Professional Documents
Culture Documents
DOI 10.1007/s00234-010-0804-4
INTERVENTIONAL NEURORADIOLOGY
Received: 9 September 2010 / Accepted: 5 November 2010 / Published online: 24 November 2010
# Springer-Verlag 2010
attempts have been made to find a reliable parameter from retrospective review of a prospectively maintained database
aneurysm configurations to predict rupture risk [2–6]. of patients presenting to two institutes with aneurysms. The
Among many geometric indices, aneurysm size is a well- study protocol included all consecutive patients in whom
documented risk factor for aneurysm rupture. Size is aneurysms were diagnosed by DSA and 3DRA in two
commonly represented as the maximum length at one viewing institutions [Kyung Hee University Hospital (KUH) from
angle of the aneurysm. However, we suggest that aneurysm June 2006 to October 2007 and Kyung Hee University
size is too simple a measurement to depict real aneurysm Hospital at Gangdong (KUHG) from June 2006 to February
configurations. Aspect ratio, the ratio of the aneurysm’s height 2008]. Brain CT, laboratory results, and clinical information
and neck width, first introduced by Ujiie et al. [7], is also well of subjects were retrospectively evaluated for the presence
recognized as a geometric risk factor for aneurysm rupture. of subarachnoid hemorrhage.
As with size, the aspect ratio uses length that is measured on In patients with multiple aneurysms, ruptured aneurysms
two-dimensional (2D) sources. Therefore, measurements were identified by direct observation at the time of surgery.
taken with different viewing angles and observers could Patients with multiple aneurysms were excluded from this
provide a potential bias. Additionally, measurements study if ruptured aneurysm was not confirmed by direct
obtained this way may not reflect the real aneurysm observation in surgery. Aneurysms with poor image quality
configuration because it is based on the assumption that of 3DRA and extradural aneurysms located proximal to the
the aneurysm is round or ellipsoid [8]. Therefore, in some opthalmic segment were also excluded.
previous studies [3, 9], bi- or multi-lobulated aneurysms Two hundred fourteen aneurysms of 195 patients (mean
were excluded from subjects to avoid the bias. age, 57.44±11.21 years; males, 69; females, 126) were
Today, we can determine the real aneurysm configura- included in this study; 52 patients presented to KUH, and
tion by use of three-dimensional rotational angiography 143 patients presented to KUHG. Multiple aneurysms were
(3DRA). The system workstation provides not only 3D identified in 18 patients. One hundred five aneurysms were
reconstruction of images but also various analysis tools that categorized as ruptured, and 109 aneurysms were unrup-
are easily used by clinicians. Additionally, 3DRA allows tured. Of the 105 patients with ruptured aneurysms, 96
for the measurement of aneurysm volume [8, 10]. patients presented with a single ruptured aneurysm, and
Contrary to aneurysm size, a dimension expressed to the nine patients had at least one additional unruptured
length, aneurysm volume cannot be influenced by the shape aneurysm.
or projection of view of the aneurysm. The volume is identical
to the amount of blood into the aneurysm and is related to the Three-dimensional rotational angiography
volumetric flow rate. The aneurysm neck area, which may
reflect intra-aneurysmal inflow [11], also can be associated All of the 3DRA examinations were performed using a flat
with the hemodynamics of aneurysms. Here we propose a panel angiographic unit (Allura Xper FD 20/10 angiogra-
novel parameter, the ratio of volume to neck area, to describe phy system; Philips Medical Systems, Best, The Nether-
the hemodynamic geometrics of intracranial aneurysms. The lands) with a field of view of 16×16 cm and a frame rate of
basic concept of this index is the same as that of the aspect 30/s. The ranges of rotation included +110° to −110°. A
ratio. However, we presumed that our novel parameter is free volume of 21–35 mL of non-ionic, iso-osmolar contrast
from the bias that can occur with aspect ratio. medium (270 mg per I/mL, iodixanol; Visipaque, Nycomed
The aim of this study is to determine whether volume Imaging, Oslo, Norway) was injected through a 5–6 Fr
and the volume-to-neck ratio (VNR) is a more reliable catheter by use of an injector with a velocity of 3–5 mL/s.
