EDUCATION EXHIBIT CT Angiography of RadioGraphics ... · image quality of 3D imaging with SSD in a patient with two aneurysms (arrows in b–d)at the bifurcation of the left middle

EDUCATION EXHIBIT 637

CT Angiography ofIntracranial Aneu-rysms: A Focus onPostprocessing1

LEARNINGOBJECTIVESFOR TEST 1After reading thisarticle and takingthe test, the reader

will be able to:

� Describe the tech-nique of CT angiog-raphy performed fordetection of intracra-nial aneurysms.

� List the technicalconsiderations in andlimitations of themost common meth-ods of image postpro-cessing.

� Discuss when touse CT angiographyas the sole diagnostictest in a patient withsubarachnoid hemor-rhage before therapy.

Bernd F. Tomandl, MD ● Niels C. Kostner ● Miriam SchempershofeWalter J. Huk, MD ● Christian Strauss, MD ● Lars Anker, MD ● PeterHastreiter, PhD

Computed tomographic (CT) angiography is a well-known tool fordetection of intracranial aneurysms and the planning of therapeuticintervention. Despite a wealth of existing studies and an increase inimage quality due to use of multisection CT and increasingly sophisti-cated postprocessing tools such as direct volume rendering, CT an-giography has still not replaced digital subtraction angiography as thestandard of reference for detection of intracranial aneurysms. One rea-son may be that CT angiography is still not a uniformly standardizedmethod, particularly with regard to image postprocessing. Severalmethods for two- and three-dimensional visualization can be used:multiplanar reformation, maximum intensity projection, shaded sur-face display, and direct volume rendering. Pitfalls of CT angiographyinclude lack of visibility of small arteries, difficulty differentiating theinfundibular dilatation at the origin of an artery from an aneurysm, thekissing vessel artifact, demonstration of venous structures that cansimulate aneurysms, inability to identify thrombosis and calcificationon three-dimensional images, and beam hardening artifacts producedby aneurysm clips. Finally, an algorithm for the safe and useful applica-tion of CT angiography in patients with subarachnoid hemorrhage hasbeen developed, which takes into account the varying quality of equip-ment and software at different imaging centers.©RSNA, 2004

Abbreviations: DSA � digital subtraction angiography, dVR � direct volume rendering, FOV � field of view, ICA � internal carotid artery,MCA � middle cerebral artery, MIP � maximum intensity projection, MPR � multiplanar reformation, PICA � posterior inferior cerebellar artery,SAH � subarachnoid hemorrhage, SSD � shaded surface display, 3D � three-dimensional, 2D � two-dimensional

Index terms: Aneurysm, CT, 17.12116, 17.73 ● Aneurysm, intracranial, 17.73 ● Computed tomography (CT), angiography, 17.12116 ● Computedtomography (CT), image processing, 17.12117

RadioGraphics 2004; 24:637–655 ● Published online 10.1148/rg.243035126

1From the Department of Neurosurgery, University of Erlangen-Nuremberg, Schwabachanlage 6, D-91054 Erlangen, Germany. Presented as an edu-cation exhibit at the 2002 RSNA scientific assembly. Received May 5, 2003; revision requested June 13 and received July 30; accepted July 31. All au-thors have no financial relationships to disclose. Address correspondence to B.F.T. (e-mail: [email protected]).

©RSNA, 2004

Radio

Gra

phic

s

CME FEATURESee accompanying

test at http://www.rsna.org

/education/rg_cme.html

IntroductionSudden onset of vigorous headache typically isthe leading symptom in patients with subarach-noid hemorrhage (SAH) caused by the rupture ofan intracranial aneurysm. Computed tomography(CT) is the first step in the examination of thesepatients. Once SAH is confirmed, it is paramountto detect the source of bleeding in order to initiatetherapy. Digital subtraction angiography (DSA) isstill the most sensitive tool for the detection ofintracranial aneurysms. The selective intraarterialinjection of contrast medium ensures optimal en-hancement of the intracranial arteries with supe-rior resolution compared with that of CT or mag-netic resonance (MR) angiography. However,DSA has the disadvantage of being an invasivestudy. The risk of acquiring a permanent neuro-logic deficit with cerebral angiography in patientswith SAH is below 0.1% (1). Despite this rela-tively low risk, a noninvasive method yieldingthree-dimensional (3D) information for the plan-ning of therapeutic intervention is desirable.

The reported sensitivity of CT angiography liesin the range of 80%–97% (2–8) depending on thesize and location of an aneurysm (3). In all ofthese studies, some kind of 3D visualization wasused to analyze the CT angiography data. Little isknown about the influence of postprocessingmethods like maximum intensity projection(MIP), shaded surface display (SSD), and directvolume rendering (dVR) on the detection rate ofintracranial aneurysms. However, it can be as-sumed that the same CT angiography data maylead to varying detection rates when different vi-sualization strategies, computer platforms, andgraphics hardware are used (9–12).

In the first section of this article, technical as-pects of CT angiography with a focus on data ac-quisition are discussed. In the second section,different methods for the postprocessing of CTdata are presented, including the analysis ofsource images and the methods currently avail-able for two-dimensional (2D) and 3D postpro-cessing, such as high-resolution dVR. Then, typi-cal pitfalls encountered while working with CTangiography data are demonstrated. Finally, wepropose a reasonable paradigm for the use of CTangiography in patients with SAH, taking intoaccount that this method is still not a standard-ized procedure.

Technique of Intra-cranial CT Angiography

CT angiography can be defined as a fast thin-sec-tion volumetric spiral (helical) CT examinationperformed with a time-optimized bolus of con-trast medium in order to enhance the cerebralarteries (13). In order to visualize the intracranialarteries, the examination includes the region fromthe first vertebral body up to the vertex. It is im-portant to include the atlas in the study to ensureincorporation of the posterior inferior cerebellarartery (PICA), which has an extracranial originfrom the vertebral arteries in about 18% of cases(14).

