on the lumped parameter model of CoW with anatomical
variations. The aim is to investigate how the collateral circulation works in the CoW when the severity
of ICA stenosis increases.

2. Methods

2.1. Mathematical model

The mathematical model is based on the linear, 1D equations of pressure and flow wave propagation
in a compliant vessel, which is analogous to an electric component shown in Fig. 1 [12]. The
governing equations result from conservation of mass and momentum applied to a 1D impermeable
tubular control volume of incompressible and Newtonian fluid, which take the form:
dPo
C = Qi − Qo
dt (1)
dQi
L = Pi − Po − Qi R
dt (2)
where Qi and Qo are inflow and outflow of the vessel, respectively; Pi and Po are blood pressures
upstream and downstream the vessel, respectively. The resistance, inertia of the blood flow, and the
 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations 493

compliance of the vessel are represented as R, L, and C respectively. The value of R, L, and C are
calculated using equations based on the geometrical data [3,18]:
8μl
R= (3)
πr 4
9 ρl
L= (4)
4πr 2

3πlr 3
C= (5)
2 Eh
where l, r and h represent the length, radius and thickness of the vessel, respectively; E is the elastic
parameter of the vascular;  is the blood density;  is the viscosity of the blood. The values of  and 
are set as 1.06 g/ml and 0.004 Pays respectively in this study.
In order to investigate the compensatory ability of CoW, 4 lumped parameter models of CoW were
constructed based on the governing equations (Fig. 2). Model A was a normal CoW while model B
and C were an abnormal CoW with the absence of ACoA or unilateral PCoA respectively. ACoA and
one PCoA were absent simultaneously in Model D.

Fig. 1. Electric circuit analogy of a blood vessel. The flow rate through the vessel was defined as Q (ml/s). The blood
pressures in the vessel was defined as P (mmHg). Resistors were denoted by R (mmHgys/ml), while capacitors and
inductances by C (ml/mmHg) and L (mmHgys2/ml), respectively.

Fig. 2. The model of CoW. A, a normal case; B, a case without ACoA; C, a case without left
PCoA; D, a case without ACoA and left PCoA at the same time.

2.2. Physiological data

The geometric and physiological data of the vessels in CoW were listed in Table 1. The data, aiming
at healthy young adults, were referred to the reports by Fahrig et al [9]. To simulate the unilateral
stenosis in ICA, the geometric data of ICA was changed according to the severity of the stenosis. The
severity of the stenosis () was defined as the percentage reduction of cross-section in the obstructed
vessels:
494 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations

As
β = (1 − ) ×100%
Ao (6)
where As and A0 were the cross-section area of the artery after and before the stenosis respectively.
For each model, the collateral circulation in CoW was simulated with the stenosis degree from 0% to
90% in the left ICA.
Table 1
The geometric and physiological data of arteries in CoW
Arterial Length Radius Thickness Elastic modulus Peripheral resistance
Code*
segment (cm) (cm) (cm) (106Pa) (mmHgys/ml)
ICA 0102, 1011 17.7 0.200 0.050 0.8 --
BA 0304 2.9 0.162 0.040 1.6 --
MCA 0209, 1116 11.9 0.143 0.036 1.6 22.4
ACA1 0207, 1114 1.2 0.117 0.029 1.6 --
ACA2 0708, 1415 10.3 0.120 0.030 1.6 43.7
PCA1 0405, 0412 0.5 0.107 0.027 1.6 --
PCA2 0506, 1213 8.6 0.105 0.026 1.6 29.1
ACoA 0714 0.3 0.074 0.019 1.6 --
PCoA 0502, 1211 1.5 0.073 0.018 1.6 --
* The codes represented the arteries shown in Fig. 2.
The input boundary condition for CoW models was set as the blood pressure at the beginning of
ICA and VA shown in Fig. 3, when ICA was not constricted [8]. The cardiac cycle was set as 0.8s.
The blood flow through the contralateral ICA and BA was assumed to be constant when the stenosis
occurred in left ICA. The fourth-order Runge-Kutta algorithm was used to solve the differential
equations of the CoW models. The calculation was performed for 5 cycles. The data of the last cycle
was considered as the simulation results to make sure the convergence of the calculation. The blood
flow in each artery simulated by Model A without stenosis was compared with the data measured by
Transcranial Doppler (TCD), to validate the accuracy of the model. The TCD measurement was
performed on a healthy adult female. Since TCD could only provide the velocity information of flow,
the data was integrated over the area of the artery to calculate the flow rate.