index than 2D indices, including the aspect ratio, to predict Image acquisition was started 1–3 s after contrast material
aneurysm rupture status. We compared the discriminative injection. The acquisition time of the images was 4.1 s.
capacity and interobserver variability of VNR to those of Volume-rendered 3D images were reconstructed with a
the aspect ratio using a receiver operating characteristic 50% magnification (a field of 37.56 cm2) and a matrix of
(ROC) curve and the interobserver correlation coefficient 256 pixels3 using the 3DRA volumetric measurement of the
(ICC), respectively. system software. The threshold for the volume-rendering
image was fixed as a default value provided by the
software.
Materials and methods
Image analysis
Subjects
We recorded the aneurysm locations as the anterior cerebral
This study was performed according to the ethical standards artery (ACA; ACA and anterior communicating artery);
of the Institutional Review Board. This study represents a middle cerebral artery (MCA; MCA trunk and MCA
Neuroradiology (2011) 53:883–889 885
bifurcation); posterior communicating artery (PCoA; PCoA an index's ability to discriminate between ruptured and
and anterior choroidal artery); paraclinoid internal carotid unruptured aneurysms. Thus, the larger the AUC is, the
artery (ICA); and posterior circulation (basilar top, basilar better the discrimination. The cut-point was determined at
trunk, and vertebral artery). the point where the Youden index [12] shows the maximum
Aneurysms were classified into simple-lobed and value (Youden Index=sensitivity+specificity−1). Likewise,
irregular aneurysms according to their shape. The the comparison of the AUC between the aspect ratio and
irregular shape was defined as an aneurysm with the VNR was performed for simple-lobed and irregular
daughter sac and multilobular aneurysms. The relation- aneurysms groups.
ship between aneurysm shape and rupture status was To assess the correlation between values obtained from
analyzed by the χ2 test. two observers, the ICC between two observers was tested
The geometric indices of aneurysms on 3D volume- for VNR and the aspect ratio. The calculated ICCs were
rendering images of the workstation were measured by compared between the VNR and aspect ratio.
two neuroradiologists (CR and EK) independently. The The Pearson correlation coefficient and linear regression
following five dimensions of aneurysms were manually were used to analyze the relationships between VNR and
measured on the volume-rendering 3D images with the the aspect ratio. P <0.05 was considered statistically
ruler tool of the system software: (1) height (H, the significant.
maximum distance between the neck and dome of an
aneurysm), (2) width (W, the longest diameter of an
aneurysm perpendicular to H), (3) depth (D, the longest
diameter of an aneurysm perpendicular to the axis of the
parent artery, (4) neck width (N), and (5) another neck
width (N′) was obtained perpendicular to the axis of the
parent artery. The neck area (Nd) was calculated as
follows: Nd (mm2)=π×(N/2×N′/2). H, W, and N were
measured in the plane parallel to the blood flow of the
parent arteries (Fig. 1).
We calculated the following secondary geometric indi-
ces: (1) aspect ratio, the ratio of H to N; (2) height–width
ratio, the ratio of H to W; and (3) bottleneck factor, the ratio
of W to N. These indices were previously evaluated in the
identification of aneurysm rupture risk [3, 4].
The aneurysm volume was acquired using the aforemen-
tioned 3DRA system software. In the volume-rendering
image, the aneurysm was manually segmented from the
parent artery, and the volume was calculated. The manual
outlining and segmentation of the aneurysm from the parent
artery was performed by the same neuroradiologist. We
calculated the VNR as follows: VNR (mL/cm2)=volume×
100/Nd.