On our four-row multisection scanner (Soma-tom 4 Volume Zoom; Siemens Medical Solu-tions, Erlangen, Germany), we used the followingparameters: 120 kVp, 200 mAs, collimation of4 � 1 mm, table feed of 2.7 mm per rotation, androtation time of 0.5 seconds. Image reconstruc-tion parameters were as follows: section thicknessof 1.25 mm, overlapping steps of 0.5 mm, andfield of view (FOV) of 120 mm2. The appliednarrow FOV of 120 mm2 leads to an excellentin-plane resolution (0.23 � 0.23 mm2) and repro-duces all relevant information (Fig 1). In addi-tion, lateral parts of the skull are already elimi-nated, which simplifies the postprocessing ofsource data. It is possible to perform reconstruc-tions in steps of 0.23 mm to produce isotropicdata (15), thus yielding voxels of equal extent inall three dimensions. In our experience, this doesnot noticeably increase image quality while dou-bling the number of source images, thus leadingto an extension of time spent on postprocessingsource data.

For enhancement of intracranial arteries, 100mL of contrast medium (Ultravist 300; Schering,Berlin, Germany) was injected intravenously at aflow rate of 4 mL/sec by using a power injector(EnVision CT injector; Medrad, Indianola, Pa).A bolus tracking method (16) was used routinelyto achieve optimal synchronization of contrastmedium flow and scanning. Once the injection isstarted, the bolus tracking software measures at-tenuation values within one internal carotid artery(ICA), and the spiral scan is automatically startedas soon as a threshold of 100 HU is exceeded.

If bolus tracking is not available, the test bolusmethod should be applied to calibrate timing ofthe data acquisition (17): Ten seconds after bolusinjection of 20 mL of contrast medium, a dy-namic single-axial-section study (one scan every

638 May-June 2004 RG f Volume 24 ● Number 3

Radio

Gra

phic

s

2 seconds) at the level of the first cervical verte-bral body is started until the contrast materialappears as hyperattenuating spots in the ICAs. Byusing this technique, the time interval betweenbolus administration and the beginning of dataacquisition can also be determined individually.

Analysis of CT AngiogramsThe examples shown in this article were createdwith the regular software of the workstation sup-plied with a Somatom Volume Zoom CT scanner(Syngo Wizard version VA 40C; Siemens MedicalSolutions). The dVR images were created on aseparate workstation (Syngo Leonardo 2002B;Siemens Medical Solutions).

Prior to any kind of postprocessing, such as 3Dvisualization, a detailed review of the source im-ages that are the basis of CT angiography is man-

datory (18). These source images contain the en-tire information that is available from the data.Even the most sophisticated methods for 3D im-aging will lead to a distinct loss of data and, thus,potentially important information. Partial throm-bosis or calcification of an aneurysm will bemissed if the source images are not reviewed in ameticulous way. The interactive analysis of thesource images should be done on a workstationrather than by looking at hard copies in order todevelop a better perception of the course and therelationships of the intracranial arteries of inter-est. A wide window setting is necessary to enabledifferentiation between arteries filled with con-trast medium, bone, and calcifications (Fig 2).

Figure 1. Effect of FOV onimage quality of 3D imagingwith SSD in a patient with twoaneurysms (arrows in b–d) atthe bifurcation of the leftmiddle cerebral artery (MCA).(a) CT image shows the areascovered by three different val-ues for FOV: 200, 120, and 60mm2. (b) SSD image obtainedby reconstructing the data withan FOV of 200 mm2. Arteriesappear blurred. (c) SSD imageobtained by reconstructing thedata with an FOV of 120 mm2.Vascular anatomy is shownmore clearly than in b. ThisFOV contains all relevant intra-cranial arteries from which an-eurysms usually originate whileproviding good in-plane resolu-tion. Thus, we always use thisFOV for detection of intracra-nial aneurysms. (d) SSD imageobtained by reconstructing thedata with an FOV of 60 mm2.There is even better demonstra-tion of vascular anatomy thanin c. It is sometimes useful toperform a second reconstruc-tion with a narrow FOV such asthis when CT angiography isused for therapy planning andvery detailed information is re-quired.

RG f Volume 24 ● Number 3 Tomandl et al 639

Radio

Gra

phic

s

Figure 2. Importance of an adequate win-dow setting to demonstrate the intracranialarteries within the skull base. CT image ob-tained with a window width of 500 HU and acenter of 150 HU. Both ICAs can be clearlydifferentiated within the carotid canals (ar-rows).

Figure 3. Preparation of the volume foranalysis. (a) Posterosuperior image showslarge veins (arrowheads), which are typicallyalso visible in CT angiography of intracranialvessels and preclude an unobstructed view ofthe circle of Willis and the basilar artery (ar-row). (b) Left lateroposterior image showseasy elimination of the most obscuring venousstructures by using a clip plane (dotted whiteline) parallel to the clivus. (c) Posterosuperiorimage obtained after application of the clipplane (dotted white line) shows that thebasilar artery is demonstrated completely(arrowheads).


Radio

Gra

phic

s

Many aneurysms can already be detected byanalyzing the source images. Smaller aneurysmsbelow 5 mm in diameter are often difficult to de-tect on the basis of source images alone. There-fore, several methods for 2D and 3D postprocess-ing have been developed that allow more detailedanalysis and in addition an “angiographic” repre-sentation of CT angiography data.

Postprocessing ofCT Angiography Data

The basic principle of 2D and 3D postprocessingis to input cross-sectional images into a computerand thereby to create a so-called volume. One canimagine that this procedure is like putting backtogether the pieces of a sliced potato. Once a vol-ume is created, several methods for 2D and 3Dvisualization exist (19). The easiest way to ana-lyze a volumetric data set is multiplanar reforma-tion (MPR), in which from a given angle of view aplane is reconstructed in a defined depth of thevolume. This way it is possible to create coronal,axial, sagittal, as well as any kind of oblique sec-tions. The quality of the reconstructions dependson the voxel size. With the use of isometric data(ie, voxels have the same depth, length, andheight), all images are of the same quality as thebasic source images (20). In contrast to MIP andthe 3D methods discussed later, the recon-structed planes contain all information that iscontained in the source images. Therefore, MPRshould always be the method of first choice forthe further examination of CT angiography data(21).

