Carotid Artery Stenosis

In evaluating occlusive disease of the extracranial carotid artery, CTA complements conventional or digital subtraction angiography (DSA) and serves as an alternative to MRA (Fig. 33B.1). In the grading of carotid stenosis using the North American Symptomatic Carotid Endarterectomy Trial (NASCET) criteria, Randoux and colleagues (2001) found that the rate of agreement between 3D CTA and DSA was 95%. Relative to DSA (the reference standard), severe stenosis (70%-99%) was detected with a sensitivity and specificity of 100% and 100%, respectively, for CTA and 93% and 100%, respectively, for contrast-enhanced MRA (CE-MRA). In addition, CTA and CE-MRA were significantly correlated with DSA in depicting the length of the stenotic segment.

Fig. 33B.1 Computed tomographic angiography (CTA) compared to digital subtraction angiography (DSA) in a patient with proximal internal carotid artery (ICA) stenosis. A, Three-dimensional reconstructed CTA image of left ICA reveals severe stenosis distal to the ICA bifurcation. B, DSA confirms severe stenosis seen on CTA due to an atherosclerotic plaque.

Other investigators have reported lower sensitivity (80%-89%) yet comparable specificity (96%-100%) for CTA in detecting severe stenosis (Binaghi et al., 2001; Magarelli et al., 1998). Those investigators found that time-of-flight (TOF) MRA had higher sensitivity (92%-93%) than CTA and similar specificity (98%-100%). Binaghi and colleagues also compared CTA with DSA and showed that the sensitivity was the same (89%), whereas specificity was higher for CTA (100%) than for DSA (81%). In 2006, Wardlaw et al., performed a meta-analysis on studies comparing CTA with DSA for the diagnosis of carotid artery stenosis. For detection of severe (70% to 99%) stenoses, CTA demonstrated a pooled sensitivity and specificity of 77% and 95%, respectively. Data for moderate (50%-69%) stenoses were determined to be sparse and unreliable (Wardlaw et al., 2006). An earlier systematic review found CTA to be a reliable method for detecting severe (70%-99%) stenoses, with a sensitivity and specificity of 85% and 93%, respectively. For detection of a complete occlusion, the sensitivity and specificity were 97% and 99%, respectively (Koelemay et al., 2004). Saba et al. evaluated the use of multidetector CTA and carotid ultrasound in comparison to surgical observation for evaluating ulceration, which is a severe complication of carotid plaques. CTA was found to be superior, with 93.75% sensitivity and 98.59% specificity compared to carotid ultrasound, which demonstrated 37.5% sensitivity and 91.5% specificity (Saba et al., 2007).

Fibromuscular dysplasia (FMD), which often involves a unique pattern of stenoses in the cervical vessels, may be detected by CTA, although no large studies have evaluated the sensitivity and specificity for detection. This disorder, which characteristically demonstrates a string-of-beads pattern of vascular irregularity on angiography, has been reliably demonstrated on carotid artery evaluations from case reports. This may potentially reduce the need for more invasive angiographic imaging in the future, although further studies in this area are required (de Monye et al., 2007).

Currently, either CTA or MRA is used to evaluate suspected carotid occlusive disease, with the choice of method determined by clinical conditions (e.g., pacemaker), accessibility of CT and MR scanners, and additional imaging capabilities (CT or MR perfusion brain imaging).

Carotid and Vertebral Dissection

Dissections of the cervicocephalic arteries, including the carotid and vertebral arteries, account for up to 20% of ischemic strokes in young adults (Leys et al., 1995). CTA findings include demonstration of a narrowed eccentric arterial lumen in the presence of a thickened vessel wall, with occasional detection of a dissecting aneurysm. In subacute and chronic dissection, CTA has been shown to detect a reduction in the thickness of the arterial wall, recanalization of the arterial lumen, and reduction in size or resolution of dissecting aneurysm. CTA or MRA is superior to DSA in depicting the distal portions of the cervical internal carotid artery (ICA), a common site of dissection. CTA is likely superior to MRI alone in evaluating aneurysms at these sites because MRI findings are often complicated by the presence of flow-related artifacts. CTA depiction of dissections at the level of the skull base may be complicated in some cases because of to beam hardening and other artifacts that obscure dissection findings, including similarities in the densities of the temporal and sphenoid bones with the dissected ICA.