Fig. 3. The input pressure wave at ICA and BA.

 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations 495

3. Results

The results for the blood pressure and flow pulses in the arterials of CoW in Model A without any
stenosis were shown in Fig. 4, comparable to the published reports [21]. The results were compared
with the data from TCD measurement shown in Fig. 5, indicating that the simulation results were
consisitent with the clinical data.

Fig. 4. The blood pressure (right side) and flow rate (left side) in the arteries of CoW in Model A. The results in only half
arteries were shown, since the model was symmetric. The abscissa ranged from t=0 to t=0.8s (one full cardia cycle). The
ordinate ranged from 60 to 120 mmHg (for blood pressure) or from 0 to 8 ml/s (for flow rate).

Fig. 5. The comparison between the simulation results and the clinical data measured by TCD in left ACA, MCA, and PCA.
Left: the flow rate of simulation results in arteries. Right: the flow rate calculated with the clinical data from TCD
measurement.

Since the model was validated, the compensatory capacity were evaluated between the 4 models
with the different stenosis degrees in ICA (Fig. 6). For each model, the flow rate in the left ICA
decreased gradually to 1% of that in the normal case (without stenosis), as the stenosis degree
increased to 90%. In Model A, the decrease of the flow rate in the ipsilateral ACA, MCA, and PCA
was slow, because that the collateral circulation was built through ACoA and PCoAs. The flow rate in
the ipsilateral PCoA and ACoA increased significantly, as the stenosis degree increased, indicating
that they played a significant role in the establishment of collateral circulation (Figure 7).
496 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations

Fig. 6. The decrease of the flow rate in ACA, MCA, and PCA with the stenosis degree increasing from 0% to 90% in ICA.
The flow rate in the ipsilateral ACA, MCA, and PCA was shown in the left column (a, c, and e respectively); while the
contralateral arteries was in the right (b, d, and f respectively).

Fig. 7. The change of flow rate in communicating arteries.

 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations 497

In Model B and C, the decrease of flow rate became faster, because the ipsilateral PCoA or ACoA
was absent in the model, which diminished the compensatory ability of CoW. In Model D, the flow
rate in the ipsilateral ACA and MCA decreased dramatically, because of the simultaneous absence of
the ipsilateral PCoA and ACoA. The flow rate decreased to nearly 0 when ICA stenosis degree
increased to 90%, implying that the cerebral blood supply was severely insufficient, and the risk for
stroke was extremely elevated.

4. Discussion and Conclusion

Lumped parameter models of CoW were developed in the present study to simulate how ICA
stenosis affected the blood flow in the normal or abnormal CoW. The model of the normal case was
validated by the clinical data from TCD measurement. The simulation results from Model A were also
compared with the published reports, and showed a good agreement with them. Based on Model A,
the CoW without one PCoA or/and ACoA was also simulated by Model B to D. Therefore, the results
in the study were convincing and capable to reveal the difference in the compensatory ability caused
by the structural variation of CoW in the development of ICA stenosis.
Our results showed that, in spite of the anatomical variations of CoW, the influence of ICA stenosis
was mainly focused on the anterior circulation. Because the blood in posterior circulation was mainly
supplied by BA, the flow rate in PCA was less affected by the ICA stenosis, even when ICA was
occluded. BA, rather than ICA, was therefore considered as an important artery for the posterior
circulation. On the other hand, in the anterior circulation, the effect of ICA stenosis was much more
obvious. In Model A, which simulated the normal CoW, the ICA stenosis caused an obvious decrease
of flow rate in ACA and MCA. When the stenosis degree increased from 0% to 90%, the flow rate in
the ipsilateral ACA and MCA decreased 31.5% and 32.8% respectively. The flow rate in contralateral
arteries also decreased, but in a smaller extent, since the blood redistributed through the collateral
circulation. If ACoA was absent in CoW (Model B), the decrease was more obvious in the ipsilateral
arteries (both 41.9% in ACA and MCA). The decrease was also enlarged by the absence of the
ipsilateral PCoA (34.4% and 36.6% in ACA and MCA respectively in Model C). In Model D, the flow
rate in the ipsilateral ACA and MCA decreased dramatically (both 99.0% in ACA and MCA), because
the absence of the ipsilateral PCoA and ACoA invalidated the collateral circulation.
Our results also showed that the ACA and the ipsilateral PCoA played a significant role in the
establishment of the collateral circulation of CoW. Cassot firstly pointed out the importance of ACoA
in cerebral hemodynamics with ICA diseases [5]. Alastruey et al. systematically analyzed the effects
of ACoA and PCoA diameters on cerebral flows, and revealed the association between the diameters
of communicating arteries and the compensatory ability of CoW [1]. In the present study, although the
effect of communicating arteries diameters was not considered in the models, the effect of the absence
of communicating arteries on the cerebral flows were simulated during the development of ICA
stenosis. The decrease of blood flows in the ipsilateral ACA, MCA, and PCA caused by the absence of
the ACoA and ipsilateral PCoA were enlarged when the severity of ICA stenosis increased, indicating
that the effect of the absence of ACoA and PCoA on the compensatory ability was larger in the patient
with severe ICA stenosis.
Our study indicated that the abnormality of CoW aggravated the effect of ICA stenosis on the
cerebral blood supply. Actually, the abnormality of CoW is quite common clinically. Cassot et al.
found that PCoA was absent in 9% people while only 1% have no ACoA [20]. Raghavendra
investigated the morphological variation in Indian, and found that the absence of ACoA accounted for
498 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations

