| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 All authors: Department of Radiology, Innsbruck University Hospital, Anichstr. 35, 6020 Innsbruck, Austria.
Received January 21, 2002;
accepted after revision July 11, 2002.
Address correspondence to A. Mallouhi.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Three-dimensional time-of-flight MR angiography studies performed in 82 patients were retrospectively evaluated by two independent reviewers who were unaware of digital subtraction angiography findings, the standard of reference. Panoramic maximum-intensity-projection and volume-rendered angiograms were produced from each data set to investigate the presence of underlying aneurysms. Each detected aneurysm was then interactively evaluated with subvolume maximum-intensity-projection and targeted volume-rendering algorithms to evaluate aneurysm morphology and size. Aneurysm detection and characterization were evaluated by means of the receiver operating characteristic analysis, and aneurysm size was evaluated using the limits-of-agreement method. Image quality, aneurysm neck depiction, and vascular delineation were also compared between maximum-intensity-projection and volume-rendered images. The time required for the generation and interpretation of maximum-intensity-projection and volume-rendered images was assessed.
RESULTS. Volume rendering tended to improve the diagnostic
confidence (Az [area under the receiver operating
characteristic curve] = 0.95 vs Az = 0.90 for maximum
intensity projection) and yielded a considerable improvement in sensitivity
(89% vs 71% for maximum intensity projection), particularly in the detection
of small cerebral aneurysms. Regarding aneurysm morphology, volume rendering
performed significantly better than maximum intensity projection in lobulation
detection (p < 0.001) and slightly better in neck categorization
(p > 0.238). Limits-of-agreement analysis showed a trend toward
improved assessment of the aneurysm size by volume rendering (-0.31 ±
1.62 mm vs -1.27 ± 2.84 mm by maximum intensity projection). Overall
image quality and vascular delineation of involved vessels on volume-rendered
images were rated better than that obtained by maximum intensity projections
(p 
0.007 and p 0.001, respectively). Evaluation of
time-of-flight MR angiography data sets was significantly facilitated with
volume rendering (p < 0.001).
CONCLUSION. The volume-rendering technique facilitates the evaluation of cerebral time-of-flight MR angiography data sets and allows better detection and more reliable characterization of intracranial aneurysms than does maximum intensity projection.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cerebral MR angiography provides a 3D representation of the intracranial vasculature, which is, however, routinely captured in a two-dimensional perspective by the maximum-intensity-projection algorithm or multiplanar volume reformations that might obscure an abnormality because of vascular overlay or lead to misinterpretation of tortuous vascular structures as an aneurysm. The implementation of a volume-rendering algorithm to evaluate MR angiography data sets has been recently described by several investigators [13,14,15,16,17], but the algorithm has not yet gained clinical relevance. The purpose of this study was to compare MR angiography with volume-rendering and maximum-intensity-projection algorithms with digital subtraction angiography, the reference standard, in the detection and characterization of intracranial aneurysms and to compare both rendering algorithms for estimating aneurysm size, aneurysm neck detectability, image quality, and delineation of the involved arteries.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Technique and Image Analysis of Digital Subtraction Angiography
Intraarterial digital subtraction angiography was performed using a
transfemoral approach. After the injection of 7 mL of nonionic contrast
material at a rate of 5 mL/sec, selective internal carotid angiograms were
acquired bilaterally in the anteroposterior, lateral, and oblique views.
Vertebrobasilar angiograms were obtained after the injection of 6 mL of
nonionic contrast material at a flow rate of 4 mL/sec. When applicable,
angiography was completed with additional rotational or 3D angiograms to
optimize the visualization of the vascular anatomy, detection of an underlying
aneurysm, and categorization of detected aneurysms. For 13 patients with 16
aneurysms, 3D digital subtraction angiography was available.
Digital subtraction angiograms, the standard of reference, were assessed by two neurointerventional radiologists in consensus for the presence of an aneurysm, the morphology of the aneurysm pouch and neck, and aneurysm size using digital calipers.