A retrospective review compared combined multidetector CT/CTA with MRI/MRA among 18 patients with 25 dissected vessels in both anterior and posterior circulations. CT/CTA was preferred for diagnosis in 13 vessels, whereas MRI/MRA was preferred in 1 vessel, and the techniques were deemed equal in the remaining 11 vessels. It should be noted that such combinations of noninvasive angiographic study with other CT or MRI components is common and therefore does not reflect the role of CTA or MRA alone. A significant preference for CT/CTA was noted for vertebral artery dissections but not for ICA dissections (Vertinsky et al., 2008). When compared with DSA, a small retrospective review of patients found that multidetector CTA had a sensitivity of 100% and specificity of 95% for detecting vertebral artery dissections (Pugliese et al., 2007). A small study evaluated multislice CTA with cervical axial T1-weighted MRI and MRA among seven patients with carotid artery dissection. The combination of MRI and MRA identified dissection in five of the seven patients that were identified by CTA. Additionally, a dissecting aneurysm was identified by CTA that was missed by MRI and MRA. These findings suggest that CTA may be at the very least a complementary study to provide additional information (Elijovich et al., 2006).

Acute Ischemic Stroke

Computed tomography angiography is a reliable alternative to MRA in evaluating arterial occlusive disease near the circle of Willis in patients with symptoms of acute stroke (Knauth et al., 1997; Shrier et al., 1997) (Fig. 33B.2). CTA shows clinically relevant occlusions of major cerebral arteries and enhancement caused by collateral flow distal to the site of occlusion. CTA may be superior to transcranial Doppler (TCD) ultrasound in diagnosing atherothrombotic MCA disease in Asian patients presenting with middle cerebral artery (MCA) stroke (Suwanwela et al., 2002). CTA detected MCA stenosis measuring more than 50% in twice as many patients as TCD. The difference resulted primarily from improved detection by CTA of distal M1 and M2 stenosis. Because half of the patients studied by Suwanwela and colleagues had distal M1 and M2 disease, the authors concluded that TCD should not be used to screen for MCA stenosis.

Fig. 33B.2 Right middle cerebral artery (MCA) stenosis in a patient who subsequently received intracranial stent placement. A, Coronal image from computed tomographic angiography (CTA) shows focal distal M1 segment stenosis prior to stenting. B, 1.5T 3D time-of-flight (TOF) magnetic resonance angiography (MRA) demonstrates a focal flow gap of the right M1. MRA overestimates degree of stenosis when compared to CTA. C, Digital subtraction angiography (DSA) image after stent placement reveals right MCA restenosis.

