Imaging and anatomy of the normal intracranial venous system

Neuroimag Clin N Am 13 (2003) 1 – 12

Imaging and anatomy of the normal intracranial venous system James N. Scott, MD, MSc*, Richard I. Farb, MD Division of Neuroradiology, Department of Medical Imaging, The Toronto Western Hospital, University of Toronto, Toronto, Ontario, Canada

Current techniques for evaluating the intracranial venous system Assessment of the intracranial venous system has traditionally been made during the venous phase of conventional catheter digital subtraction angiography (DSA). In fact, DSA has remained the most definitive diagnostic tool, or reference standard, in the investigation of intracranial venous disease. Advantages include widespread availability, excellent spatial resolution, very high sensitivity and specificity, and familiarity of the images to surgeons. DSA is, however, an invasive procedure with well-known associated risks [1,2]. Additional disadvantages include a short postprocedure hospital stay, radiation exposure, and allergic or nephrotoxic effects of iodinated contrast medium.

Reported advantages of CT venography, compared with the MR techniques, include its more rapid image acquisition that reduces the negative effect of patient motion-related artefacts [3]. In addition, CT venography avoids the many patient contraindications and other artefacts that may either prevent or limit MR evaluation. CT venography, however, requires the use of iodinated contrast material and exposure to ionizing radiation, has complex postprocessing work, and provides limited visualization of skull base structures. As a result, CT venography is providing a supportive role at most centers, including our own institution, with increasing reliance on MR imaging and MR angiography as the current methods of choice to image the intracranial venous system.

Magnetic resonance cerebral venography Computed tomographic cerebral venography With the introduction of helical Computed tomography (CT), the intracranial venous system can be reliably assessed using CT venography [3]. In fact, high-resolution CT has been reported to be comparable with magnetic resonance (MR) angiographic techniques that utilize either time-of-flight or motioninduced phase shifts in diagnosing dural sinus thrombosis [4].

* Corresponding author. E-mail addresses: [email protected] (J.N. Scott), [email protected] (R.I. Farb).

Recent advances in MR angiography have made it possible to visualize the dural sinuses and cerebral veins without the use of invasive techniques or ionizing radiation [5 – 14]. Good correlation has been shown between MR venography and DSA [7], although some artefacts and potential diagnostic pitfalls exist with individual MR techniques [12,15 – 17]. Other advantages include the minimal risk of allergic or nephrotoxic effects when gadolinium (Gd) is administered intravenously. Disadvantages of MR angiography include poorer spatial resolution and slightly lower sensitivity and specificity for vascular patency relative to DSA. Contraindications to MR angiography include severe claustrophobia, the presence of implanted electronic devices (eg, pacemakers, cochlear implants, thalamic neural stimulators), non-

1052-5149/03/$ – see front matter D 2003, Elsevier Science (USA). All rights reserved. PII: S 1 0 5 2 - 5 1 4 9 ( 0 2 ) 0 0 0 6 2 - X

2

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

MR – compatible cerebral aneurysm clips, and ferromagnetic foreign bodies in critical locations (eg, eye, brain, lung). In addition, the MR angiographic technique may be suboptimal in patients with metallic clips or other devices that can obscure vessel detail. There are three general methods currently available for MR angiography of the intracranial venous system: 2D time-of-flight (TOF), 3D phase contrast (PC), and 3D Gd-enhanced. Two-dimensional time-of-flight MRA TOF MR angiography (MRA) is based on the principle of flow-related enhancement and highlights differences in magnetization between nuclei in flowing blood and those in stationary tissue [16,18 – 20]. TOF MR angiographic methods have included both 2D and 3D techniques, although the latter is considered much inferior when visualizing the intracranial venous system [11], primarily because of signal loss from in-plane saturation. The typical characteristics of TOF MRA include a conventional gradient echo (GRE) sequence, short repetition time (TR), intermediate or high flip angle, and short echo time (TE). The orientation of the acquisition plane is selected to be perpendicular to the main direction of flow and is typically coronal when imaging the intracranial venous system. Spatial presaturation pulses are commonly applied either above or below each slice to reduce signal from overlapping arteries or veins and thereby select for flow in one direction or the other. In these methods, stationary tissue reaches a steady ‘‘saturated’’ state of magnetization as it is exposed to repeated radiofrequency (RF) excitation pulses throughout the imaging process and appears dark or intermediate shades of gray. By contrast, fresh blood flowing into an image section perpendicular to the plane of the acquisition retains significantly higher magnetization strength, as it has not been subjected to these RF pulses (ie, it is ‘‘unsaturated’’) and therefore appears as high contrast ‘‘bright blood’’ signal. Maximum intensity projection (MIP) images can be created by combining the data from multiple 2D TOF sections (Fig. 1). The major advantage of 2D TOF MR angiography is that it is sensitive to slow flow and has relatively short acquisition times. Its main disadvantages include an insensitivity to in-plane flow, patient motion causing vessel misregistration among the slices, and high signal from substances with short T1 values (eg, fat, methemoglobin, Gd) that can mimic flow because of incomplete saturation and ‘‘shine through’’ on the MIP reconstruction. Magnetization transfer saturation can be used to decrease the signal intensity of back-