MR Imaging Technique
All MR examinations were performed on a 1.5-T system (Magnetom Vision;
Siemens Medical Systems, Erlangen, Germany). The receiver coil was a standard
head coil. Three overlapping slabs, each including 36 partitions, were used to
perform time-of-flight MR angiography using 3D fast imaging with steady-state
precession sequences with the following parameters: TR/TE, 35/7; flip angle,
20°; field of view, 20 cm; section thickness, 0.8 mm; matrix, 512 x
256; and acquisition time, 6.4 min.
Reconstruction Parameters
The source images were transferred to an independent workstation (Ultra 60;
Sun Microsystems, Mountain View, CA) running the Advantage Windows software
(version 4.0; General Electric Medical Systems, Milwaukee, WI).
Maximum-intensity-projection and volume-rendered images were created for each
data set to show a panoramic view of the intracranial arteries. Initially,
each maximum-intensity-projection and volume-rendered panoramic image was
interactively rotated around axial and coronal axes to examine the presence of
harbored aneurysms. Further arbitrary oblique projections were also obtained
when necessary to overcome vascular overlapping. At a second step, each
detected aneurysm on the panoramic images was separately evaluated using
subvolume maximum-intensity-projection and targeted volume-rendered images to
examine the aneurysm morphology and size.
Subvolume maximum intensity projections were obtained according to a standardized protocol using a slab with a similar width (20 mm) that was positioned on the multiplanar volume reformation images in all cases. The slab was centered on the aneurysm center, and its direction was interactively modified to present coronal, sagittal, axial, or oblique projections selected to encompass the aneurysm and the involved vessels. Targeted volume-rendered images were obtained by centering the display field of view on the aneurysm center. Likewise, oblique projections were selected to show the aneurysm and its anatomic relationships and morphologic characteristics. Overlying arteries were cut off.
Optimal display parameters (window width and level, opacity, and brightness) for generating volume-rendered images of the intracranial arteries were defined by consensus between two reviewers after a preclinical study in which different models of volume-rendered images from MR angiographic data sets were tested. The opacity transfer function was determined to maximize the visualization of the cerebral vessels and the perception of small vessels by using a low-to-high opacity curve type, the slope of which ranged from 180-250 to 350-380 MR intensity units. Opacity and brightness values of 100% were assigned to the selected material.
The operator-dependent time required to display a panoramic image of the intracranial arteries was within 1 min for volume rendering and less than 20 sec for maximum intensity projection. Each modification of parameter or any change in volume orientation was performed within 1-2 sec for volume rendering. For maximum intensity projection, in comparison, each change of volume orientation or modification of the subvolume slab took an additional 1-3 sec.
The image rendering procedure was performed independently by two radiologists trained in MR imaging and 3D reconstructions. For each data set, maximum-intensity-projection and volume-rendered images were acquired in a different randomized order to prevent a consistent bias in the interpretation based on a prior viewing of the same image set next to the other rendering algorithm. Both operators were unaware of clinical outcome and digital subtraction angiography findings and did not participate in image interpretation.
Image Analysis
After reconstruction, maximum-intensity-projection and volume-rendered
images were interactively evaluated by two senior radiologists (a
neuroradiologist and a neurointerventional radiologist) working independently
who were unaware of clinical outcome and digital subtraction angiography
findings. Although it is common to refer to source images of MR angiography,
each examining team (operator and reviewer) deliberately avoided assessing the
source images to achieve a net comparison of the performance of each rendering
algorithm.
Presence and morphology of aneurysms.—The diagnostic confidence in the presence of an intracranial aneurysm was scored using a 5-point ordinal scale on which 1 was used for definitely present; 2, probably present; 3, uncertain; 4, probably absent; and 5, definitely absent [18]. When an aneurysm was categorized as definitely present or probably present, the aneurysm pouch was categorized as 1, lobulated; 2, uncertain; or 3, smooth; and the aneurysm neck, as 1, narrow; 2, uncertain; or 3, wide.