To create useful “angiographic” representa-tions from CT angiography data, it is always nec-essary to eliminate disturbing structures from thevolume in order to ensure an unobstructed viewof the circle of Willis and its related arteries. For

this purpose, several graphical tools exist that varydepending on the software of the workstationused. To eliminate the straight sinus and otherveins that always prohibit an unobstructed view ofthe basilar artery, a so-called clip plane can beapplied parallel to the clivus (Fig 3). This kind ofdata manipulation is always necessary when MIPor 3D visualization of CT angiography data isperformed by using one of the following methods.

Maximum Intensity ProjectionThe term maximum intensity projection (MIP)means that from any given angle of view only thebrightest voxels of a volume are collected andused to create an image (22). Therefore, MIP isnot a 3D method, as it creates 2D images inwhich voxels from different locations within thevolume are collapsed into one plane. Thus, depthinformation is lost and it is not possible to tellwhether a structure is located in the front or backon the basis of a single MIP image. Because calci-fications and bone are brighter than contrast ma-terial–filled arteries, it is possible to differentiatelevels of attenuation (eg, to recognize a calcifiedartery).

The use of MIP as a method to create CT an-giograms is limited due to the fact that the skullbase has a much higher attenuation than the in-tracranial arteries and therefore has to be elimi-nated when MIP is used for image reconstruction(23). When dealing with intracranial aneurysms,it is often not possible to clearly depict the rela-tions of the aneurysm to its adjacent arteries. Fur-thermore, with the use of MIP, small aneurysmswill often be missed as they are eclipsed by thesignal of their parent vessels averaged into thesame 2D plane (Fig 4).

Figure 4. Influence of 3D visualization techniques on the detection of intracranial aneurysms. (a) MIP image (su-perior view) shows the bifurcation of the left MCA (arrow). Owing to the lack of depth information, the image doesnot allow visualization of two aneurysms at this site. (b, c) SSD (b) and dVR (c) images (superior views) of the bifur-cation of the left MCA show the two aneurysms (arrows).


Radio

Gra

phic

s

Figure 5. Analysis of CT angiography data with MPR and thin-section MIP. (a–c) Sagittal (a), coro-nal (b), and axial (c) MPR images show a small aneurysm at the bifurcation of the right MCA (arrow).Note the large intracerebral hematoma (arrowheads in a), which is usually not demonstrated on thresh-old-based 3D images. (d–f) Sagittal (d), coronal (e), and axial (f) MIP images obtained with thin sec-tions of 20 mm show the aneurysm more clearly (arrow) and show the intracerebral hematoma as well(arrowheads in d).


Radio

Gra

phic

s

In contrast to the two methods for 3D visual-ization described later, MIP is not threshold de-pendent and therefore is relatively easy to use. Ithas to be kept in mind that MIP uses only about10% of the information contained in a given vol-ume. In our experience, MIP is of minor use forthe creation of CT angiograms in order to searchfor and analyze aneurysms but is often very help-ful when used interactively on the workstation inthin sections of about 10–20 mm in addition toMPR (24) (Fig 5). In contrast to threshold-de-pendent methods, smaller arteries are displayedwithout user interaction (22).

Threshold-dependentMethods for 3D VisualizationShaded surface display (SSD) and direct volumerendering (dVR) are more difficult to use thanMIP. They require the user to define thresholdsfor the selection of voxels on the basis of their at-tenuation (measured in Hounsfield units). ForSSD, typically upper and lower thresholds aredefined and from a chosen angle of view the firstlayer of voxels with an attenuation within the de-fined parameters is displayed. Therefore, the im-ages show the surface of these structures and pro-vide valuable information about the 3D shape ofan object (25). On the other hand, all structures

are shown in the same color and informationabout the attenuation of a structure is lost com-pletely. For example, it is not possible to see calci-fications within an artery on SSD images. SinceMIP retains information about the attenuation ofobjects yet does not allow the depth perceptionprovided by SSD, both methods may be used tocomplement one another.

The definition of the thresholds is performedinteractively by the user and significantly influ-ences the appearance of the vascular structures(7) (Fig 6). Setting the lower threshold to a lowvalue (eg, 100 HU) will result in an image show-ing many vascular structures, including the veinsand small arteries. When the lower threshold isincreased (eg, to 200 HU), structures of low at-tenuation such as intracranial veins and small ar-teries will disappear completely and the majorarteries will appear smaller. The “ideal” thresholdto depict intracranial arteries has to be found in-teractively and depends on several parameters,including the injection rate of the contrast me-dium and cardiac output, both of which influencethe attenuation of the contrast material–filled vas-culature (26).

Figure 6. Threshold-dependent 3D visualization with SSD. (a) Superoposterior view obtainedwith a lower threshold of 100 HU shows smaller arteries like the left PICA (arrow) and venousstructures (arrowheads). (b) Superoposterior view obtained by increasing the lower threshold to200 HU shows arteries that appear thinner compared with those in a and even demonstrate dis-continuities (arrow). The venous structures are nearly eliminated (arrowheads), resulting in a lesscomplex image.


Radio

Gra

phic

s

From the point of view of a computer scientist,MIP and SSD are types of “volume rendering” inwhich only one layer of voxels is used for visual-ization. In the medical literature, the term volumerendering is reserved for the technique describednext, in which all voxels of a volume are consid-ered for visualization (27).

Direct volume rendering (dVR) is the mostsophisticated method for 3D visualization. Thebasic principle is to select several groups of voxelsaccording to their attenuation in Hounsfield unitsand to assign them a color and a so-called opacity(28,29) (Fig 7). When dVR is used to create CTangiograms, the voxels of high attenuation con-taining information about bony structures areselected separately from those voxels with an at-

tenuation between 100 and 300 HU containinginformation about contrast-enhanced vascularstructures. This selection allows the creation of3D images showing red arteries and white bone.A high opacity will lead to images that look simi-lar to those produced by SSD. Use of a low opac-ity can result in the creation of transparent objects(eg, it is possible to make intracranial arteries vis-ible beneath a layer of skull bone) (Fig 8). Select-ing only a small group of voxels with a high opac-ity allows creation of a “virtual endoscopic” viewin which the thin layer of voxels resembles thevessel wall (Fig 8e) (30).