ground tissue without affecting blood, thereby further increasing contrast between them and possibly helping to overcome this limitation. Three-dimensional phase contrast MRA Phase-contrast MRA is based on the principle of using velocity-induced phase shifts to depict flowing blood [16,19 – 22]. In contrast with stationary tissue, spins moving through a magnetic field will experience a phase shift that is proportional to the velocity of flow, amplitude of the bipolar gradients, and time interval between the gradient lobes. The typical characteristics of PC angiography include a conventional GRE sequence, bipolar (flow encoding) gradients along one or more axis, and a velocity encoding variable (VENC) that adjusts the bipolar gradient strength to produce a phase shift of 180°. PC images are reconstructed from two or more data sets that are usually acquired simultaneously in an interleaved fashion. One data set acts as a reference GRE phase without flow encoding, whereas the others are acquired when the bipolar gradients are applied along the x-, y-, and z-axes. In this way, the 3D PC angiographic data set combines flow-sensitive transverse, coronal, and sagittal images to visualize the intracranial venous system. The major advantages of PC MRA are the greater suppression of background (stationary) tissues, and the ability to quantify flow and determine flow direction. Although 3D PC MRA is comparable to 2D TOF techniques in visualization of intracranial venous structures [11], its disadvantages include relatively long acquisition times and the need to predict the optimal VENC which is generally not known in advance. Gadolinium-enhanced three-dimensional MRA Gadolinium-enhanced 3D MRA is gaining wide clinical acceptance in a variety of vascular applications including the intracranial venous system [13,14,23]. Unlike the previous two MR methods, it is not a TOF technique and does not depend on the inflow of unsaturated spins. Rather, the paramagnetic effect of Gd shortens the intravascular T1 relaxation time, thus increasing the signal intensity of blood with no saturation effects [19,20]. In this way, the contrast between blood and stationary tissue becomes relatively flow-independent. In a manner analogous to conventional catheter angiography, Gd-enhanced 3D MRA thereby produces a ‘‘lumenogram’’ with contrast enhancement of the intravascular space. The typical characteristics of Gd-enhanced 3D MRA include a conventional GRE sequence, keeping

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

3

Fig. 1. Normal 2D TOF venogram. Thirty-five-year-old male patient with nonlocalizing headache symptoms. TOF MIP images in (A) sagittal, (B) postero-anterior, and (C) right posterior oblique projections demonstrate patency of the superior sagittal sinus, straight sinus, torcular Herophili, and left transverse sinus. But a common flow gap involving approximately two thirds the size of the right transverse sinus (open short arrows) creates some difficulty in confidently discriminating between a thrombosed versus hypoplastic sinus. (D) ATECO MR venographic MIP image in the same patient demonstrates the hypoplastic right transverse sinus to be patent (short arrows) with a duplication at its proximal aspect, and contain a small intraluminal arachnoid granulation (open long arrow) within its lateral aspect.