Quantitative analysis.—Using digital calipers, aneurysm size, defined as the maximal spherical diameter between the neck and the fundus, was measured and compared with that measured on digital subtraction angiography. The mean size of each detected aneurysm was separately determined from the values measured by the two reviewers. In addition, the total time required to reach the final diagnosis, including the generation of images and their interpretation, was calculated by the reviewers for each reconstruction algorithm.
Semiquantitative analysis.—For further assessment of maximum-intensity-projection and volume-rendering algorithms, images were analyzed according to a 4-point scale. General criteria included the impression of overall image quality assessed as 1, very good vascular visibility enabling detailed and reliable evaluation; 2, good vascular visibility enabling adequate evaluation; 3, average visibility with compromised evaluation; or 4, unsatisfactory visibility with inadequate evaluation. Depiction of the aneurysm neck was categorized as 1, sufficiently depicted with good analysis of relationship between the aneurysm and the parent vessel; 2, scarcely depicted with no clear relationship with the parent vessel; or 3, relationship with the parent vessel could not be determined. In addition, delineation of the parent vessel was scored as 1, well delineated; 2, vaguely delineated but with definite visualization of the vessel; or 3, the parent vessel was not shown.
Statistical Analysis
Data entry procedures and statistical analyses were performed using a
statistical software system (SPSS for Windows, version 10.0.7; Statistical
Package for the Social Sciences, Chicago, IL). In the first step, receiver
operating characteristic curves were generated to show the relative accuracy
of maximum-intensity-projection and volume-rendered images for the detection
of aneurysms and aneurysm lobulation by comparing the areas under the curves
(Az) of each reviewer. From the observed data points,
sensitivity and specificity were calculated on a per-aneurysm (the ability to
correctly identify all aneurysms) basis with the use of only "definitely
present" as positive and all other categories as negative. In the second
step of analysis, the Spearman's rank correlation coefficient test was used to
investigate the correlation between neck category depicted on
maximum-intensity-projection and volume-rendered images and that depicted on
digital subtraction angiography.
Next, interval data arising from mean aneurysm size were analyzed using the limits-of-agreement method [19] by determining the bias as the mean difference between methods and the error as ± 2 SDs. Interval data arising from evaluation duration were analyzed using a two-tailed paired Student's t test. Ordinal data arising from image quality, depiction of the aneurysm neck, and delineation of parent vessels were analyzed using Wilcoxon's signed rank test. Values for r of less than 0.05 were regarded as significant.
Finally, the kappa statistic
[20] was used for comparison
of observer performance for the categorization of the aneurysm pouch and neck
and for the assessment of image quality, neck depiction, and vascular
delineation. Interobserver agreement was considered as slight for 
0.2; fair, = 0.21-0.40; moderate, = 0.41-0.60; substantial,
= 0.61-0.80; or almost perfect, = 0.81-1.00.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Presence of Aneurysm
Three aneurysms that had an intraluminal signal void were not depicted on
MR angiography. These included a 20-cm large partially thrombosed aneurysm of
the internal carotid artery and two 2-mm aneurysms of the same dissected
posteroinferior cerebellar artery. The parametric receiver operating
characteristic models obtained for the detection of any aneurysm and the
results of calculating the Az values for both rendering
techniques are shown in Figure
1A. The reviewers achieved better performance for detection of
intracranial aneurysms with the volume-rendering algorithm. The
Az values for volume-rendered images were not
significantly greater than those for maximum-intensity-projection images. The
use of only "definitely present" as positive for the detection of
intracranial aneurysms yielded a substantial increase in sensitivity, with the
volume-rendering algorithm associated with a considerable improvement of the
negative predictive value (Table
1). When the definition of "positive" was enlarged to
include both definitely present and probably present, the advantage of volume
rendering decreased slightly (Table
2).
