| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Departments of Diagnostic and Interventional Neuroradiology (R.A., S.B., L.P., M.B., F.S., A.L.) and Neurology (X.D.), University Hospital, 29 avenue du Maréchal de Lattre de Tassigny, 54035 Nancy, France; and GE Medical Systems Europe, Buc, France (Y.T., L.L., E.K., R.V.). Received March 7, 2000; revision requested April 11; revision received July 17; accepted July 25. Address correspondence to R.A. (e-mail: r.anxionnat@chu-nancy.fr).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: In 22 ruptured aneurysms, neck visualization, aneurysm shape, and EVT feasibility were analyzed at 2D DSA (anteroposterior, lateral, and rotational views) and at maximum intensity projection (MIP) and surface shaded display (SSD) 3D DSA. The possibility of obtaining a working view for EVT at 3D DSA and the relevance of measurements in choosing the first coil also were assessed.
RESULTS: Two-dimensional DSA images clearly depicted the aneurysm neck in four of 22 aneurysms; MIP images, in 10; and SSD images, in 21, but SSD led to overestimation of the neck size in one aneurysm. Aneurysm shape was precisely demonstrated in five of 22 aneurysms at 2D DSA, in eight at MIP, and in all cases at SSD. In two of 22 aneurysms, EVT seemed to be nonfeasible at 2D DSA; however, SSD demonstrated feasibility and EVT was successfully performed. In one aneurysm, only SSD demonstrated the extension of the neck to a parent vessel, which was proved at surgery. Working views for EVT were deduced from 3D DSA findings in 20 of 21 aneurysms. The choice of the first coil was correct in 19 of 21 aneurysms.
CONCLUSION: Three-dimensional DSA is valuable for evaluating the potential for EVT, finding a working view, and performing accurate measurements.
Index terms: Aneurysm, intracranial, 17.73 • Aneurysm, rupture, 17.73 • Digital subtraction angiography, comparative studies, 17.12483, 17.127 • Digital subtraction angiography, technology, 17.124 • Digital subtraction angiography, three dimensional, 17.124 • Interventional procedures, 17.1267, 17.1269
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
DSA provides only two-dimensional (2D) projections of the cerebral vessels. The use of a biplanar system improves the visualization of the pathologic entity, but numerous oblique views often are necessary to find intracranial aneurysms and precisely analyze their angioarchitecture. These multiple oblique views may be automatically obtained by using rotational angiographic acquisition. This type of acquisition improves the analysis of aneurysms compared with conventional angiography (13–15,16). Three-dimensional DSA images can be reconstructed from rotational sequences. Three-dimensional DSA has already been evaluated in experimental conditions (17) and in short selected series of patients by using a prototype (18–21). The clinical results of those studies demonstrated a clear superiority of 3D DSA over 2D DSA, especially in the analysis of the aneurysm neck and the search for branches emerging from the pouch in complex cases.
The purpose of this study was to validate a fully automated system designed to provide both DSA and 3D reconstruction of cerebral vessels for endovascular treatment and to use 3D DSA as a supplement to 2D DSA in 20 patients with acute subarachnoid hemorrhage due to intracranial aneurysm rupture.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Fifteen milliliters of 300 mg/mL of iodinated contrast material is injected at a rate of 3 mL/sec in either the internal carotid or vertebral artery; this provides continuous filling of the intracranial arteries during rotational angiography. The injection is programmed to start 1.5 seconds before the acquisition of the first opacified image, but this default delay may be modified according to the hemodynamics of previous anteroposterior and lateral biplanar acquisitions.
This protocol results in a rotational series of 44 subtracted images. These images are sent to a multimodality workstation (AW 3.1; GE Medical Systems), on which the 3D reconstruction is automatically performed. Less than 7 minutes after the end of the rotational acquisition, a 512 x 512 x 512 3D image that shows all the vessels included in the field of view is displayed.