4

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

TR and TE as short as possible, and using an intermediate flip angle to keep background tissue sufficiently saturated and provide an image of the passing contrast bolus. For optimal image quality, the injection profile must be timed so that the contrast bolus is maximally present within the vessels of interest during acquisition of the low k-space frequencies. An acquisition that is too early or too late may miss the peak passage of contrast bolus and produce inadequate vascular visualization. To this end, MR-compatible injectors and automated bolus tracking software to trigger scan acquisition may be employed. The major advantages of Gd-enhanced 3D MRA are the much superior visualization of intracranial venous morphology (Fig. 2), greater suppression of unwanted background signal, and avoidance of saturation effects that are often problematic with the TOF technique [13,14]. This point should be emphasized. Although the artifactual signal loss encountered using TOF techniques [12,14] may occur at predictable locations within the intracranial venous anatomy (ie, posterior sagittal sinus, transverse sinus, and transverse sigmoid junctions), its presence only increases the difficulty in confidently discriminating a hypoplastic from a thrombosis dural sinus (Fig. 1). The 3D nature of the data set also permits different postprocessing techniques that can optimize visualization of venous structures [24]. Perceived disadvantages of Gd-enhancing 3D MRA might include issues of cost of the contrast agent, accompanying cost of the power injector and supplies, patient discomfort of obtaining antecubital intravenous (IV) access, and training of the MR technologists. ATECO 3D Gd-enhanced MR venography We have had recent success using a novel autotriggered elliptic centric ordered (ATECO) 3D Gd-enhanced MR venography [14]. Details of the fully automated detection and triggering system have been previously described for intracranial [25] and spinal [26] vascular imaging. In brief, the ATECO MR venographic examination consists of four integral stages: (1) initiation of an axial 2D bolus detection sequence located at the level of the cavernous carotid arteries; (2) power injection of an IV contrast material; (3) automated detection of the arrival of intra-arterial Gd-based contrast material at the cavernous carotid level, resulting in automatic termination of the detection sequence, and triggering of (4) a fast 3D GRE MR angiographic sequence with elliptical centric ordered phase encoding. Coordination of the IV injection and 3D centrically encoded scan is performed using a softwarebased ‘‘triggering tool’’ developed at our institution

[25 – 27]. Interactive slice positioning at the level of the skull base permits the MR technologist to position a region of interest (ROI) over the carotid and basilar arteries and determine a signal intensity-triggering threshold. Saturation bands active during the detection sequence extinguish unwanted flow-related arterial enhancement. The subsequent IV bolus injection of Gd-based contrast material results in a rapid rise in signal intensity within the ROIs that exceeds the set threshold. The 3D MR angiographic sequence is automatically triggered after an empirically determined eight-second delay to ensure sufficient contrast material perfusion has advanced well into the intracranial venous system prior to initiation of the centric filling of k-space. The view for this 3D MR angiographic sequence is in ascending order of radial k-space distance from the k-space origin, similar to that previously reported [25,28,29]. We use the following technique for the elliptical centric-ordered MR angiographic sequence on a 1.5-T system: TR 7.0 ms; TE 1.6 ms; flip angle 35°; matrix 320  320; field of view 25 cm; section thickness 1.3 mm; bandwidth of 62.5.0 kHz; and a slab thickness of 16 cm. All sequences are oriented in the sagittal plane. Thirty mL of Gd is injected at a rate of 3 mL per second, followed immediately by a saline flush at the same rate. Total imaging time is approximately 4 minutes 38 seconds. Maximum intensity projection images include a segmented MIP, following removal of the major arteries, which is rotated about its superoinferior axis in 15° increments through 180°. The resultant MIP image sets and source images are then transferred to a picture archiving and communications system (PACS) workstation for review and interpretation and, if required, can undergo further image segmentation specifically tailored to address any clinical questions particular to that case.

Anatomy of the intracranial venous system The intracranial venous system is a complex threedimensional structure that is often asymmetric and considerably more variable that the arterial anatomy. A more detailed description of its many possible variants requires a level of embryologic understanding that is beyond the purpose of this article and the reader is therefore referred to the excellent works of Lasjaunias et al [30]. In this article, we discuss the intracranial venous system in a craniofugal approach, moving from deep to cortical, and finally to dural. In this section, the supratentorial compartment is considered first, followed by the infratentorial compartment.