In the detection of intracranial steno-occlusive disease, Hirai and colleagues (2002) have shown that combined CTA and MRA provide substantially higher sensitivity, specificity, and accuracy than MRA alone. Review of the CTA depiction of vessels in conjunction with the 3D time-of-flight (TOF) MRA reduced the frequency of overestimation of stenosis when compared with MRA alone. In the identification of 50% or greater stenosis, the sensitivity, specificity, and accuracy for the combined CTA and MRA evaluation were 100%, 99%, and 99%, respectively, and the values for 3D TOF-MRA alone were 92%, 91%, and 91%, respectively. The grading of stenosis by the combined approach agreed with DSA grading in 98% of cases. In a retrospective review of their cases, the authors found that CTA did not always correctly delineate arterial lumina with circumferential calcification and the cavernous portion of the ICA. Nguyen et al. evaluated 475 vessel segments in 41 patients who received both CTA and DSA studies and found that for arterial occlusions, CTA had demonstrated 100% sensitivity and specificity. For detection of 50% or more stenosis, CTA had 97.1% sensitivity and 99.5% specificity. There was no difference observed in CTA accuracy for vessel segments in the anterior versus posterior circulation (Nguyen-Huynh et al., 2008). In a recent review of published studies, Latchaw et al. found that in comparison to conventional angiography, CTA demonstrated sensitivities ranging from 92% to 100% and specificities of 82% to 100% for the detection of intracranial vessel occlusion. The sensitivities for detection of intracranial stenoses range from 78% to 100%, with specificities of 82% to 100% (Latchaw et al., 2009). CTA is considered to be superior to TCD in detecting intracranial stenoses and occlusions, with a high false-negative rate noted for Doppler ultrasound. Studies also suggest that CTA has a higher sensitivity when directly compared with 3D TOF-MRA. Bash et al. found that CTA had a sensitivity of 98% while MRA had a sensitivity of 70% for detection of intracranial stenosis. The sensitivity for detecting occlusions on CTA was 100% and only 87% on MRA. Additionally, CTA was noted to superior to both MRA and DSA in detecting posterior circulation stenoses when slow or balanced flow states were present, possibly owing to longer scan time, which allows for more contrast to pass through a critical stenosis. Although previous studies noted decreased accuracy with the presence of atheromatous calcifications, the sensitivity and specificity of CTA for stenosis quantification were not compromised by this when appropriate window and level adjustments were made to account for the blooming artifacts that are frequently associated with heavy calcifications (Bash et al., 2005).

Computed tomography angiography source images (CTA-SI) may be used to provide an estimate of perfusion by taking advantage of the contrast enhancement in the brain vasculature that occurs during a CTA, possibly making it unnecessary to perform a separate CT perfusion study with a second contrast bolus. In normal perfused tissue, contrast dye fills the brain microvasculature and appears as increased signal intensity on the CTA-SI. In ischemic brain regions with poor collateral flow, contrast does not readily fill the brain microvasculature. Thus, these regions demonstrate low attenuation (Schramm et al., 2002). The hypoattenuation seen on CTA-SI correlate with abnormality on diffusion-weighted MRI (DWI) and have been found to be more sensitive than noncontrast CT scans for the detection of early brain infarction (Camargo et al., 2007). The sensitivity of CTA-SI and DWI when directly compared has been found to be similar in detecting ischemic regions, but DWI is better at demonstrating smaller infarcts and those in the brainstem and posterior fossa. Such findings may be useful for patients with symptoms of acute infarction who cannot undergo MRI (Latchaw et al., 2009).

In addition to perfusion status, CTA imaging may potentially be used for prognostication in patients undergoing acute stroke intervention. The 10-point Clot Burden Score (CBS) was devised as a semiquantitative analysis of CTA to help determine prognosis in acute stroke (Fig. 33B.3). The CBS subtracts 1 or 2 points each for absent contrast opacification on CTA in the infraclinoid internal carotid artery (ICA) (1), supraclinoid ICA (2), proximal M1 segment (2), distal M1 segment (2), M2 branches (1 each), and A1 segment (1). The CBS applies only to the symptomatic hemisphere. A CBS below 10 was associated with reduced odds of independent functional outcome (odds ratio (OR) 0.09 for a CBS of 5 or less; OR 0.22 for CBS 6 to 7; OR 0.48 for CBS 8 to 9; all versus CBS 10). The quantification of intracranial thrombus extent with the CBS predicts functional outcome, final infarct size, and parenchymal hematoma risk acutely. This scoring system requires external validation and could be useful for patient stratification in stroke trials (Puetz et al., 2008).

Fig. 33B.3 The Clot Burden Score (CBS) on computed tomographic angiography (CTA). This is a 10-point imaging-based score where two points are subtracted for thrombus found on CTA in the supraclinoid internal carotid artery (ICA) and each of the proximal and distal segments of the middle cerebral artery (MCA) trunk. One point is subtracted for thrombus in the infraclinoid ICA and A1 segment and for each M2 branch.