A multiresolution implementation of the algebraic reconstruction technique is used to reconstruct this huge high-spatial-resolution volume. To accelerate the computation, the algorithm uses the fact that the vessels occupy a very small proportion of the total volume. This hypothesis is relevant because subtracted projection images are used; thus, only opacified structures are reconstructed. An initial reconstruction is made at a low spatial resolution, and voxels with high reconstructed intensities are selected. A second and final reconstruction is made at the highest spatial resolution for this set of selected voxels (which correspond to vessels) and at a lower spatial resolution for all other voxels (which correspond to background tissue). The voxel size is isotropic. In the most widely used fields of view, 22 and 16 cm, the voxel sizes are 0.3 and 0.2 mm, respectively. This high spatial resolution allows precise visualization of small vessels such as perforating arterial branches and anterior choroidal arteries. Only the vessels seen during most of the rotational sequence—that is, over a wide angle—are reconstructed. Because the rotation time is 5 seconds, the venous structures may be seen on only the last images of the rotational sequences and therefore are not reconstructed, except in the case of an arteriovenous shunt.
The geometry of acquisition is estimated during weekly regular calibration of the system. This calibration is based on rotational acquisitions of two dedicated phantoms. The first phantom is a grid, which is used to provide an estimate of the image intensifier geometric distortions. The second phantom is a Lucite cylinder (GE Medical Systems), in which is included a set of bullets along a helicoid curve. The second phantom allows the estimation of the conic projection geometry (22,23).
3D DSA Display
Three-dimensional DSA angiographic volumes are reviewed by the neuroradiologist on a workstation inside the angiographic room during the course of the angiographic procedure.
Display methods.—Three-dimensional images are obtained by analyzing the reconstructed intensity along parallel rays emitted through the 3D DSA volume: Maximum intensity projection (MIP) views are obtained when only the maximum intensity is considered. Surface shaded display (SSD) requires the prior definition of a threshold by the user; this results in an isointense surface of the 3D DSA volume, which ideally corresponds to the lumen of the opacified vessels. The displayed value is a function of the angle between the ray and the normal vector to this surface at the first (closest to the observer) intersection point between the ray and the surface.
As a default, the volume is displayed by using MIP. No preprocessing is required to obtain such a view, but this view conveys little information about the relative depth of the vessels. The SSD method provides true 3D views. The surfaces of the vessels are first extracted by applying a user-defined threshold on the volume intensities. The effect of this threshold on the vessels can be interactively appreciated and may be corrected if necessary. This surface is then displayed as if it were illuminated by a light source. The position, signal intensity, and color of this source can be modified to improve the realism of the display.
It is possible to navigate inside the vessels and in the aneurysm pouch by using an endovascular display algorithm after applying a threshold that is usually the same as that used for SSD. The volume rendering technique attributes to each voxel an opacity value based on the volume’s intensity and on a user-defined signal intensity profile. At 3D DSA, the resultant image can appear transparent, opaque, or semitransparent.
User interface.—MIP and SSD views are usually simultaneously displayed to analyze a 3D DSA volume. An intuitive and efficient user interface allows the manipulation (eg, rotation, translation, zoom, electronic scalpel) of these views in real time on the screen. The neuroradiologist first reviews the following six predefined orientations: anterior, posterior, superior, inferior, left, and right views. Regions of interest can be focused on by reducing the field of view and cutting away parts of the 3D volume with an electronic scalpel. The analysis of the aneurysm anatomy is then refined by interactively moving the 3D DSA image in as many directions as necessary.
Three-dimension–assisted positioning.—During analysis, each 3D view is precisely defined in space by two angles: in cranial or caudal orientations and in right anterior oblique or left anterior oblique orientations. These angles can be automatically sent back to the acquisition system to perform fluoroscopy or acquire additional 2D DSA series by using the orientation defined at 3D imaging. In the few cases in which the chosen view is too extreme to be reproduced on the C arm, the message "unreachable" appears on the workstation screen.
Measurements.—Multiplanar reformations with a high spatial resolution due to small voxel size also can be obtained in real time. These reformations can be used to perform measurements of the aneurysm neck and pouch in at least two diameters. In the current study, the precision of these measurements was evaluated on phantoms and found to be within plus or minus 0.5 mm.