|
|
|
Considering only those aneurysms of 3 mm or larger, the two reviewers, on maximum-intensity-projection images, detected with certainty 28 (85%) and 29 (88%) of 33 aneurysms found on digital subtraction angiography. One and two aneurysms were not definitely detected by the two reviewers on volume-rendered images. Of 10 aneurysms smaller than 3 mm, seven and five aneurysms were definitely detected by the two reviewers on volume-rendered images compared with one and two aneurysms by the two reviewers on maximum-intensity-projection images (Fig. 2A,2B,2C).
|
|
|
Aneurysm Morphology
Volume-rendered images were more efficient than
maximum-intensity-projection images for the detection of aneurysm lobulation.
The receiver operating characteristic curves revealed significantly greater
Az values for volume-rendered images
(Fig. 1B). Furthermore, the
sensitivity and negative predictive value for depicting daughter lobules
increased significantly on volume-rendered images, yielding a substantial
increase in accuracy (Table
3).
|
|
In terms of categorization of aneurysm neck morphology, the reviewers found no significant differences between targeted volume-rendered and subvolume maximum-intensity-projection images (for all comparisons, p > 0.238; Wilcoxon's signed rank test) (Fig. 3A,3B,3C). Furthermore, Spearman's rank correlation coefficient test showed a statistically significant correlation between aneurysm neck category assessed on digital subtraction angiography and that assessed by reviewers of maximum-intensity-projection images (r = 0.64, p < 0.001; and r = 0.56, p = 0.001) and reviewers of volume-rendered images (r = 0. 82, p < 0.001; and r = 0. 79, p < 0.001).
|
|
|
Quantitative Image Analysis
The mean aneurysm size difference (bias) between the two reviewers was -0.1
mm with a 95% confidence interval (CI) of -0.5 to 0.3 mm for maximum intensity
projection and -0.4 mm (95% CI, -0.7 to -0.1 mm) for volume rendering,
suggesting excellent interobserver agreement. Therefore, the average of their
two interpretations was compared with the findings from digital subtraction
angiography (Fig.
4A,4B).
The limits-of-agreement analysis showed a trend toward improved assessment of
the aneurysm size by volume rendering. Good agreement was seen between volume
rendering and digital subtraction angiography, with a bias of -0.31 mm
(±1.62 mm); the agreement was less good between maximum intensity
projection and digital subtraction angiography, with a bias of -1.27 mm
(±2.84 mm).
|
|
The mean time required to evaluate a data set, including image rendering and interactive interpretation, differed significantly between reviewers of maximum-intensity-projection and volume-rendered images (p < 0.001 for all comparisons, Student's t test). For maximum-intensity-projection images, the two reviewers required a mean time of 5.27 min (95% CI, 4.77-5.77 min) and 4.31 min (95% CI, 3.74-4.87 min) for a data set. In comparison, for volume-rendered images the two reviewers required 2.35 min (95% CI, 2.01-2.69 min) and 3.11 min (95% CI, 2.70-3.52 min).
Semiquantitative Image Analysis
Comparisons of maximum-intensity-projection and volume-rendering algorithms
are summarized in Table 4.
Maximum intensity projection and volume rendering enabled the generation of
high-quality images of intracranial aneurysms with no vascular overlay because
of the interactive viewing of the vasculature from any orientation. However,
the 3D appearance on volume-rendered images was preserved and thus enabled a
more precise analysis of the relationship between the aneurysm pouch and the
parent vessels as well as the vessels arising from the pouch in all three
detected cases. Because of the availability of editing procedures, image
quality of subvolume maximum-intensity-projection and volume-rendered images
was not hampered in patients with subarachnoid hemorrhage. Both reviewers
considered the overall image quality with volume rendering significantly
greater than that with maximum intensity projection (p 0.007 for
all comparisons, Wilcoxon's signed rank test). However, the image quality of
subvolume maximum-intensity-projection images obtained to evaluate the
harbored aneurysms was not rated significantly lower (p = 0.035-0.157
for all comparisons) than that of targeted volume-rendered images.
|
Both subvolume maximum-intensity-projection and targeted volume-rendered
images provided sufficient depiction of the aneurysm neck in most cases (Fig.