The Alberta Stroke Program Early CT Score (ASPECTS) is a 10-point analysis of topographic CT scan score used in patients with MCA stroke (Fig. 33B.4 and Box 33B.1). Segmental assessment of MCA territory is made, and 1 point is removed from the initial score of 10 if there is evidence of infarction in the following regions: putamen, internal capsule, insular cortex, anterior MCA cortex, MCA cortex lateral to insular ribbon, posterior MCA cortex, anterior MCA territory immediately superior to M1, lateral MCA territory immediately superior to M2 and posterior MCA territory immediately superior to M3. An ASPECTS score of 7 or less predicts worse functional outcome at 3 months as well as symptomatic hemorrhage. Puetz et al. sought to determine whether the ASPECTS scoring system could be applied to CTA-SI and combined with the CBS system for improved prognostication. A 10-point ASPECTS score based on CTA-SI and the 10-point CBS were combined to form a 20-point score for patients presenting acutely with stroke who received thrombolysis treatment. For patients with a combined score of 10 or less, only 4% were functionally independent, and mortality was 50%. In contrast, 57% of patients with scores of 10 or greater were functionally independent, and mortality was 10%. Additionally, parenchymal hematoma rates were 30% versus 8%, respectively (Puetz et al., 2010). A similar semiquantitative scoring system for CTA-SI was devised for patients presenting with acute basilar artery occlusion and termed the posterior circulation (pc)-ASPECTS (Fig. 33B.5). This 10-point scoring system subtracts 1 or 2 points each for areas of hypoattenuation in the left or right thalamus, cerebellum, or posterior cerebral artery (PCA) territory, respectively (1 point), or any part of the midbrain or pons (2 points). Median follow-up pc-ASPECTS was lower in patients with a CTA-SI pc-ASPECTS less than 8 compared with patients with a CTA-SI pc-ASPECTS of 8 or higher, respectively. Hemorrhagic transformation rates were 27.3% versus 9.5%, respectively, for patients who received thrombolysis. The results indicate that such analysis can predict a larger final infarct extent in patients with basilar artery occlusion. Larger prospective trials are required for validation, but the systematic acute evaluation of CTA along with CTA-SI may potentially be used to help guide future stroke treatments (Puetz et al., 2009).

Fig. 33B.4 Axial noncontrast head computed tomography (CT) demonstrating middle cerebral artery (MCA) territory regions defined by the Alberta Stroke Program Early CT Score (ASPECTS). C, Caudate, I, insular ribbon, IC, internal capsule, L, lentiform nucleus.

Fig. 33B.5 Cerebral map defining the posterior circulation Acute Stroke Prognosis Early CT score (pc-ASPECTS) territories. From 10 points, 1 or 2 points each (as indicated) are subtracted for early ischemic changes or hypoattenuation on computed tomographic angiography source images (CTA-SI) in left or right thalamus, cerebellum, or posterior cerebral artery (PCA) territory, respectively (1 point); and any part of midbrain or pons (2 points).

Intracerebral Hemorrhage

Patients presenting acutely with intracerebral hemorrhage (ICH) within the first few hours of symptom onset are known to be at increased risk for hematoma expansion. However, only a fraction of such patients arrive at a hospital within this time frame, so alternative means of identifying potential hemorrhage expansion have been sought because it is an important predictor of 30-day mortality. One such prognostic marker has been identified on CTA: the spot sign, defined as a tiny, enhancing foci seen within hematomas, with or without clear contrast extravasation. A prospective study by Wada et al. of 39 consecutive patients with spontaneous ICH within 3 hours of symptom onset identified this sign in 33% of cases. Sensitivity was found to be 91%, and specificity was 89% for predicting hematoma expansion. In patients with the spot sign, mean volume change was greater, extravasation was more common, and median hospital stay was longer (Wada et al., 2007).

A larger retrospective analysis determined that the presence of three or more spot signs, a maximum axial dimension of 5 mm or greater, and maximum attenuation of 180 Hounsfield units or more were independent predictors of significant hematoma expansion (Delgado et al., 2009). In another study, Delgado et al. noted that the presence of any spot sign increased the risk of in-hospital mortality (OR, 4.0) and poor outcome among survivors at 3-month follow-up (OR, 2.5). This was determined to be an independent predictor of both measures. The spot sign currently requires further validation but remains a promising use of CTA in guiding acute ICH management (Delgado et al., 2010).