There are so many possibilities to create im-ages that it is impossible to compare studies ofCT angiography with volume rendering from dif-ferent institutions. As with SSD, the appearanceof the intracranial arteries strongly depends onthe selected thresholds. Lower values will show

Figure 7. Basic principles of volume rendering. Groups of voxels are selected according totheir Hounsfield unit values. Every group has its own color and opacity. A low opacity makesthe objects transparent. dVR image (superior view) of a patient with two aneurysm clips (a)and photograph of the workstation screen (b). The voxels representing the metal clips (ar-rows in a) are colored blue with a high opacity. Bone (voxels between 200 and 2,000 HU) iscolored white with an opacity of 49%. Finally, the voxels between 90 and 300 HU that con-tain the vascular information are colored red with a 50% opacity.

‹Figure 8. Different possibilities for examining an aneurysm of the left MCA with dVR. (a) Frontal image obtainedwithout shading. (b) Frontal image obtained with shading (addition of an artificial light source), which gives the ob-jects more depth. (c) Frontal image obtained by selecting only a small group of voxels with a low opacity. The vesselsappear transparent, thus allowing visualization of a branch of the MCA running behind the aneurysm (arrow).(d) Left frontolateral transparent image allows the orifice of the feeding artery to be seen through the aneurysm (ar-row). (e) Frontal “virtual endoscopic” image, obtained by selecting only a small group of voxels with a high opacity,shows a pseudoconnection between the aneurysm and an adjacent artery (arrow). This “kissing vessel” artifact is apartial volume problem that is often seen on CT angiograms of intracranial aneurysms. (f) Anterocaudal image ob-tained with high opacity shows the close relationship of the aneurysm to the lower branch of the MCA (arrow).


Radio

Gra

phic

s


Radio

Gra

phic

s

Figure 9. Standard 3D projections ob-tained by using dVR interactively on theworkstation (left) and diagrams of the cor-responding arterial anatomy (right). Thissystematic analysis is of extreme impor-tance, especially when an aneurysm is de-tected at first glance, to ensure that addi-tional aneurysms are not missed. ACA �anterior cerebral artery, PCA � posteriorcerebral artery, PCom � posterior com-municating artery, VA � vertebral artery.(a) Superior view of all of the intracranialarteries. In many cases, larger aneurysmsare immediately visible on this overview.(b) Posterior view of the basilar and verte-bral arteries. Aneurysms of the PICA andthe basilar artery tip can be detected onthis view. AICA � anterior inferior cer-ebellar artery. (c) Lateral view of the intra-cranial part of the ICA. Note that the ICAis partially obscured by osseous structures,making detection of aneurysms in this areadifficult with 3D images alone. (d) Unob-structed view of the MCA bifurcation ob-tained from a superior angle. (e) Unob-structed view of the anterior communicat-ing artery (ACom) obtained from a superiorangle. (f) Unobstructed view of the ante-rior communicating artery (ACom) and thebifurcation of the left MCA obtained froman inferior angle after elimination of theskull base by using a clip plane (Fig 3).


Radio

Gra

phic

s

more vessels including venous structures and thuslead to a more complex image. It very much de-pends on the experience of an individual user toselect these thresholds in an efficient way. To en-sure a good detection rate for aneurysms, it ismandatory to evaluate the 3D models in a stan-dardized way (Fig 9). The areas where aneurysmsare most likely to occur have to be inspected inparticular detail. This is extremely important be-cause there is more than one aneurysm in about20% of patients with SAH (31). Although it iscustomary to use red for the group of voxels con-taining the vascular structures, it is not clearwhether the colors have an influence on aneurysmdetection (Fig 10).

Unfortunately, the quality of dVR highly de-pends on many factors like the quality of theworkstation and the applied rendering algorithm(29). Therefore, dVR is not a unique method andavailable studies are still not comparable.

Detection of IntracranialAneurysms with CT Angiography

In a systematic review of studies published be-tween 1988 and 1998, the average sensitivity ofCT angiography for the detection of intracranialaneurysms was about 90%. Aneurysms with a sizeof 3 mm or below were detected in 61% of cases,whereas the sensitivity for aneurysms with a largerdiameter was 96% (32). More recent studiesfound even higher overall detection rates of up to

97%, and some authors already solely rely on thefindings of CT angiography in patients with SAH(21,33,34). Because the detection rate dependson technical requisites and user-dependent post-processing techniques, this is not yet acceptedpractice in most radiology departments.

For the increasing number of aneurysmstreated by using an endovascular approach, CTangiography may be used as a tool for therapeuticdecision making and therapy planning: A majoradvantage of CT angiography compared withDSA is the generation of 3D information on theexact anatomy of the intracranial arteries. In addi-tion to the mere detection of aneurysms, these 3Dmodels can be most helpful for therapy planningas well (21). When intravascular coiling is thetherapy chosen on the basis of the CT angio-graphic findings, additional pretherapeutic DSAis not necessary, as it is part of the coiling proce-dure. In addition, CT angiography can help pre-dict the ideal angle for the endovascular ap-proach. Information on the exact dimensions ofan aneurysm provided by CT angiography can beused to determine the diameter of the first coil(Fig 11). For the neurosurgeon, informationabout adjacent vascular structures and the possi-bility of simulating the intraoperative view priorto surgery are often helpful (Fig 12) (35).

Figure 10. Visualization of the intracranial arteries with 3D dVR performed by using differentcolors. Superior views show the arteries colored red (a) and blue (b). It is not known whether thecolors affect the detection rate of intracranial aneurysms with CT angiography.