Evaluation of 3D DSA Images
Patients.—We performed 3D DSA in 20 patients (14 women, six men; mean age at diagnosis, 42 years; age range, 19–76 years) with acute subarachnoid hemorrhage who underwent an angiographic examination, between September and November 1998.
Angiographic protocol.—Angiograms were acquired in all 20 patients by using the biplanar C-arm system. Our angiographic protocol, which was systematically applied for each internal carotid artery and at least one vertebral artery, started with the acquisition of simultaneous anteroposterior and lateral biplanar images, which depicted the global hemodynamics of the vessels. Rotational angiography was then performed, and 3D DSA images were automatically obtained on the multimodality workstation inside the angiographic room less than 7 minutes later. The neuroradiologist reviewed the anteroposterior and lateral angiograms, rotational angiograms, and 3D DSA images to analyze the aneurysm anatomy and to decide whether to perform endovascular treatment. When a treatment decision was made, the neuroradiologist, by manipulating the 3D DSA images on the workstation screen, found a view that clearly depicted the aneurysm neck and could be used for the treatment. This view, as depicted at 3D DSA, was then reproduced on the C arm for the endovascular treatment.
In expectation of endovascular treatment, all of the patients with subarachnoid hemorrhage were put under general anesthesia during the angiographic procedure to avoid subtraction artifacts due to motion.
Aneurysms.—Twenty-two aneurysms were analyzed at both 2D DSA (anteroposterior, lateral, and rotational views) and 3D DSA (MIP and SSD) according to our angiographic protocol. These aneurysms were at the internal carotid artery (n = 9), anterior cerebral artery (n = 8), middle cerebral artery (n = 3), tip of the basilar artery (n = 1), and inferior cerebellar artery (n = 1). In 21 aneurysms (in 19 patients), selective endovascular occlusion with Guglielmi detachable coils (Target Therapeutics, Fremont, Calif) was performed during the same angiographic session. The time from diagnosis of subarachnoid hemorrhage to endovascular treatment varied from 1 to 4 days. One middle cerebral artery aneurysm was treated by means of surgery because of branches arising from the aneurysm pouch. Two patients each presented with two aneurysms, and because it was not possible to assess which aneurysm was bleeding, both were treated with detachable coil occlusion.
Evaluation method.—The following parameters regarding aneurysm anatomy were evaluated and pertinent for the endovascular treatment:
1. The visualization and exact location of the aneurysm neck, with special attention given to possible extension of the neck to the division branches of the parent vessel. These characteristics were given a classification of poor visualization or misinterpretation, ambiguous visualization, or precise visualization.
2. The shape of the aneurysm and the possible incorporation of vessels into the aneurysm, which were given a classification of poor visualization, ambiguous visualization, or precise visualization.
3. The possibility of performing endovascular treatment on the basis of the aneurysm anatomy. This possibility was classified as follows: insufficient information provided, endovascular treatment probably possible or probably not possible, or endovascular treatment definitely possible or definitely not possible. The treatment actually performed was recorded in each case.
Two senior neuroradiologists (S.B., R.A.) retrospectively analyzed these parameters independently at 2D DSA and 3D DSA and resolved differences in assessment by means of consensus. Three-dimensional DSA images were simultaneously displayed in both MIP and SSD views in all cases. The described three parameters allowed assessment of whether 3D DSA provided additional information to 2D DSA. However, this information needed to be confirmed. In the case of surgery, direct confirmation was obtained by means of visualization of the aneurysm. In the cases of endovascular treatment, indirect confirmation was partially provided by using the following two additional parameters, which were evaluated at 3D DSA:
1. The possibility of obtaining, at 3D DSA, a view that delineated the aneurysm neck from the adjacent vessels and could be reproduced on the C arm; this possibility was classified as either possible or impossible. Such a view, as used to perform the endovascular treatment, was thus referred to as the working view.