Brain Death

The absence of cerebral circulation is an important confirmatory test for brain death, and CTA is emerging as an important alternative means of testing. A novel 4-point score was devised, with points subtracted based on the lack of opacification of the cortical segments of the MCAs and internal cerebral veins. Frampas et al. used this system to prospectively evaluate 105 patients who were clinically brain dead and found a sensitivity of 85.7% and specificity of 100%. This appears to be a possible alternative means of detecting cerebral circulatory arrest, and given that it is a fast and noninvasive technique, it may become a useful confirmatory test (Frampas et al., 2009; Escudero et al., 2009).

Cerebral Venous Thrombosis

The diagnosis of cerebral venous thrombosis (CVT) was previously often made with conventional angiography and more recently by MRI techniques. Magnetic resonance venography (MRV) is commonly considered the most sensitive noninvasive test in diagnosing CVT. However, given the prolonged imaging time and often limited availability, CTA has been studied as a potential alternate means of detecting CVT. Spiral CT with acquisition during peak venous enhancement has been implemented with single-section systems but remains limited in spatial and temporal resolution. One study directly comparing CTV with MRV demonstrated a sensitivity and a specificity of 75% to 100%, depending on the sinus or venous structure involved (Khandelwal et al., 2006). Multidetector-row CTA (MDCTA) offers higher spatial and temporal resolution, which allows for high-quality multiplanar and 3D reformatting. Two recent small studies found 100% specificity and sensitivity with MDCTA when compared to MRV. The venous sinuses could be identified in 99.2% and the cerebral veins in 87.6% of cases. MDCTA may be equivalent to MRV in visualizing cerebral sinuses, but further studies are needed to evaluate the diagnostic potential of MDCTA in specific types of CVT such as cortical venous thrombosis, thrombosis of the cavernous sinus, and thrombosis of the deep cerebral veins. The advantages of MDCTA include the short exam duration and the possible simultaneous visualization of the cerebral arterial and venous systems with a single bolus of contrast. MDCTA visualizes thrombus via contrast-filling defects and remains less prone to flow artifacts. A potential problem with this technique lies in the fact that in the chronic state of a CVT, older organized thrombus may show enhancement after contrast administration and may not produce a filling defect, leading to a false-negative result. The addition of a noncontrast CT with the MDCTA is sometimes used to remove another potential to obtain false-negative results from the presence of a spontaneously hyperattenuated clot that could be mistaken for an enhanced sinus. This phenomenon is known as the cord sign and may be seen in 25% to 56% of acute CVT cases (Gaikwad et al., 2008; Linn et al., 2007).

Cerebral Aneurysms

Digital subtraction angiography has been the standard imaging method for diagnosis and preoperative evaluation for patients with ruptured and unruptured cerebral aneurysms. However, DSA is invasive and subject to complications resulting from catheter manipulation. Thus, in asymptomatic patients at greater risk for cerebral aneurysms, the use of noninvasive techniques such as MRA and CTA to screen for aneurysms is particularly attractive. These techniques have advantages and disadvantages. The most thoroughly investigated MRA technique is 3D TOF-MRA, and its main disadvantages are long scanning times, difficulty in detecting very small aneurysms, difficulty in establishing the relationship of the aneurysm to adjacent (and surgically important) osseous anatomy, and occasional difficulty in distinguishing between patent lumen, high-grade stenosis, and occlusion.

The main disadvantages of CTA are radiation exposure, the use of iodinated contrast material, and difficulty in detecting very small aneurysms and imaging artifacts from endovascular coils in treated aneurysms. CTA is nondiagnostic for determining the presence of a residual lumen and the size/location of the remnant neck of a treated aneurysm because of the streak artifacts caused by coils.