2A,2B,2C).
Despite the absence of marked differences in the mode scores between the
reviewers (p = 0.013-0.109 for all comparisons), volume-rendered
images facilitated the identification and evaluation of the aneurysm neck
(Fig.
5A,5B,5C).
In terms of delineation of parent vessels, statistical analysis revealed
significantly better performance with volume rendering than with maximum
intensity projection (p 0.001) (Fig.
6A,6B,6C).
|
|
|
|
|
|
Interobserver Agreement
For both MR angiographic rendering techniques, some disagreement was
observed among the reviewers in the confidence level for the presence of an
aneurysm ( = 0.65 and 0.71 for maximum-intensity-projection and
volume-rendered images, respectively) and in the categorization of an aneurysm
pouch ( = 0.46 and 0.77 for maximum-intensity-projection and
volume-rendered images, respectively) and aneurysm neck ( = 0.60 and
0.84 for maximum-intensity-projection and volume-rendered images,
respectively). Kappa values for the assessment of image quality, aneurysm neck
depiction, and vascular delineation were 0.75, 0.56, and 0.83, respectively,
for maximum intensity projection and 0.67, 0.47, and 0.77 for volume
rendering. The disagreement was probably related to the variations involved in
the interactive reconstruction of maximum-intensity-projection and
volume-rendered images.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, two postprocessing algorithms of cerebral 3D time-of-flight MR angiography were compared: the routinely used maximum intensity projection and the less frequently used volume rendering. The application of the volume-rendering algorithm to cerebral 3D time-of-flight MR angiography contributed to the improved perceptibility of intracranial arteries and, consequently, to the improved detection and comprehensive characterization of intracranial aneurysms. The principle, as reported in previous articles [21, 22], is that the volume-rendering algorithm incorporates the entire data set into a 3D image, thus enabling the visualization of the vascular surface and intravascular details while preserving spatial relationships. Because volume rendering is based on the percentage classification technique, volume rendering provides accurate determination of the amounts of materials when the voxel consists of two or more materials being volume averaged. The volume-averaged voxels are included in the final image because the volume-rendering algorithm calculates a weighted sum of data from all voxels along a ray projected through the data set. Therefore, volume rendering improves the delineation of small-caliber vessels (parent and branch vessels) and the perception of a lobulated aneurysm surface and contributes to precision in assessing aneurysm size.
As opposed to volume rendering, the maximum-intensity-projection algorithm selects only the voxel with the highest signal intensity along a ray projected through the data set, of which only approximately 10% are displayed in the final image [23]. Thus, the maximum-intensity-projection algorithm does not take spatial location into account, displays the vasculature without surface or depth cues, and is affected by volume averaging [24]. In addition, volume-averaged voxels may be erroneously excluded from the final maximum-intensity-projection images, resulting in misrepresentation and decreased delineation of the small-caliber vessels.
Our study showed that MR angiography with the volume-rendering technique provides better visualization of intracranial aneurysms than does the maximum-intensity-projection algorithm. The calculated areas under the receiver operating characteristic curves indicated that the implementation of the volume-rendering algorithm augments, in general, the diagnostic confidence of cerebral time-of-flight MR angiography and, in particular, the discerning of very small aneurysms. Studies of the natural history of aneurysms show that the risk of rupture increases with aneurysm size [25,26,27,28,29] and that the annual rupture rate for aneurysms smaller than 10 mm is low (0.05%) [30]. Nevertheless, the detection of all harbored aneurysms, even those smaller than 3 mm, is important diagnostically because small aneurysms tend to enlarge in unpredictable spurts and, ultimately, to rupture. Indeed, seven of eight aneurysms smaller than 3 mm that did not cause an intraluminal signal void were detected with certainty by volume rendering compared with two aneurysms detected with maximum intensity projection.