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

5

Fig. 2. Normal ATECO MR venogram. (A) Sagittal MIP image illustrates the typical segmentation performed (dashed line) to remove major arteries at the central skull base with (B) resultant antero-posterior and (C) left anterior oblique projections. Note the robust signal and visualization of the transverse sinuses (short arrows) and transverse-sigmoid junctions (open short arrows). Also noted are facial veins (large arrow) and the ptyergoid plexus (open large arrow). Nonexcluded portions of the internal carotid and vertebral arteries (arrowheads) are also visible on these views.

6

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

Deep venous system The deep venous system (Fig. 3) is concerned with centripetal venous drainage of deep cerebral white matter and basal ganglia and, using a classification suggested by Lasjaunias, can be considered at two separate levels: (1) the internal cerebral vein, the basal vein (of Rosenthal), and the great cerebral vein (of Galen), and (2) the transcerebral venous system [30]. These two levels operate in a hemodynamic balance such that some overlap and variation in their venous drainage is common. The paired internal cerebral veins are located near the midline within the tela choroidea in the roof of the third ventricle. The internal cerebral vein originates at the interventricular foramen of Monro where it is formed by the confluence of the septal, anterior caudate, ventricular, choroidal, and terminal (thalamostriate) subependymal veins, although anatomical variation in this region is common [31]. The internal cerebral veins run posteriorly to where they unite in the rostral part of the quadrigeminal cistern and form the great cerebral vein (of Galen). The septal vein drains the deep structures of the frontal lobes and courses around the anteromedial aspect of the lateral ventricle before passing posterior to the foramen of Monro to join the internal

cerebral vein. The caudate nucleus is drained by several caudate veins, which drain into the thalamostriate vein or directly into the internal cerebral vein. The thalamostriate vein drains the posterior frontal and anterior parietal lobes, caudate nucleus, and internal capsule, and is therefore usually a prominent tributary feeding the internal cerebral vein. The choroid plexus is drained by two choroidal veins (superior and inferior) that empty either directly into the internal cerebral vein, or first to the thalamostriate vein. The basal vein of Rosenthal originates deep within the sylvian fissure, near the medial part of the anterior temporal lobe, and receives veins draining the insula, cerebral peduncles, and multiple cortical (temporal) tributaries. The basal vein courses posteriorly at the base of the brain and curves around the cerebal peduncles to its junction with the great cerebral vein (of Galen) or internal cerebral vein. The basal vein has important anastamoses with its openings into the deep middle cerebral vein anteriorly, great cerebral vein (of Galen) posteriorly, and petrosal veins inferiorly. The great cerebral vein (of Galen) is a short, unpaired, midline structure that curves posteriorly beneath the splenium of the corpus callosum. It unites with the inferior sagittal sinus at the tentorial apex and forms the straight sinus.

Fig. 3. Postsegmentation ATECO MR venographic MIP images in (A) sagittal and (B) right posterior oblique projections demonstrate deep venous system structures, including variant anterior septal vein anatomy. The internal cerebral veins (open long arrows) collect the thalamostriate veins (open short arrows). The paired anterior septal veins (short arrows) join the main stem of the internal cerebral vein far beyond the foramen of Monro. After coursing superolaterally around the midbrain, the basal veins (of Rosenthal) (long arrows) join the internal cerebral veins to form the great cerebral vein (of Galen) (open large arrow). The confluence of the great cerebral vein (of Galen) and inferior sagittal sinus (arrowheads) at the tentorial apex (large arrow) is also shown.

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

The transcerebral veins are a group of superficial and deep medullary veins that drain the cerebral hemispheric white matter. The transcerebral veins are typically not visualized, however, during angiography because of their small caliber unless certain conditions are met that alter their normal hemodynamic state.

7

Superficial cerebral veins The superficial cerebral veins course over the surface of the brain, draining the cortex and a portion of the subjacent white matter. Despite a highly variable appearance, several large cortical veins can

Fig. 4. Three different patients in whom superficial cortical veins are noted on postsegmentation ATECO MR venographic MIP images. (A) The superficial middle cerebral vein (arrowheads) drains anteromedially into the cavernous or sphenoparietal sinus on this submento-vertex view. (B) The superior anastomotic vein of Trolard (short arrows) opens superiorly into the superior sagittal sinus along with multiple superior superficial cortical veins. (C) The inferior anastomotic vein of Labbe´ (open short arrows) opens into the transverse sinus.