Although 3D TOF-MRA and dynamic 3D CE-MRA have similar sensitivity and specificity to CTA for detection of intracerebral aneurysms at least 5 mm in diameter, they have lower sensitivity for aneurysms smaller than 5 mm (Villablanca et al., 2002; White et al., 2001) (Figs. 33B.6 and 33B.7). The results of the International Study of Unruptured Intracranial Aneurysms (ISUIA) (Wiebers et al., 2003) suggest that MRA, despite its lower sensitivity for smaller aneurysms, would not significantly change management, because incidental aneurysms smaller than 10 mm should not be treated (exceptions may be made for individuals with daughter aneurysm formation, a family history of subarachnoid hemorrhage, and young patients). However, in general, the accuracy of CTA is felt to be at least equal if not superior to that of MRA in most circumstances, and in some cases, its overall accuracy approaches that of DSA (Latchaw et al., 2009).

Fig. 33B.6 Right middle cerebral artery aneurysm seen on both computed tomographic angiography (CTA) and magnetic resonance angiography (MRA). A, Coronal section on CTA reveals aneurysm in right middle cerebral artery (MCA) bifurcation. B, MRA also displays aneurysm with less definition. C, Three-dimensional reconstruction of CTA better defines saccular appearance of this aneurysm.

Fig. 33B.7 Left internal carotid artery (ICA) aneurysm. Comparison of computed tomographic angiography (CTA) postprocessed images with catheter angiography. A, Catheter angiography lateral view, following left ICA injection, shows aneurysm originating from supraclinoid portion of ICA. B, CTA axial source image reveals lobulated aneurysm (arrow). C-E, CTA three-dimensional (3D) volume-rendered images with transparency feature for user-selected tissue regions (called 4D angiography). C, Lateral view from left side of patient demonstrates relationship of the aneurysm, measuring 14 mm from neck to dome, to the anterior clinoid process. D, View of aneurysm (arrow), skull base, and circle of Willis from above. E, Same view as D but edited to remove most of skull base densities and improve visibility of vessels.

Using optimized helical CTA acquisition and post-processing protocols that included 3D volume-rendered images, 3D thick-slab and grayscale 2D single-section images, and thick-slab multiplanar reformatted 2D images, Villablanca and colleagues (2002) reported a sensitivity for the detection of small aneurysms (<5 mm diameter) of 98% to 100%, compared with 95% for DSA. The specificity of both CTA and DSA was 100%. CTA image analysis times ranged from 6 to 36 minutes (mean, 16 minutes). The smallest aneurysm detected was 1.9 × 1.6 × 1.3 mm³, and 48% of aneurysms were detected in the presence of subarachnoid hemorrhage. The sensitivity of CTA exceeded that of DSA, primarily because the optimal projection necessary to visualize some aneurysms could be displayed on the post-processed CTA images but was not or could not be displayed by the 2D DSA. Other disadvantages of DSA that have been noted by investigators include superimposition of normal vessels, obscuring a small aneurysm, and the lack of an internal image scale for estimating the aneurysm sac and neck dimensions. Villablanca et al. (2002) showed that CTA can provide quantitative information such as dome-to-neck ratios and aneurysm characterization such as the presence of mural thrombi or calcium, branching pattern at the neck, and the incorporation of arterial segments in the aneurysm. The 3D images in particular provided a surgically useful display of the aneurysm sac in relation to skull base structures (see Fig. 33B.7). The authors concluded that clinically relevant aneurysms can be detected by CTA using published protocols, routine scanners, and commercially available image-processing workstations. Furthermore, CTA can be a reliable source of information for treatment planning.

Cerebral veins show much more anatomical variation than arteries. The presence of an unexpected vein or the lack of collateral drainage from a region drained by a vein that may need to be sacrificed during surgery can alter the approach to resection of an aneurysm. Kaminogo and colleagues (2002) used 3D CTA to demonstrate the venous anatomy accurately. They showed the usefulness of this information in selecting a therapeutic procedure (surgery versus endovascular coiling) and in planning the approach for surgical treatment.