Radio

Gra

phic

s

Figures 11, 12. (11) Use of CT angiography for planning endovascular therapy. (a) Anterior 3DdVR image obtained with a high opacity shows an irregular aneurysm of the left intracranial ca-rotid bifurcation (arrow). (b) Anterior transparent dVR image obtained with a low opacity showsthe measured diameters of the dome and neck of the aneurysm (arrow). The maximal diameter ofthe dome was almost 7 mm. (c) Posteroanterior DSA image of the left ICA obtained after place-ment of the first coil. Because the exact measurements of the aneurysm (arrow) were determinedwith CT angiography, a 7-mm-diameter platinum coil was used first. (12) Use of CT angiographyfor planning surgical therapy in a patient with two small aneurysms at the bifurcation of the rightMCA. Arrows � aneurysms. (a) Three-dimensional dVR image obtained with a high opacityshows anatomic information nearly identical to the intraoperative findings (c). (b) TransparentdVR image obtained with a low opacity shows the orifice of the main stem of the MCA (arrow-head). Such images are often helpful in demonstrating the anatomy behind the vascular structures.(c) Photograph shows the intraoperative findings.


Radio

Gra

phic

s

Perimesencephalic hemorrhage is a distinctentity characterized by subarachnoid bleedingthat is confined to the peripontine and perimesen-cephalic cisterns. A source of bleeding will befound in only 5% of these patients (36). Althoughthe cause remains unclear in the other patients,their prognosis is usually excellent and rebleedingis uncommon. It has been proposed that thesepatients should be investigated with CT angiogra-phy alone (37). As CT angiography does not al-low exclusion of nonaneurysmal lesions such asdural arteriovenous fistulas or small angiomas,this management would lead to patients beingsubjected to both CT angiography and DSA in95% of cases (38). Therefore, we would performDSA as the primary and only study in these pa-tients.

Shortcomings andPitfalls of CT Angiography

When performing CT angiography for the detec-tion and therapy planning of intracranial aneu-rysms, knowledge about several potential pitfallsis essential. The in-plane resolution of CT is rela-tively high depending on the FOV that is used forsource image reconstruction; for example, a FOVof 120 mm2 combined with a 512 matrix results

in an in-plane pixel size of 0.23 � 0.23 mm2. Thethrough-plane resolution, which is defined by thesection thickness, typically lies around 1 mm. Inorder to produce more homogeneous reconstruc-tions, the section images are usually recon-structed in overlapping steps of 0.5 mm or less.Therefore, small perforating arteries with a diam-eter below 0.5 mm are not visible on CT angio-grams (39). It is often difficult or even impossibleto differentiate the infundibular dilatation of theorigin of an artery from an aneurysm, which canlead to false-positive results (7,40) (Fig 13a).This problem frequently occurs concerning theorigin of the posterior communicating artery fromthe ICA. Often, the situation can be clarified bylowering the thresholds on the 3D images (Fig13b) or by careful inspection of the source im-ages.

In about 10% of cases, it will not be possible tosee a clear margin between an aneurysm and adja-cent arteries, which results in the erroneous im-pression that a connection between these vascularstructures might exist. This “kissing vessel” arti-fact (41) is more likely to appear with larger aneu-rysms (Fig 14). During daily clinical work, this

Figure 13. Infundibular dilatation at the origin of the right PICA. (a) SuperoposteriordVR image shows an aneurysmal structure along the right vertebral artery (arrow). (b) Su-peroposterior dVR image obtained by lowering the threshold for the group of voxels thatcontain the vascular information shows that the structure is an infundibular dilatation at theorigin of the right PICA (large arrow). Note that some small arteries like the left PICA (largearrowhead) and the right anterior inferior cerebellar artery (small arrow, small arrowhead)have also become visible.


Radio

Gra

phic

s

Figure 14. Kissing vessel artifact. (a) CT angiogram (superior view) shows an aneurysmof the left anterior communicating artery (arrowhead) adjacent to the right ICA (arrow).An aneurysm at the bifurcation of the left MCA is also seen. (b) Right frontal dVR imageshows a large connection (arrow) between the right ICA and the aneurysm (arrowhead).(c) Transparent dVR image clearly shows the connection (arrow). Arrowhead � aneurysm.(d, e) Transparent (d) and virtual endoscopic (e) dVR images show a hole (arrow) betweenthe right ICA and the aneurysm. (f) DSA image shows no connection between the right ICAand the aneurysm (arrow). A separate angiogram was obtained for each carotid artery, andthe images were combined manually. To demonstrate the lack of a connection, the spacebetween the right ICA and the aneurysm is shown as larger than it actually was.


Radio

Gra

phic

s

artifact is easily recognized by following thecourse of an artery adjacent to an aneurysm andshould not cause problems in most cases whenhigh-opacity dVR or SSD is used for display.When transparent low-opacity dVR images areused and even more so when “virtual endoscopy”is performed (42), the closely related vascularstructures will show “holes” that can easily bemisinterpreted as connections. Therefore, virtualendoscopy is of minor use for both investigationand therapy planning of intracranial aneurysms.

The time that elapses between the arterial andthe venous phase of a contrast material bolusflowing through the intracranial circulation isabout 5–6 seconds. Therefore, even with four-row multisection scanners, it is not possible toproduce pure arterial phase CT angiography.Thus, the depiction of venous structures cannotbe avoided, and when they appear adjacent toarteries they can sometimes be mistaken for aneu-

rysms (Fig 15). On the other hand, informationabout the location of intracranial veins may be ofinterest for surgical treatment planning (43). Thelatest scanner prototypes, which are equippedwith up to 16 detector rows, accomplish data ac-quisition even faster, and pure arterial phase CTangiography of the intracranial vasculature willsoon become possible (44).

Aneurysms involving the skull base often donot show very well on 3D images. Analysis ismore easily performed by using the section im-ages. This is especially true when there is intraan-eurysmal thrombosis or calcification of the aneu-rysm wall. These important findings are easilyrecognized on the section images but are oftennot seen even when dVR is used for 3D visualiza-tion (Fig 16).

Figure 15. Obscuring venous structure in a patient with SAH. Superior (a) and posterior (b)CT angiograms show only part of the right MCA owing to an adjacent vascular structure (arrow).DSA was performed to exclude an aneurysm and showed normal findings.