Also important is the detection of daughter lobules, a substantial indicator of a prior rupture or the threat of a rupture [18, 31]. The recognition of these areas of weakness in the aneurysm pouch has an essential impact on treatment if rupture is to be avoided. In this context, we observed a significantly improved detectability of lobulated aneurysms by volume-rendered MR angiography because of a twofold advantage: the virtual embodiment of the aneurysm surface that allows better estimation of the aneurysm morphology, and the improved intraluminal delineation that allows better perception of small lobules. In contrast, the capability of maximum-intensity-projection images in the diagnosis of aneurysm lobulation was not adequate, as indicated by its low sensitivity and negative predictive value, despite a rather high specificity.
Another advantage of volume rendering over maximum intensity projection is that the rendering and interpretation times of targeted volume-rendered images are significantly less than those for subvolume maximum-intensity-projection images. Our empiric results indicate that the volume-rendering algorithm significantly facilitates the evaluation and interpretation of cerebral time-of-flight MR angiography. Although standard maximum-intensity-projection images (off-line rendered images around axial and coronal axes and subsequently viewed in a cine loop) of the intracranial arteries could be obtained in a relatively short time, lack of depth cue and uncorrectable vessel overlap may lead to disorientation of the operator or misinterpretation of findings. The disadvantages of the maximum-intensity-projection algorithm have been repeatedly reported [12, 18, 32,33,34]. To compensate for these disadvantages, we evaluated our MR angiography data sets with an interactive maximum-intensity-projection technique. However, this method, as previously pointed out by Adams et al. [12], requires that all projection rays of the entire image volume be recalculated for each manipulation, requiring a considerably longer reconstruction time. The extra rendering of subvolume maximum-intensity-projection images to highlight particular features of the aneurysm (size, pouch and neck morphology, and parent and exiting vessels) caused a further substantial delay in image evaluation. As opposed to maximum intensity projection, once a volume-rendered image of the entire data set has been created, it can be evaluated in real-time from any angle.
Whereas volume-rendered images share with subvolume maximum-intensity-projection images the advantage of good depiction of the aneurysm neck, the inherent technical properties of the volume-rendering algorithm provide an appreciable advantage with regard to improved overall image quality and vascular delineation of parent and arising vessels. The attractive and well-defined vasculature on volume-rendered images contributes to accuracy in assessing the relationship between the aneurysm and the involved branches. This accuracy may be of particular importance if 3D digital subtraction angiography, the state-of-the-art method, is not available for diagnostic workup of patients undergoing surgery or endovascular coiling.
Successful application of volume rendering, however, is predicated on the correct choice of display parameters (window width and level, opacity, and brightness), which remains a crucial subjective decision. In addition, as reported by several studies [12, 23, 24], interactive control over the volume-rendering transfer function and display parameters may increase operator variability. Generating a reliable volume-rendered angiogram can be achieved, in our experience, by interactive modification of the window level so that it concurs with the mean intensity of source data. To display distal, fine, low-intensity arteries, we used a decreased window level. A decreased window level caused an increase in the paravascular noise, which was ultimately cut off or filtered out by editing. Operator variability could be minimized if the rendering process is performed by a trained operator. Furthermore, combining technical and neuroradiologic expertise can improve the accuracy of interactive volume rendering of intracranial MR angiography.
In clinical practice, it is common to correlate rendered images with source images of MR angiography. This correlation is particularly important for volume rendering because an inappropriate threshold setting can result in loss of visualization of some intracranial arteries, particularly low-intensity small peripheral cerebral arteries, and thus result in a potential aneurysm being missed. However, the interpretation of the source images was deliberately excluded during image rendering and interpretation in order to perform a net comparative analysis, in which we assessed the independent ability of each rendering algorithm in the evaluation of intracranial aneurysms.