8

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

Fig. 5. Visualization of the major dural sinuses with ATECO MR venographic MIP images in (A) sagittal, (B) antero-posterior, (C) right anterior oblique, and (D) left anterior oblique projections postsegmentation to remove major arteries located at the skull base. The superior sagittal sinus (large arrows) becomes progressively larger as it courses posteriorly, and is joined by the straight sinus (short arrows) and transverse sinuses (open short arrows) at the confluence of sinuses (torcular Herophili) (open large arrow). The hypoplastic inferior sagittal sinus (arrowheads) and left jugular bulb (long arrow) are also labeled. A round intraluminal filling defect within the distal right transverse sinus represents a normal arachnoid granulation (open long arrow). Nonexcluded portions of the internal carotid and vertebral arteries (open arrowheads) are also visible on these views.

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

often be identified individually and include the (1) superficial middle cerebral vein, (2) superior anastomotic vein, and (3) inferior anastomotic vein (Fig. 4). The latter two anastomotic veins are often in a reciprocal relationship such that if one is dominant, the other is usually hypoplastic or absent. The superficial middle cerebral vein runs anteriorly along the lateral (sylvian) fissure and receives smaller veins draining the lateral surface of the hemisphere. This large vein curves around the anterior temporal pole and either drains medially into the cavernous sinus or inferiorly into the pterygoid plexus. Anastomotic channels allow the superficial middle cerebral vein to drain in other directions. These include the superior anastomotic vein (of Trolard), which opens into the superior sagittal sinus, and the inferior anastomotic vein (of Labbe´), which opens into the transverse sinus. There are a variable number of other superiorly directed superficial cortical veins, often 10 – 12 in number, that also empty into the superior sagittal sinus along with the vein of Trolard.

9

Dural venous sinuses The cerebral veins empty into the venous sinuses, from which blood eventually flows into the internal jugular veins (Fig. 5). Dural venous sinuses are enclosed between the periosteal and meningeal layers of dura, and they lack valves. The superior sagittal sinus lies along the attached border of the falx cerebri and extends from the foramen cecum to the torcular Herophili. As it extends posteriorly, the superior sagittal sinus increases in caliber as it collects the superficial cerebral veins draining the cerebral convexities. Arachnoid granulations, contained within venous lacunae, are found protruding into the superior sagittal sinus along its course and may produce normal filling defects on imaging studies. The inferior sagittal sinus lies along the inferior free margin of the falx cerebri and drains the falx, anterior part of the corpus callosum, and medial aspects of the cerebral hemispheres. The inferior sagittal sinus extends posteriorly and is joined by the great cerebral vein (of Galen) to form the straight sinus.

Fig. 6. Postsegmentation ATECO MR venographic MIP images in (A) antero-posterior and (B) left anterior oblique projections demonstrate cavernous sinus relationships. Although partially obscured by basilar and cavernous carotid artery segments (open arrowheads) on these static images, the cavernous sinus (large arrow) can be seen to communicate across midline to its opposite cavernous sinus via the intercavernous sinus (open large arrow). The superior petrosal sinus (short arrows) drains posterolaterally from the cavernous sinus into the transverse sinus (long arrow), whereas the inferior petrosal sinus (open short arrows) opens more inferiorly into the jugular bulb (open long arrow). Delineation of these anatomic structures would improve during assessment of the routine 180° rotational MIP dataset.