2. The accuracy of measurements of the aneurysm neck and pouch in choosing the diameter of the first coil to place in the aneurysm; this parameter was classified as the correct or incorrect choice of the first coil.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
SSD imaging was more effective in demonstrating the relationship between the aneurysm and the adjacent vessels than MIP imaging and 2D DSA. MIP was less demonstrative than SSD in 11 (50%) of 22 aneurysms because of superimposition between the neck and the vessels. In three supraclinoid internal carotid artery aneurysms, the SSD images clearly demonstrated the relationship between the neck and the origin of the anterior choroidal artery, whereas the MIP and 2D DSA images did not (Fig 1). In one large aneurysm at the tip of the basilar artery, neither 3D MIP nor the anteroposterior, lateral, or rotational 2D views showed clearly the relationship between the neck and both posterior cerebral arteries, whereas this relationship was precisely shown at SSD. In the same way, the large neck of an anterior cerebral artery aneurysm was well defined on only the posterior SSD view (Fig 2). In addition, only SSD imaging clearly depicted the extension of the aneurysm neck to a main branch of the division of a middle cerebral artery (Fig 3).
|
|
|
|
|
|
|
|
|
|
|
|
|
Aneurysm Shape
As detailed in Tables 3 and 4, the aneurysm shape was precisely assessed in four (18%) of 22 aneurysms and poorly assessed in five (23%) at biplanar 2D DSA; visualization was ambiguous in 13 (59%) aneurysms. The numerous 2D rotational views only slightly improved these results: Visualization of the shape was precise in five (23%) of 22 aneurysms, poor in three (14%), and ambiguous in 14 (64%). MIP allowed precise visualization of the shape in eight (36%) of 22 aneurysms, but visualization remained ambiguous in 13 (59%) and poor in one (4%). These limitations of both 2D DSA and 3D MIP in the visualization of the aneurysm shape were due to superimposition and lack of depth on these images, factors that necessitated a mental reconstruction of the aneurysm shape in complex cases. In contrast, SSD imaging clearly demonstrated the shape and was effective in depicting the complex bilobated or irregular shapes of all the aneurysms. Thus, SSD was the only imaging mode of those studied that enabled precise visualization of the aneurysm shape in all cases.
|
|
|
In one internal carotid artery aneurysm, MIP imaging was not able to delineate the aneurysm neck from the origin of the anterior choroidal artery although this information was easily demonstrated at SSD imaging (Fig 1); this aneurysm underwent endovascular treatment. In an anterior cerebral artery aneurysm at the A1 segment–A2 segment junction, the 2D and MIP views showed a large aneurysm neck extending to both anterior cerebral arteries, which is a contraindication to endovascular treatment. However, the posterior view at SSD imaging showed the neck to be distinct from both anterior cerebral arteries—an indication that endovascular treatment could be performed (Fig 2). The subsequent endovascular treatment was successful. In one intracavernous internal carotid artery aneurysm, SSD imaging greatly overestimated the neck size (Fig 4) and would have led to the use of a remodeling technique (25). MIP analysis enabled the correction of this measurement, which was confirmed at 2D DSA acquisition in the same orientation. Endovascular treatment was performed without the need for a remodeling technique.
In the last three aneurysms described, the analysis of SSD images alone or MIP images alone might have led to an inappropriate treatment decision, whereas simultaneous analysis of both sets of images allowed us to make the appropriate decision for all aneurysms, as shown in Table 6.