Figure 16. Partially thrombosed and calcified aneurysm of the right ICA. (a) dVR image (supe-roposterior view) shows parts of the aneurysm (arrow) but does not allow identification of its ori-gin. (b) Source CT image shows the relationships between the perfused part of the aneurysm (�),the thrombosed part (small arrow), the calcified wall (arrowheads), and the ICA (large arrow).


Radio

Gra

phic

s

Patients with clipped aneurysms represent aspecific problem. Beam hardening artifacts pro-duced by the aneurysm clips preclude a clear de-piction of nearby intracranial arteries dependingon which material the clips are made of (39).Therefore, CT angiography is of minor value forthe postoperative follow-up of patients after aneu-rysm surgery in most cases. When CT angiogra-phy is used for this purpose, it is mandatory tocarefully inspect the source images prior to 3Dvisualization to check for the occurrence of arti-facts around the clip (Fig 17).

DiscussionThe increasing quality of CT provides excellentsource images, and there are already reports ofcases where CT angiography showed aneurysmsthat were not seen at standard DSA (Fig 18) (7).Since methods for standardized and reproducible3D visualization are not available as yet, the realvalue of CT angiography for the detection andevaluation of intracranial aneurysms remains un-clear despite the wealth of studies focusing on thistopic. The use of any commercially availableworkstation for postprocessing requires intensivefamiliarization with both the techniques of 3Dvisualization as well as the different manipulationtools available on the workstation used. Their

actual value strongly depends on the individualexperience of the investigator (19). For a mean-ingful delineation of the target structures, correctadjustment of transfer functions as well as effi-cient interactive manipulation of volume data arenecessary to make the aneurysms clearly visible.Therefore, studies performed so far evaluate spe-cific systems and the results depend on the expe-rience of individual users. Thus, the results fromthese different studies cannot be readily com-pared.

In order to allow a comparison of scientificstudies, it is essential to eliminate external influ-ences and to reduce the number of uncontrolledvariables as much as possible. A model for a stan-dardized postprocessing algorithm has been pro-posed by our research group (45). Our approachis based on an automated 3D image reconstruc-tion generated by a remote high-end workstation.CT data are transferred to the remote system viathe Internet, and standardized video sequencesare returned. The idea to produce standardizedvideos based on high-quality volume renderingcould also be realized locally on modern work-stations. Such a software tool would be a usefuladdition to widely used CT angiography equip-ment by allowing the reproduction of studieswithin a department and the comparison of re-sults between different institutions.

Figure 17. CT angiography of a patientwith two intracranial aneurysm clips.(a) Three-dimensional dVR image (poste-rior view) shows an apparent occlusion ofthe distal basilar artery (arrow). Arrow-heads � aneurysm clips. (b, c) Sagittal (b)and axial (c) MPR images show that theapparent occlusion is due to a beam hard-ening artifact (arrows) caused by the aneu-rysm clips (arrowheads in c).


Radio

Gra

phic

s

The example of the aneurysm involving theskull base (Fig 16) demonstrates that althoughthe CT angiography data contained informationabout thrombosis and calcification, the visualiza-tion strategy applied was not sufficient to depictthe lesion accurately. Thus, it would be useful tofind methods for improved depiction of aneu-rysms located within the cavernous sinus and nearthe skull base that are clinically applicable. Forthis purpose, special postprocessing techniquesaiming at the suppression of irrelevant bonystructures are required. So far, two solutions forthis problem have been proposed. The first isbased on the acquisition of two separate sets ofdata–one before and one after the administrationof contrast media–which then allow digital sub-traction of the images, resulting in “CT-DSA”(4,46). Patients have to lie absolutely still duringscanning and are exposed to a greater amount ofradiation. The second relies on the automatedseparation of bony and vascular structures (47).This approach is technically even more challeng-ing, since threshold-based elimination of the boneis difficult due to partial volume effects.

MR angiography is another possibility for thenoninvasive investigation of intracranial aneu-rysms. Its main advantage when compared to CTangiography is that the bone does not disturb theimages. The reported sensitivity of MR angiogra-phy for the detection of intracranial aneurysms iscomparable to that of CT angiography for aneu-rysms with a diameter greater than 3 mm but issignificantly lower for smaller aneurysms (32). Incontrast to CT angiography, for which at leastdata acquisition is carried out with relative consis-

tency throughout the published studies, the pro-tocols for MR angiography show a wide variabil-ity. An interesting application for MR angiogra-phy is the follow-up investigation of patients whohave been treated with platinum coils (48). WhereasCT angiography is not useful in these patients dueto severe beam hardening artifacts, MR angiogramsare not degraded by the coils (49).

In 2002, the results of the International Sub-arachnoid Aneurysm Trial (ISAT) (50) showedthat for patients with a ruptured intracranial an-eurysm for which both endovascular coiling andneurosurgical clipping were therapeutic options,after 1 year the outcome in terms of survival with-out disability was significantly better with endo-vascular coiling. This will lead to an increasingnumber of aneurysms being treated with an endo-vascular approach.

CT angiography can be performed immedi-ately following the confirmation of SAH withnonenhanced CT. Three-dimensional informa-tion on the aneurysm that caused the bleedingcan be obtained within a few minutes and usedfor therapy planning. The findings and therapeu-tic options can then be discussed among neuro-surgeons and endovascular neurointerventional-ists. Once coiling is the therapeutic option cho-sen, additional DSA is not necessary as it is partof the procedure. If endovascular treatment is notpursued, DSA may be necessary in addition toCT angiography, depending on the experience ofa specific institution.

Figure 18. Aneurysm of the basilar artery. (a) Superoposterior dVR image shows a 2-mm-diameter aneurysm ofthe basilar artery at the origin of the right superior cerebellar artery (arrow). (b) On frontal (left) and lateral (right)DSA images, the aneurysm is barely visible (arrow in left image). It was recognized only after the CT angiographyresults were known.


Radio

Gra

phic

s

The resulting algorithm is shown in Figure 19.It takes into account the fact that the quality ofCT angiography may vary even within a depart-ment, depending on the experience of the investi-gator. Negative CT angiography findings in a pa-tient with SAH must always be corroborated withDSA.