10

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

The straight sinus lies within the attachment of the falx cerebri and tentorium cerebelli. It runs posteroinferiorly, terminates at the internal occipital protruberance, and is usually continuous with the left transverse sinus. The torcular Herophili (confluence of sinuses) is formed by the union of the superior sagittal sinus, straight sinus, and transverse sinuses, although the confluence is often asymmetric and quite variable in appearance [32,33]. The transverse sinuses lie along the attached margin of the tentorium cerebelli within a groove on the occipital bone. Each transverse sinus courses anterolaterally and, on reaching the base of the petrous portion of the temporal bone, turns inferomedially to form the sigmoid sinus that lies in the sigmoid sulcus of the temporal bone. The transverse sinus receives several important veins from the inferolateral temporal and occipital lobes, the cerebellum, and, when present, the anastamotic cortical vein of Labbe´. The transverse sinuses are commonly asymmetric, with the right transverse sinus being dominant in the majority of cases. Other common variations include a unilateral atretic segment [32,33]

and normal intraluminal filling defects (Fig. 5) resulting from arachnoid granulations [34 – 36], similar to those seen in the superior sagittal sinus. The sigmoid sinuses represent the anteroinferior continuation of the transverse sinuses and, in turn, drain into the jugular bulbs and terminate by becoming the internal jugular veins at the jugular foramen. The cavernous sinuses are situated on each side of the sphenoid body and represent an important confluence of intracranial and extracranial venous structures (Fig. 6). Each cavernous sinus is a multi-compartmental extradural space that extends from the superior orbital fissure to the petrous portion of the temporal bone. This sinus encloses the cavernous segment of the internal carotid artery and the abducens nerve, whereas the lateral wall of the sinus contains the oculomotor, trochlear, and ophthalmic division of the trigeminal nerves between its dural leaves. The superior petrosal sinus extends from the posterior aspect of the cavernous sinus to the transverse sinus, running along the attachment of the tentorium cerebelli to the petrous part of the temporal bone. The inferior petrosal sinus also extends from the posterior

Fig. 7. Two different patients in whom infratentorial venous structures are noted on (A) antero-posterior and (B) sagittal postsegmentation ATECO MR venographic MIP images. (A) The midline occipital sinus (short arrows) drains from the torcular Herophili into the left jugular fossa (arrowhead), although minimal flow is still identified within the hypoplastic left transverse and sigmoid sinuses (open arrowheads). Also noted is a relative paucity of pial venular structures over the left hemispheric convexity in this patient with ipsilateral encephalotrigeminal angiomatosis (Sturge-Weber syndrome). (B) The paired inferior vermian veins (open short arrows) course posterosuperiorly and usually open into the tentorial sinus. Also note an occipital transcalvarial emissary vein (long arrow) and suboccipital venous plexus (large arrow).

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12

aspect of the cavernous sinus but runs posterolaterally in a groove along the petro-occipital fissure to usually terminate by joining the jugular bulb. The sphenoparietal sinus lies along the lesser wing of sphenoid and drains usually the superficial middle cerebral (sylvian) vein into the cavernous sinus. Less common variations include the sphenoparietal sinus bypassing the cavernous sinus to drain into either the pterygoid plexus, or inferior petrosal or transverse sinus. The small variable occipital sinus lies in the midline at the attachment of the falx cerebelli, and it extends from the foramen magnum to drain upward into the torcular Herophili. Alternatively, with variant anatomy, the occipital sinus may drain toward the foramen magnum or into either the jugular fossa or suboccipital veins (Fig. 7). Infratentorial veins The veins of the posterior fossa collect into three major drainage systems: (1) superiorly into the vein of Galen and related tributaries, (2) anteriorly into the petrosal sinuses, and (3) posteriorly into dural sinuses bordering on or between the leaves of the tentorium cerebelli. In contrast with the dural sinuses and supratentorial veins, MR venography does not commonly visualize most of these deep and superficial venous structures such that only the most important ones are discussed here, and the reader is referred to previous works based on DSA technique [30,37] for a more complete discussion. The precentral cerebellar vein originates in the midline, between the lingula and central lobule of the cerebellar vermis, and runs superiorly, paralleling the roof of the fourth ventricle before opening into the great cerebral vein (of Galen). The superior vermian vein collects numerous small intrafissural venous tributaries of the upper vermis, and it empties into the great cerebral vein (of Galen), either directly or after uniting with the precentral cerebellar vein to first form the superior cerebellar vein. The anterior pontomesencephalic vein runs along the anterior surface of the pons and commonly drains superiorly via the peduncular vein into the posterior mesencephalic vein, which in turn runs around the upper midbrain to drain into the great cerebral vein (of Galen). The petrosal vein lies in the cerebellopontine angle cistern and receives drainage from the anterior cerebellar veins, in addition to other venous tributaries from the pons and medulla, before emptying into the middle portion of the superior petrosal sinus. The inferior vermian veins are important paired paramedian veins that course posterosuperiorly along the inferior vermis

11

and open into the tentorial sinuses (Fig. 7). Each receives numerous small venous tributaries draining the inferior cerebellar hemispheres.