Statistical analysis
Geometric index Unruptured (109) Ruptured (105) Mean difference (95% CI) P value
aneurysm size and was based on the assumption that the not prove this hypothesis. Despite the pitfalls of aspect
aneurysm is a round cavity [7]. Ambiguously, an aneurysm ratio, which were outlined in the introduction, this study
with a high aspect ratio can be figured two different demonstrated that the aspect ratio had a reliable discrimi-
configurations; one is an aneurysm with the larger volume nating ability between unruptured and ruptured aneurysms
for neck area and the other is a vertically elongated and had a high interobserver reliability. We suggest that
aneurysm. Which of the two shapes are more related to height would be a reliable substitute for the volume of an
the geometrical characteristics of the ruptured aneurysm? aneurysm and the aspect ratio would be a more simplified
The present results suggest that the larger volume for neck index of the VNR.
area could be correlated with rupture. The other geometric We have not analyzed other risk-related factors such as
feature, vertically elongated aneurysm, could also be smoking, hypertension, age, gender, and familial predispo-
independently correlated with aneurysm rupture. The sition. The present study evaluated only morphologic
vertical ellipticity of aneurysm can be represented as the factors related to aneurysm rupture. Morphologic analysis
height–width ratio, which we found to be significantly is a useful way to assess aneurysm characteristics indepen-
different between unruptured and ruptured aneurysms. dently without patient-level confounding variables. In this
However, the shape of ruptured aneurysms was almost study, however, the discriminating ability of morphologic
spherical (the height–width ratio is near 1.0), and the analysis was not suitable for the decision-making involved
height–width ratio had nearly identical mean values in the risks associated with observation. The VNR is yet
between the ruptured and unruptured aneurysms despite another variable that at best adds further weight to the array
the statistical difference observed previously [4, 9] and in of variables used to assess the risk of aneurysm rupture. A
the present study. Therefore, the VNR rather than the more comprehensive multivariate analysis, including the
vertical ellipticity should be given a great deal of weight on above parameters, should be performed for additional study.
reading of aspect ratio. We found that, in several previous To determine treatment of an asymptomatic unruptured
reports, the schematic illustrations showing the geometric aneurysm, other contributing factors, such as the patient’s
characteristic of the aspect ratio were misinterpreted; an age, previous history of subarachnoid hemorrhage, aneu-
aneurysm with a low aspect ratio is dumpy, and an rysm location, the patient’s history of smoking and familial
aneurysm with a high aspect ratio is lanky [1, 4, 18]. In history, and aneurysm shape (multi-lobulation and daughter
the planning of this study we hypothesized that, because sacs), should be taken into account. Another bias of this
aneurysm volume is a more realistic parameter than height, study is that the shape or size of ruptured aneurysms might
VNR would be superior to the aspect ratio in determining have changed due to surrounding clots or intra-aneurysmal
risk factors of aneurysm rupture. However, our result did thrombus. Therefore, the measured geometric indices of the
Geometric index AUC (95% CI) Cut-off value Sensitivity Specificity Positive likelihood ratio Negative likelihood ratio
Fig. 2 a Graph depicting the ROC curves for aspect ratio (bold) and irregular-shaped aneurysm group. The ROC area for VNR (0.735;
VNR (dotted line) in the simple-lobed aneurysm group. The ROC area 95% CI 0.671–0.793) tends to be larger than the AUC of the aspect
for aspect ratio is 0.757 and for VNR is 0.734. b Graph depicting the ratio (0.735; 95% CI 0.671–0.793; P=0.072)
ROC curves for aspect ratio (bold) and VNR (dotted line) in the
ruptured aneurysms could be underestimated as compared characteristics could be more simplified by the aspect
to the indices immediately before rupture. This concern has ratio, of which the value was not inferior to VNR for
also been presented in a previous study [2]. discriminating between the ruptured and unruptured
In conclusion, ruptured aneurysms had a significantly aneurysms. ROC analysis also revealed that the aspect
larger VNR than that of unruptured aneurysms. The ratio had a substantially greater ability than other
larger volume in proportion to the neck of the aneurysm secondary geometric indices. Thus, the aspect ratio is as
would be one of the morphological characteristics of a good geometric index for the assessment of aneurysm
aneurysms indicating a higher risk of rupture. These rupture risk as ever.