|
|
|
|
|
|
Size of Aneurysm Neck and Pouch
In the 21 aneurysms treated by using detachable coils, measurements of the aneurysm neck and the largest depth and width of the pouch were performed on magnified multiplanar reformations (Figs 4, 5). These measurements facilitated our choice of the first coil. The diameter of this coil was chosen according to the largest width of the aneurysm and to the neck size. As shown in Table 7, the first choice was appropriate in 19 (90%) of 21 aneurysms. For two (10%) aneurysms, we had to change the first coil. In one of these, the aneurysm length, largest width, and neck size were 3.8, 2.6, and 2.2 mm, respectively. We first tried to use a detachable coil with a diameter slightly larger than the neck size—that is, 3 mm in diameter and 6 cm in length. It was replaced with a 2-mm-diameter, 8-cm-long detachable coil, which entered the aneurysm pouch. In the other aneurysm, the length, largest width, and neck size were 2.6, 3.5, and 2.0 mm, respectively. We chose a 3-mm-diameter, 8-cm-long detachable coil. This coil was easily introduced into the aneurysm pouch but judged by the operator to be too small; this resulted in replacement with a 4-mm-diameter, 6-cm-long coil, which was a better size for the aneurysm pouch.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
For endovascular treatment, precise analysis of the aneurysm anatomy and a working view that delineates the neck from adjacent vessels are required. For this purpose, 2D DSA necessitates multiple oblique views and, when available, rotational acquisition (13–15,16). However, in complex cases it still may be hard to precisely depict the neck of the aneurysm and its relationship with adjacent vessels on these 2D projections (8,16–18).
Selective detachable coiling (26) is the preferred treatment for intracranial aneurysms at our institution. Since December 1996, 3D DSA has been part of our angiographic protocol in cases of subarachnoid hemorrhage (27,28). The introduction of 3D DSA led to a simple and reproducible angiographic protocol. This protocol starts with a simultaneous anteroposterior and lateral biplanar acquisition followed by a rotational angiographic series; this enables the reconstruction of a 3D DSA model. In this study, 3D DSA allowed precise analysis of the aneurysm anatomy in all 22 aneurysms, whereas 2D DSA, including anteroposterior, lateral, and rotational views, allowed precise analysis in only four of 22 aneurysms regarding neck visualization and in five of 22 aneurysms regarding aneurysm shape. Although rotational acquisition affords an opportunity to view the aneurysm in various projections, these projections are in only a single plane of rotation and do not include any projection in the cranial or caudal orientation; thus, in our series, the rotational views only slightly improved the anteroposterior and lateral analysis of the aneurysm anatomy.
Comparison of MIP and SSD imaging showed that SSD was the best mode of display to precisely visualize the neck and clearly depict the shape of the aneurysm, because it gives a true 3D view, whereas at MIP imaging, all structures are superimposed as in x-ray projections. The SSD images always showed the exact location of the neck and demonstrated the shape of the aneurysm. However, SSD may lead to a slight overestimation of the aneurysm and vessel sizes, whereas MIP images, which are produced from the original nonthreshold 3D model, allow a visual assessment of the sizes of the neck and pouch.
In this series, the neck size of an internal carotid artery aneurysm was greatly overestimated at SSD imaging compared with that at MIP and the corresponding 2D studies performed on the C arm with the same angles. On the basis of these results, we recommend simultaneous display of both SSD and MIP views. Our analysis of 3D DSA images with MIP and SSD proved to add confidence in the decision of treatment and even prompted us to perform an endovascular treatment in two aneurysms in which this treatment was not indicated at 2D analysis.
For endovascular treatment, one must find a 2D view that clearly depicts the aneurysm neck from adjacent vessels and could be used to perform the treatment in optimal conditions. Such a working view was found by using 3D DSA in all cases. Thus, instead of the multiple oblique acquisitions that are usually needed with standard 2D DSA, especially with complex aneurysms, a single optimized 2D acquisition based on 3D angiographic analysis was performed. This 2D view consistently corresponded well with the 3D view and was satisfactory to perform endovascular treatment. The angles that clearly delineated the neck from adjacent vessels were selected by interactively rotating the 3D DSA images on the workstation, and they often happened to be complex.
The choice of the diameter of the first detachable coil is important to ensure correct deployment of the coil into the aneurysm pouch. Precise measurements of the aneurysm pouch that were obtained by using multiplanar reformations at 3D DSA imaging helped us in this important and difficult choice, especially in determining the coil diameter.