9.1% of the cases with variations while PCoA accounted for 18.2% [22]. Li et al. also statistically
analyzed the variation of CoW in Chinese [19]. He classified the abnormality of CoW into 2 classes,
partly complete and incomplete CoW. The incomplete CoW was defined as the CoW with the absence
of arteries in both anterior and posterior circulation, which accounted for 5.9% in Chinese [17]. The
value was even higher (18.0%) in the patients with cerebrovascular diseases. The present study
indicated that, for the people with incomplete CoW, ICA stenosis would be very dangerous, since the
blood flow in anterior circulation were easily affected by ICA stenosis.
The findings in the present study were based on the computations for the CoW of a healthy adult
without concomitant cardiac or other arterial diseases. However, the physiological situation might be
complicated, as the stenosis in carotid arteries might also be associated with coronary heart disease or
other vascular diseases. In addition, although the simulation results were comparable with the
experimental data, the quantitative comparison was difficult to achieve because of the measurement
errors of TCD and the individual variations. Regarding the gap to the complexities of realistic cases,
our results have limitations, which point to the future study, but do not challenge our fundamental
findings concerning the influence of ICA stenosis on the blood flow in CoW. Further efforts should be
focused on the effect of the whole cardiovascular system on the establishment of collateral circulation
in CoW.
In conclusion, the effect of unilateral ICA stenosis with different degrees on normal and abnormal
CoW has been assessed with the lumped parameter models. The results show that the ipsilateral PCoA
and ACoA play a significant role in maintaining the cerebral blood supply, especially in anterior
circulation. The absence of the ipsilateral PCoA or ACA would aggravate the decrease of flow rate in
the ipsilateral ACA and MCA caused by ICA stenosis. The simultaneous absence of the ipsilateral
PCoA and ACoA would be very dangerous for the patients with ICA stenosis, whose blood flow
decreases dramatically in anterior circulation. Therefore, the abnormality of CoW should be an
important screening factor for the precaution of ICA stenosis and stroke.

5. Acknowledgement

This study was supported by the National Key Technology R & D Program (2012BAI19B04),
National Natural Science Foundation of China (No. 61190122) and the grant from the State Key
Laboratory of Virtual Reality Technology and Systems, Beihang University. The author would like to
acknowledge Lihong Yang and Feiyan Chang for their contributions.

Reference

[1] J. Alastruey, K.H. Parker, J. Peiro, S.M. Byrd, S.J. Sherwin, Modelling the circle of Willis to assess the effects of
anatomical variations and occlusions on cerebral flow, Journal of Biomechanics 40 (2007), 1794-1805.
[2] B.J. Alpers, R.G. Berry, R.M. Paddison, Anatomical studies of the circle of Willis in normal brain, Archives of
Neurology 81 (1959), 409-418.
[3] A.P. Avolio, Multi-branched model of the human arterial system, Medical & Biological Engineering & Computing 18
(1980), 709-718.
[4] E. Behzad, D. Majid, A Saeed, G. Mohammad, N. Bashir, K. Ebrahim, Are the distributions of variation of circle of
Willis different in different populations? Results of an anatomical study and review of literature, BMC Neurology 6
(2006), 22-30.
[5] F. Cassot, V. Vergeur, P. Bossuet, B. Hillen, M. Zagzoule, J.P. MarcVergnes, Effects of anterior communicating artery
diameter on cerebral hemodynamics in internal carotid artery disease, Circulation 92 (1995), 3122-3131.
 C. Zhang et al. / Modeling the circle of Willis to assess the effect of anatomical variations 499