Acknowledgment This work was supported by the program of saccular aneurysms: a possible index for surgical treatment of
Kyung Hee University for the Young Researcher in Medical Science intracranial aneurysms. Neurosurgery 45:119–129
(KHU-2010719). 8. Piotin M, Daghman B, Mounayer C, Spelle L, Moret J (2006)
Ellipsoid approximation versus 3d rotational angiography in the
Conflict of interest statement We declare that we have no conflict volumetric assessment of intracranial aneurysms. AJNR Am J
of interest. Neuroradiol 27:839–842
9. Parlea L, Fahrig R, Holdsworth DW, Lownie SP (1999) An
analysis of the geometry of saccular intracranial aneurysms.
AJNR Am J Neuroradiol 20:1079–1089
References 10. Yamura M, Hirai T, Korogi Y, Ikushima I, Yamashita Y (2005)
Quantitative evaluation of measurement accuracy for three-
dimensional angiography system using various phantoms. Radiat
1. Raghavan ML, Ma B, Harbaugh RE (2005) Quantified aneurysm Med 23:175–181
shape and rupture risk. J Neurosurg 102:355–362 11. Byun HS, Rhee K (2003) Intraaneurysmal flow changes affected
2. Ujiie H, Tamano Y, Sasaki K, Hori T (2001) Is the aspect ratio a by clip location and occlusion magnitude in a lateral aneurysm
reliable index for predicting the rupture of a saccular aneurysm? model. Med Eng Phys 25:581–589
Neurosurgery 48:495–502 12. Fluss R, Faraggi D, Reiser B (2005) Estimation of the Youden
3. Nader-Sepahi A, Casimiro M, Sen J, Kitchen ND (2004) Is aspect Index and its associated cutoff point. Biom J 47:458–472
ratio a reliable predictor of intracranial aneurysm rupture? 13. Ozdamar N, Celebi G (1978) Pressure distribution on the wall of
Neurosurgery 54:1343–1347 experimental aneurysms. Acta Neurochir 45:27–34
4. Hoh BL, Sistrom CL, Firment CS, Fautheree GL, Velat GJ, 14. Kerber CW, Heilman CB (1983) Flow in experimental berry
Whiting JH, Reavey-Cantwell JF, Lewis SB (2007) Bottleneck aneurysms: method and model. AJNR Am J Neuroradiol 4:374–377
factor and height–width ratio: association with ruptured aneur- 15. Tateshima S, Chien A, Sayre J, Cebral J, Vinuela F (2010) The
ysms in patients with multiple cerebral aneurysms. Neurosurgery effect of aneurysm geometry on the intra-aneurysmal flow
61:716–722 condition. Neuroradiology. doi:10.1007/s00234-010-0687-4
5. Dhar S, Tremmel M, Mocco J, Kim M, Yamamoto J, Siddiqui AH, 16. Beck J, Rohde S, El Beltagy M, Zimmermann M, Berkefeld J,
Hopkins LN, Meng H (2008) Morphology parameters for intracra- Seifert V, Raabe A (2003) Difference in configuration of ruptured
nial aneurysm rupture risk assessment. Neurosurgery 63:185–196. and unruptured intracranial aneurysms determined by biplanar
doi:10.1227/01.NEU.0000316847.64140.81 digital subtraction angiography. Acta Neurochir 145:861–865.
6. Sadatomo T, Yuki K, Migita K, Taniguchi E, Kodama Y, Kurisu K doi:10.1007/s00701-003-0124-0
(2008) Morphological differences between ruptured and unruptured 17. Weir B, Amidei C, Kongable G, Findlay JM, Kassell NF, Kelly
cases in middle cerebral artery aneurysms. Neurosurgery 62:602– J, Dai L, Karrison TG (2003) The aspect ratio (dome/neck) of
609. doi:10.1227/01.NEU.0000311347.35583.0C ruptured and unruptured aneurysms. J Neurosurg 99:447–451
7. Ujiie H, Tachibana H, Hiramatsu O, Hazel AL, Matsumoto T, 18. Ma B, Harbaugh RE, Raghavan ML (2004) Three-dimensional
Ogasawara Y, Nakajima H, Hori T, Takakura K, Kajiya F (1999) geometrical characterization of cerebral aneurysms. Ann Biomed
Effects of size and shape (aspect ratio) on the hemodynamics of Eng 32:264–273