There were some limitations in this study. First, we did not directly compare 3D DSA with 2D DSA. Such a comparison would necessitate a prospective study in which optimally performed 2D DSA—that is, consisting of as many views in varying planes as necessary—was compared with 3D DSA. Such a study would also include a comparison of the amount of contrast material administered, radiation dose, and examination time, with an increase in these three items, and would necessitate ethical considerations. Our purpose was to describe, as precisely as possible, our use of 3D DSA as a clinically routine method for the analysis and endovascular treatment of intracranial aneurysms.
Second, we did not obtain absolute confirmation of the aneurysm anatomy as demonstrated at 3D DSA, except that of the middle cerebral aneurysm that was operated on. The satisfactory choice of the working view for endovascular treatment, as well as the use of measurements on the 3D images to choose the diameter of the first coil, indirectly confirmed the reliability of the 3D information.
In conclusion, 3D DSA seems to be a valuable supplement to standard biplanar DSA for precise demonstration of the aneurysm anatomy and a reliable interactive planning tool to select a working view and the first coil and thereby assist endovascular treatment.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, R.A., S.B., L.P.; study concepts and design, R.A., S.B., L.P.; definition of intellectual content, R.A., S.B., L.P., Y.T.; literature research, R.A., Y.T.; clinical studies, R.A., S.B., L.P., X.D.; data acquisition, R.A., S.B., L.P., M.B., F.S., A.L.; data analysis, R.A., S.B., L.P., Y.T., L.L., E.K., R.V.; manuscript preparation and editing, R.A.; manuscript review, Y.T., L.L., E.K.; manuscript final version approval, R.A., S.B., L.P., Y.T., E.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  W.J. van Rooij, M.E. Sprengers, A.N. de Gast, J.P.P. Peluso, and M. Sluzewski 3D Rotational Angiography: The New Gold Standard in the Detection of Additional Intracranial Aneurysms AJNR Am. J. Neuroradiol., May 1, 2008; 29(5): 976 - 979. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. McKinney, C. L. Truwit, C. S. Palmer, M. Teksam, K. Papke, C. K. Kuhl, M. Fruth, C. Haupt, M. Schlunz-Hendann, D. Sauner, et al. Intracranial Aneurysms: Is the Diagnostic Accuracy Rate of Multidetector CT Angiography Equivalent to That of Three-dimensional Rotational Conventional Angiography? Radiology, March 1, 2008; 246(3): 982 - 984. [Full Text] [PDF]
|
|
|
|
|
|
M. Romijn, H.A.F. G. van Andel, M.A. van Walderveen, M.E. Sprengers, J.C. van Rijn, W.J. van Rooij, H.W. Venema, C.A. Grimbergen, G.J. den Heeten, and C.B. Majoie Diagnostic Accuracy of CT Angiography with Matched Mask Bone Elimination for Detection of Intracranial Aneurysms: Comparison with Digital Subtraction Angiography and 3D Rotational Angiography AJNR Am. J. Neuroradiol., January 1, 2008; 29(1): 134 - 139. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H.-J. Priebe Aneurysmal subarachnoid haemorrhage and the anaesthetist Br. J. Anaesth., July 1, 2007; 99(1): 102 - 118. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Kakeda, Y. Korogi, N. Ohnari, Y. Hatakeyama, J. Moriya, N. Oda, K. Nishino, and W. Miyamoto 3D Digital Subtraction Angiography of Intracranial Aneurysms: Comparison of Flat Panel Detector with Conventional Image Intensifier TV System Using a Vascular Phantom AJNR Am. J. Neuroradiol., May 1, 2007; 28(5): 839 - 843. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H Takao, I Doi, and T Watanabe Superselective transcatheter arterial chemoembolisation of an unresectable hepatocellular carcinoma using three-dimensional rotational angiography Br. J. Radiol., May 1, 2007; 80(953): e85 - e87. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R R Bridcut, E Murphy, A Workman, P Flynn, and R J Winder Patient dose from 3D rotational neurovascular studies Br. J. Radiol., May 1, 2007; 80(953): 362 - 366. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Brisman, J. K. Song, and D. W. Newell Cerebral aneurysms. N. Engl. J. Med., August 31, 2006; 355(9): 928 - 939. [Full Text] [PDF] |