Summary The intracranial venous system is a complex three-dimensional structure that is often asymmetric and considerably more variable than the arterial anatomy. The traditional approach has been to evaluate venous phase of catheter angiography. However, non-invasive imaging is now playing a greater role in evaluating the intracranial venous system in both healthy and diseased states. MR angiography, and especially Gd-enhanced 3D MRA, has recently emerged and offers excellent visualization of venous morphology from multiple orientations. An overview of the current non-invasive MRA methods and their applications has been provided during depiction of normal venous anatomy.

References [1] Heiserman JE, Dean BL, Hodak JA, et al. Nurologic complications of cerebral angiography. Am J Neuroradiol 1994;15:1401 – 7. [2] Spies JB, Berlin L. Complications of femoral artery puncture. Am J Roentgenol 1998;170:9 – 11. [3] Casey SO, Alberico RA, Patel M, et al. Cerebral CT venography. Radiology 1996;198(1):163 – 70. [4] Ozsvath RR, Casey SO, Lustrin ES, et al. Cerebral venography: comparison of CT and MR projection venography. Am J Roentgenol 1997;169:1699 – 707. [5] Chakeres DW, Schmalbrock P, Brogan M, et al. Normal venous anatomy of the brain: demonstration with gadopentate dimeglumine in enhanced 3-D MR angiography. Am J Neuroradiol 1990;11:1107 – 18. [6] Creasy JL, Price RR, Presbrey T, et al. Gadoliniumenhanced MR angiography. Radiology 1990;175(1): 280 – 3. [7] Mattle HP, Wentz KU, Edelman RR, et al. Cerebral venography with MR. Radiology 1991;178(2):453 – 8. [8] Tsuruda JS, Shimakawa A, Pelc NJ, et al. Dural sinus occlusion: evaluation with phase-sensitive gradientecho MR imaging. AJNR Am J Neuroradiol 1991; 12:481 – 8. [9] Yuh WTC, Simonson TM, Wang AM, et al. Venous sinus occlusive disease: MR findings. Am J Neuroradiol 1994;15:309 – 16. [10] Lewin JS, Masaryk TJ, Smith AS, et al. Time-of-flight intracranial MR venography: evaluation of the sequential oblique section technique. Am J Neuroradiol 1994; 15:1657 – 64. [11] Liauw L, van Buchem MA, Spilt A, et al. MR angiog-

12

[12]

[13]

[14]

[15]

[16]

[17]

[18] [19] [20]

[21] [22]

[23]

[24]

[25]