[6] J.R. Cebral, M.A. Castro, O. Soto, R. Lohner, N. Alperin, Blood-flow models of the circle of Willis from magnetic
resonance data, Journal of Engineering Mathematics 47 (2003), 369-386.
[7] D.R. Craig, K. Meguro, C. Watridge, J.T. Robertson, H.J. Barnett, A.J. Fox, Intracranial internal carotid artery stenosis,
Stroke 13 (1982), 825-828.
[8] G. Ding, C. Lu, W. Yao, Y. Wang, S. Cheng, J. Cheng, J. Gao, X. Wang, A hemodynamic model and a mathematical
method to calculate the dynamics index for cerebral circulation, Journal of Hydrodynamics 4 (1997), 71-81.
[9] R. Fahrig, H. Nikolov, A.J. Fox, D.W. Holdsworth, A three-dimensional cerebrovascular flow phantom, Medical
Physics 26 (1999), 1589-1599.
[10] R.D. Henderson, M. Eliasziw, A.J. Fox, P.M. Rothwell, H.J.M. Barnett, Angiographically defined collateral circulation
and risk of stroke in patients with severe carotid artery stenosis, Stroke 31 (2000), 128-132.
[11] B. Hillen, H. Hoogstraten, L. Post, A mathematical model of the flow in the circle of Willis, Journal of Biomechanics
19 (1986), 187-194.
[12] B. Hillen, A.H. Drinkenburg, H.W. Hoogstraten, L. Post, Analysis of flow and vascular resistance in a model of the
circle of Willis, Journal of Biomechanics 21 (1988), 807-814.
[13] A.W.J. Hoksbergen, C.B.L. Majoie, F.J.H. Hulsmans, D.A. Legemate, Assessment of the collateral function of the
circle of Willis: three-dimensional time-of-flight MR angiography compared with transcranial color-coded duplex
sonography, American Journal of Neuroradiology 24 (2003), 456-462.
[14] Y.N. Huang, S. Gao, S.W. Li, Y. Huang, J.F. Li, K.S. Wond, R. Kay, Vascular lesions in Chinese patients with transient
ischemic attacks, Neurology 48 (1997), 534-535.
[15] S. Kamath. Observations on the length and diameter of vessel forming the circle of Willis, Journal of Anatomy 133
(1981), 419-423.
[16] K.N. Kayembe, M. Sasahara, F. Hazama, Cerebral aneurysms and variations in the circle of Willis, Stroke 15 (1984),
846-850.
[17] M.J. Krabbe-Hartkamp, J. van-der-Grond, F.E. de-Leeuw, J.C. de-Groot, A. Algra, B. Hillen, M.M. Breteler, W.P. Mali,
circle of Willis: morphologic variation on three-dimensional time-of-flight MR angiograms, Radiology 207 (1998),
103-111.
[18] X. Li, J. Bai, S. Cui, S. Wang, Simulation study of the cardiovascular functional status in hypertensive situation,
Computers in Biology and Medicine 32 (2002), 345-362.
[19] X. Li, X. Zhou, C. Tang, L. Zhu, T. Xiang, D. Chen, R. Li, Morphologic and functional study of circle of Willis,
Chinese Journal of Clinical Anatomy 24 (2006), 640-644.
[20] D.J. Marzewski, A.J. Furlan, P.S. Louis, J.R. Little, M.T. Modic, G. Williams, Intracranial internal carotid artery
stenosis: longterm grognosis, Stroke 13 (1982), 821-824.
[21] K.T. Moorhead, C.V. Doran, J.G. Chase, T. David, Lumped parameter and feedback control models of the
auto-regulatory response in the circle of Willis, Computer Methods in Biomechanics and Biomedical Engineering 7
(2004), 121-130.
[22] K. Raghavendra, A morphological study of circle of Willis and its variations in adult human cadavers of North
Karnataka, Ph.D. Dissertation, KLE University, 2012.
[23] A. Viedma, C. Jimenez, V. Marco, Extended Willis circle model to explain clinical observations in periorbital arterial
flow, Journal of Biomechanics 30 (1997), 265-272.
[24] C. Zhang, S. Xie, S. Li, F. Pu, X. Deng, Y. Fan, D. Li, Flow patterns and wall shear stress distribution in human internal
carotid arteries: the geometric effect on the risk for stenoses, Journal of Biomechanics 45 (2012), 83-89.
Copyright of Bio-Medical Materials & Engineering is the property of IOS Press and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.