Time-of-flight MR angiography has inherent limitations in the depiction of intracranial aneurysms. The perception of aneurysms accompanied by spin saturation resulting from slow flow or turbulent flow [35] is severely limited. Furthermore, hemorrhage may obscure some segments of the intracranial arteries and thus limit the evaluation of the involved arteries. Because those confounding factors affect the acquisition of the original time-of-flight MR angiography data set, no differences were seen between volume rendering and maximum intensity projection in this regard.
A main limitation of our investigation is its retrospective design. It was beyond the scope of this study to assess the accuracy of 3D time-of-flight MR angiography in the detection of cerebral aneurysms. Rather, we showed the performance differences between maximum intensity projection and volume rendering as postprocessing techniques of an already-acquired time-of-flight MR angiography data set. Another limitation of our study is that the standard reference, in most patients, was two-dimensional digital subtraction angiography, which in a few cases did not optimally show the aneurysm neck. Therefore, the good correlation of maximum intensity projection and volume rendering with digital subtraction angiography regarding neck categorization might be overestimated.
In summary, we conclude that volume rendering as a postprocessing technique for cerebral MR angiography has several advantages over maximum intensity projection. Volume rendering facilitates the evaluation of cerebral time-of-flight MR angiography data sets; augments diagnostic confidence in the detection of cerebral aneurysms, particularly those smaller than 3 mm; improves the perception of aneurysm morphology; allows better assessment of aneurysm size; and provides better image quality and vascular delineation. Volume rendering conveys faithfully the original data acquired by cerebral MR angiography and supports its noninvasive role in the evaluation of intracranial aneurysms.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Ries, S. Siemonsen, G. Thomalla, U. Grzyska, H. Zeumer, and J. Fiehler Long-Term Follow-Up of Cerebral Aneurysms after Endovascular Therapy Prediction and Outcome of Retreatment AJNR Am. J. Neuroradiol., October 1, 2007; 28(9): 1755 - 1761. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Papke, C. K. Kuhl, M. Fruth, C. Haupt, M. Schlunz-Hendann, D. Sauner, M. Fiebich, A. Bani, and F. Brassel Intracranial Aneurysms: Role of Multidetector CT Angiography in Diagnosis and Endovascular Therapy Planning Radiology, August 1, 2007; 244(2): 532 - 540. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Dachman Advice for Optimizing Colonic Distention and Minimizing Risk of Perforation during CT Colonography. Radiology, May 1, 2006; 239(2): 317 - 321. [Full Text] [PDF]
|
|
|
|
 |
|
 A Persson, N Dahlstrom, O Smedby, and T B Brismar Volume rendering of three-dimensional drip infusion CT cholangiography in patients with suspected obstructive biliary disease: a retrospective study Br. J. Radiol., December 1, 2005; 78(936): 1078 - 1085. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Bash, J. P. Villablanca, R. Jahan, G. Duckwiler, M. Tillis, C. Kidwell, J. Saver, and J. Sayre Intracranial Vascular Stenosis and Occlusive Disease: Evaluation with CT Angiography, MR Angiography, and Digital Subtraction Angiography AJNR Am. J. Neuroradiol., May 1, 2005; 26(5): 1012 - 1021. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. F. Tomandl, N. C. Kostner, M. Schempershofe, W. J. Huk, C. Strauss, L. Anker, and P. Hastreiter CT Angiography of Intracranial Aneurysms: A Focus on Postprocessing RadioGraphics, May 1, 2004; 24(3): 637 - 655. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Dachman and M. E. Zalis Quality and Consistency in CT Colonography and Research Reporting Radiology, February 1, 2004; 230(2): 319 - 323. [Full Text] [PDF] |