J.N. Scott, R.I. Farb / Neuroimag Clin N Am 13 (2003) 1–12 raphy of the intracranial venous system. Radiology 2000;214(3):678 – 82. Ayanzen RH, Bird CR, Keller PJ, et al. Cerebral MR venography: normal anatomy and potential diagnostic pitfalls. Am J Neuroradiol 2000;21:74 – 8. Liang L, Korogi Y, Sugahara T, et al. Evaluation of the intracranial dural sinuses with a 3D contrast-enhanced MP-RAGE sequence: prospective comparison with 2D-TOF MR venography and digital subtraction angiography. Am J Neuroradiol 2001;22:481 – 92. Farb RI, Scott JN, Willinsky RA, et al. Gadoliniumenhanced MR venography of the intracranial venous system using an auto-triggered elliptical centric ordered (ATECO) sequence: initial experience. Radiology 2003;206:203 – 9. Tsuruda J, Saloner D, Norman D. Artifacts associated with MR neuroangiography. Am J Neuroradiol 1992; 13:1411 – 22. Huston III J, Ehman RL. Comparison of time-of-flight and phase-contrast MR neuroangiographic techniques. Radiographics 1993;13(1):5 – 19. Lee VS, Martin DJ, Krinsky GA, et al. Gadoliniumenhanced MR angiography: artifacts and pitfalls. Am J Roentgenol 2000;175:197 – 205. Keller PJ. Time-of-flight magnetic resonance angiography. Neuroimaging Clin N Am 1992;2(4):639 – 56. Mitchell DG. Vascular techniques. In: MRI principles. Philadelphia: WB Saunders; 1999. p. 237 – 48. Pelc NJ, Alley MT, Listerud J, et al. Fundamentals of flow and hemodynamics. In: Atlas SW, editor. Magnetic resonance imaging of the brain and spine. ed 3. Philadelphia: Lippincott Williams & Wilkins; 2002. p. 118 – 22. Dumoulin CL. Phase-contrast magnetic resonance angiography. Neuroimag Clin N Am 1992;2(4):657 – 76. Turski P, Korosec F. Technical features and emerging clinical applications of phase-contrast magnetic resonance angiography. Neuroimag Clin N Am 1992;2(4): 785 – 800. Stevenson J, Knoop EA, Litt AW. MP-RAGE subtraction venography: a new technique. J Magn Reson Imaging 1995;5:239 – 41. Davis CP, Hany TF, Wildermuth S, et al. Postprocessing techniques for gadolinium-enhanced three-dimensional MR angiography. Radiographics 1997;17(5):1061 – 77. Farb RI, McGregor C, Kim JK, et al. Intracranial arteriovenous malformations: real-time auto-triggered elliptica centric-ordered 3D gadolinium-enhanced MR angiography—initial assessment. Radiology 2001; 220(1):244 – 51.

[26] Farb RI, Kim JK, Willinsky RA, et al. Spinal dural arteriovenous fistula localization using a technique of real-time auto-triggered elliptical centric ordered 3D Gd-MRA: an initial assessment. Radiology 2002; 222(3):843 – 50. [27] Derbyshire JA, Kim JK, Farb RI, et al. Automated contrast bolus detection and triggered true centric 3D MRA of the carotid arteries [abstract 766]. In: Proceedings of the 6th Meeting of the International Society for Magnetic Resonance in Medicine. Berkeley; 1998. [28] Maki JH, Prince MR, Londy FJ, et al. The effects of time varying intravascular signal intensity and k-space acquisition order on three-dimensional MR angiography image quality. J Magn Reson Imaging 1996;6: 642 – 51. [29] Wilman AH, Riederer SJ, King BF, et al. Fluoroscopically triggered contrast-enhanced three-dimensional MR angiography with elliptical centric view order: application to the renal arteries. Radiology 1997;205: 137 – 46. [30] Lasjaunias P, Berenstein A, ter Brugge KG, et al. Intracranial venous system. In: Lasjaunias P, Berenstein A, ter Brugge KG, editors. Surgical neuroangiography. ed 2. Berlin: Springer-Verlag; 2001. p. 631 – 95. [31] Tu¨re U, Yasßargil MG, Al-Mefty O. The transcallosaltransforaminal approach to the third ventricle with regard to the venous variations in this region. J Neurosurg 1997;87:706 – 15. [32] Bisaria KK. Anatomic variations of venous sinuses in the region of the torcular Herophili. J Neurosurg 1985; 62:90 – 5. [33] Cure´ JK, Van Tassel P, Smith MT. Normal and variant anatomy of the dural venous sinuses. Semin Ultrasound CT MRI 1994;15(6):499 – 519. [34] Mamourian AC, Towfighi J. MR of giant arachnoid granulations: a normal variant presenting as a mass within the dural venous sinus. Am J Neuroradiol 1995;16:901 – 4. [35] Roche J, Warner D. Arachnoid granulations in the transverse and sigmoid sinuses: CT, MR, and MR angiographic appearance of a normal anatomic variation. Am J Neuroradiol 1996;17:677 – 83. [36] Leach JL, Jones BV, Tomsick TA, et al. Normal appearance of arachnoid granulations on contrastenhanced CT and MR of the brain: differentiation from dural sinus disease. Am J Neuroradiol 1996;17: 1523 – 32. [37] Osborn AG. The cerebral veins. In: Diagnostic cerebral angiography. Philadelphia: Williams & Wilkins; 1999. p. 217 – 37.