ConclusionsCT angiography is a promising method for bothdetection and therapy planning of intracranialaneurysms. Recent technical developments suchas the introduction of multisection CT and high-resolution dVR provide CT angiographic imagesof increasing quality. For institutions with ampleexperience using state-of-the-art equipment, itmay be safe to rely on the findings provided byCT angiography alone for both therapeutic deci-sions and therapy planning. As long as CT an-giography is not a standardized method with re-gard to postprocessing and 3D visualization, it isnot yet possible to clearly define the real value ofCT angiography with respect to the detection rateof intracranial aneurysms. Future developmentsshould focus on standardized 3D visualizationbased on workstations that are widely used. Onlythen will reliable comparative studies concerningthe sensitivity and specificity of the method bepossible.

References1. Cloft HJ, Joseph GJ, Dion JE. Risk of cerebral angiog-

raphy in patients with subarachnoid hemorrhage, ce-rebral aneurysm, and arteriovenous malformation: ameta-analysis. Stroke 1999; 30:317–320.

2. Alberico RA, Patel M, Casey S, Jacobs B, MaguireW, Decker R. Evaluation of the circle of Willis withthree-dimensional CT angiography in patients withsuspected intracranial aneurysms. AJNR Am J Neuro-radiol 1995; 16:1571–1578; discussion 1579–1580.

3. Korogi Y, Takahashi M, Katada K, et al. Intracranialaneurysms: detection with three-dimensional CT an-giography with volume rendering—comparison withconventional angiographic and surgical findings. Ra-diology 1999; 211:497–506.

4. Imakita S, Onishi Y, Hashimoto T, et al. SubtractionCT angiography with controlled-orbit helical scan-ning for detection of intracranial aneurysms. AJNRAm J Neuroradiol 1998; 19:291–295.

5. Ogawa T, Okudera T, Noguchi K, et al. Cerebral an-eurysms: evaluation with three-dimensional CT an-giography. AJNR Am J Neuroradiol 1996; 17:447–454.

6. Strayle-Batra M, Skalej M, Wakhloo AK, ErnemannU, Klier R, Voigt K. Three-dimensional spiral CTangiography in the detection of cerebral aneurysm.Acta Radiol 1998; 39:233–238.

7. Kato Y, Katada K, Hayakawa M, et al. Can 3D-CTAsurpass DSA in diagnosis of cerebral aneurysm? ActaNeurochir (Wien) 2001; 143:245–250.

8. Young N, Dorsch NW, Kingston RJ, Markson G,McMahon J. Intracranial aneurysms: evaluation in200 patients with spiral CT angiography. Eur Radiol2001; 11:123–130.

9. Mallouhi A, Felber S, Chemelli A, et al. Detectionand characterization of intracranial aneurysms withMR angiography: comparison of volume-renderingand maximum-intensity-projection algorithms. AJRAm J Roentgenol 2003; 180:55–64.

10. Addis KA, Hopper KD, Iyriboz TA, et al. CT angiogra-phy: in vitro comparison of five reconstruction methods.AJR Am J Roentgenol 2001; 177:1171–1176.

11. Choyke PL, Yim P, Marcos H, Ho VB, Mullick R,Summers RM. Hepatic MR angiography: a multiob-server comparison of visualization methods. AJRAm J Roentgenol 2001; 176:465–470.

12. Tsuchiya K, Katase S, Yoshino A, Hachiya J, YodoK. Preliminary evaluation of volume-rendered three-dimensional display of time-of-flight MR angiographyin the diagnosis of intracranial aneurysms. Neuroradi-ology 2001; 43:633–636.

13. Katz DA, Marks MP, Napel SA, Bracci PM, RobertsSL. Circle of Willis: evaluation with spiral CT angiog-raphy, MR angiography, and conventional angiogra-phy. Radiology 1995; 195:445–449.

14. Chen CJ, Chen ST. Extracranial distal aneurysm ofposterior inferior cerebellar artery. Neuroradiology1997; 39:344–347.

15. Mahesh M. Search for isotropic resolution in CTfrom conventional through multiple-row detector.RadioGraphics 2002; 22:949–962.

16. Kirchner J, Kickuth R, Laufer U, Noack M, Lier-mann D. Optimized enhancement in helical CT: ex-periences with a real-time bolus tracking system in628 patients. Clin Radiol 2000; 55:368–373.

17. van Hoe L, Marchal G, Baert AL, Gryspeerdt S,Mertens L. Determination of scan delay time in spiralCT-angiography: utility of a test bolus injection.J Comput Assist Tomogr 1995; 19:216–220.

Figure 19. Algorithm for use of CT an-giography in patients with SAH.


Radio

Gra

phic

s

18. Rubin GD. Three-dimensional helical CT angiogra-phy. RadioGraphics 1994; 14:905–912.

19. Udupa JK. Three-dimensional visualization and anal-ysis methodologies: a current perspective. Radio-Graphics 1999; 19:783–806.

20. Rydberg J, Buckwalter KA, Caldemeyer KS, et al.Multisection CT: scanning techniques and clinicalapplications. RadioGraphics 2000; 20:1787–1806.

21. Villablanca JP, Hooshi P, Martin N, et al. Three-di-mensional helical computerized tomography angiog-raphy in the diagnosis, characterization, and manage-ment of middle cerebral artery aneurysms: compari-son with conventional angiography and intraoperativefindings. J Neurosurg 2002; 97:1322–1332.

22. Prokop M, Shin HO, Schanz A, Schaefer-ProkopCM. Use of maximum intensity projections in CTangiography: a basic review. RadioGraphics 1997;17:433–451.

23. Napel S, Marks MP, Rubin GD, et al. CT angiogra-phy with spiral CT and maximum intensity projec-tion. Radiology 1992; 185:607–610.

24. Tomandl BF, Klotz E, Handschu R, et al. Compre-hensive imaging of ischemic stroke with multisectionCT. RadioGraphics 2003; 23:565–592.

25. Liang EY, Chan M, Hsiang JH, et al. Detection andassessment of intracranial aneurysms: value of CTangiography with shaded-surface display. AJR Am JRoentgenol 1995; 165:1497–1502.

26. Addis KA, Hopper KD, Iyriboz TA, Kasales CJ, LiuY, Wise SW. Optimization of shaded surface displayfor CT angiography. Acad Radiol 2001; 8:976–981.

27. Hastreiter P, Rezk-Salama C, Tomandl B, EberhardtKE, Ertl T. Interactive and intuitive visualization ofsmall and complex vascular structures in MR andCT. In: Proceedings of the Engineering in Medicineand Biology Society (EMBS), Institute of Electricaland Electronics Engineers. Vol 20. New York, NY:Springer-Verlag, 1998; 532–535.

28. Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fish-man EK. Three-dimensional volume rendering ofspiral CT data: theory and method. RadioGraphics1999; 19:745–764.

29. Tomandl BF, Hastreiter P, Rezk-Salama C, et al. Lo-cal and remote visualization techniques for interactivedirect volume rendering in neuroradiology. Radio-Graphics 2001; 21:1561–1572.

30. Bontozoglou NP, Spanos H, Lasjaunias P, Zarifis G.Intracranial aneurysms: endovascular evaluation withthree-dimensional-display MR angiography. Radiol-ogy 1995; 197:876–879.

31. Kaminogo M, Yonekura M, Shibata S. Incidence andoutcome of multiple intracranial aneurysms in a de-fined population. Stroke 2003; 34:16–21.

32. White PM, Wardlaw JM, Easton V. Can noninvasiveimaging accurately depict intracranial aneurysms? Asystematic review. Radiology 2000; 217:361–370.

33. Velthuis BK, Van Leeuwen MS, Witkamp TD,Ramos LM, Van Der Sprenkel JW, Rinkel GJ. Com-puterized tomography angiography in patients withsubarachnoid hemorrhage: from aneurysm detectionto treatment without conventional angiography.J Neurosurg 1999; 91:761–767.

34. Villablanca JP, Martin N, Jahan R, et al. Volume-rendered helical computerized tomography angiogra-phy in the detection and characterization of intracra-nial aneurysms. J Neurosurg 2000; 93:254–264.

35. Gonzalez-Darder JM, Pesudo-Martinez JV, Feliu-Tatay RA. Microsurgical management of cerebral

aneurysms based in CT angiography with three-di-mensional reconstruction (3D-CTA) and withoutpreoperative cerebral angiography. Acta Neurochir(Wien) 2001; 143:673–679.

36. van Gijn J, Rinkel GJ. Subarachnoid haemorrhage:diagnosis, causes and management. Brain 2001; 124:249–278.

37. Velthuis BK, Rinkel GJ, Ramos LM, Witkamp TD,van Leeuwen MS. Perimesencephalic hemorrhage:exclusion of vertebrobasilar aneurysms with CT an-giography. Stroke 1999; 30:1103–1109.

38. Cloft HJ. Perimesencephalic hemorrhage. Stroke1999; 30:1979–1981.

39. Kato Y, Nair S, Sano H, et al. Multi-slice 3D-CTA:an improvement over single slice helical CTA for ce-rebral aneurysms. Acta Neurochir (Wien) 2002; 144:715–722.

40. Ng SH, Wong HF, Ko SF, et al. CT angiography ofintracranial aneurysms: advantages and pitfalls. Eur JRadiol 1997; 25:14–19.

41. Tomandl BF, Hastreiter P, Eberhardt KE, Rezk-Salama C, Nimsky C, Buchfelder M. The “kissingvessel”-artifact: a problem occurring in the visualiza-tion of intracranial aneurysms using volume renderingand virtual endoscopy (abstr). Radiology 1999;213(P):311.

42. Marro B, Valery CA, Bitard A, et al. Intracranial an-eurysm on CTA: demonstration using a transparencyvolume-rendering technique. J Comput Assist To-mogr 2000; 24:96–98.

43. Kaminogo M, Hayashi H, Ishimaru H, et al. Depict-ing cerebral veins by three-dimensional CT angiogra-phy before surgical clipping of aneurysms. AJNRAm J Neuroradiol 2002; 23:85–91.

44. Lell M, Wildberger JE, Heuschmid M, et al. CT-an-giography of the carotid artery: first results with anovel 16-slice-spiral-CT scanner. Rofo Fortschr GebRontgenstr Neuen Bildgeb Verfahr 2002; 174:1165–1169.

45. Tomandl B, Hastreiter P, Iserhardt-Bauer S, et al.Standardized evaluation of CT angiography with re-mote generation of 3D video sequences for the detec-tion of intracranial aneurysms. RadioGraphics 2003;23:e12.

46. Gorzer H, Heimberger K, Schindler E. Spiral CTangiography with digital subtraction of extra- and in-tracranial vessels. J Comput Assist Tomogr 1994; 18:839–841.

47. Moore EA, Grieve JP, Jager HR. Robust processingof intracranial CT angiograms for 3D volume render-ing. Eur Radiol 2001; 11:137–141.

48. Brunereau L, Cottier JP, Sonier CB, et al. Prospec-tive evaluation of time-of-flight MR angiography inthe follow-up of intracranial saccular aneurysmstreated with Guglielmi detachable coils. J ComputAssist Tomogr 1999; 23:216–223.

49. Nome T, Bakke SJ, Nakstad PH. MR angiography inthe follow-up of coiled cerebral aneurysms after treat-ment with Guglielmi detachable coils. Acta Radiol2002; 43:10–14.

50. Molyneux A, Kerr R, Stratton I, et al. InternationalSubarachnoid Aneurysm Trial (ISAT) of neurosurgi-cal clipping versus endovascular coiling in 2143 pa-tients with ruptured intracranial aneurysms: a ran-domised trial. Lancet 2002; 360:1267–1274.


Radio

Gra

phic

s

This article meets the criteria for 1.0 credit hour in category 1 of the AMA Physician’s Recognition Award. To obtaincredit, see accompanying test at http://www.rsna.org/education/rg_cme.html.

Documents

EDUCATION EXHIBIT CT Angiography of RadioGraphics ... · image quality of 3D imaging with SSD in a patient with two aneurysms (arrows in b–d)at the bifurcation of the left middle