Recent data indicate that CTA is a safe and accurate imaging technique for evaluating most extracranial and intracranial vessels to detect the presence of stenoses or occlusions, as well as for the detection of intracranial aneurysms. The accuracy of CTA appears equal to or superior to MRA imaging in most circumstances, and its accuracy sometimes approaches that of DSA. The development of new advanced CT scanners with more detectors may further enhance the accuracy of this technique (Latchaw et al., 2009).

Time-of-Flight Mra

Earlier clinical reports outlined the advantages and pitfalls of TOF-MRA for imaging of the extracranial circulation and estimation of degree stenosis at the carotid bifurcation (Norris and Rothwell, 2001). The degree of stenosis tends to be overestimated by the traditional 2D TOF-MRA method. A corollary of this observation is that a 2D TOF study with normal or near-normal findings effectively excludes the possibility of severe (70% to 99%) stenosis. The most accurate results are obtained when short TE and small voxel size are used. Second, a consensus estimate of stenosis derived from a combination of the 2D and 3D TOF methods results in greater specificity than 2D TOF alone. This improvement results primarily from the inclusion of the 3D TOF method, in which stenosis is less likely to be overestimated, particularly if original (Fig. 33B.9) or reformatted source images are evaluated rather than the MIP images. In the combined TOF approach, the 2D TOF method is used primarily to distinguish slow flow from occlusion and, in general, the combined TOF approach is considered superior to DSA in differentiating high-grade stenosis from occlusion. A flow gap, which is a segmental dropout of signal from the carotid (or other vessels) caused by intravoxel phase dispersion or saturation, is often taken as a sign of stenosis measuring 70% or more. This association should be viewed with caution, however, because in one published series of patients, flow gap was observed with the 2D TOF technique at sites of 50% to 60% stenosis as determined by DSA.

Fig. 33B.9 Carotid stenosis. Comparison of magnetic resonance angiogram and source image to catheter angiogram. A, Catheter angiogram of the right carotid system shows narrowing (arrow) at the origin of the internal carotid artery. Findings suggest intraluminal thrombus. B, Three-dimensional (3D) time-of-flight (TOF) angiogram demonstrates similar narrowing (arrow). C, Axial 3D TOF source image at the site of stenosis shows clearly that the lumen is narrowed (arrow) by approximately 50%. In this case, there is agreement between the narrowing shown on source image and that detected by the maximum-intensity projection image (B).

Third, in detecting stenosis appropriate for carotid endarterectomy, TOF-MRA is less sensitive than DSA (75% MRA, 87% DSA) but more specific (88% MRA, 46% DSA); however, when stenosis estimates by TOF-MRA and DSA are concordant and are then taken together, the combined MRA and DSA examination is more sensitive (96%) and specific (85%) than either study alone. Furthermore, when patients are classified as to whether carotid endarterectomy is indicated by the noninvasive examination and then judged against the results of DSA, the misclassification rate for the concordant MRA and DSA results is much lower than that of either test alone (MRA and DSA 7.9%, MRA 18%, DSA 28%) (Johnston and Goldstein, 2001). Therefore, surgical decisions are more likely to be correct when based on concordant TOF-MRA and DSA results.

Three-Dimensional Contrast-Enhanced Mra

Compared with 2D and 3D TOF-MRA, 3D CE-MRA delineates carotid arterial stenosis better (Willig et al., 1998) (Fig. 33B.10). Surface morphology (e.g., ulcerated plaque) and nearly occluded vessels (e.g., “string sign”) are more easily identified, and arterial occlusions are more confidently identified. Severe carotid bifurcation stenosis may be detected by 3D CE-MRA with high sensitivity (93%-100%) and specificity (88%-96%) (Huston et al., 1998; Johnson et al., 2000; Lenhart et al., 2002; Remonda et al., 2002; Wutke et al., 2002) using DSA as the standard method of diagnosis. TOF techniques yield similar results to 3D CE-MRA for mild to moderate stenosis, thus obviating time spent on CE-MRA setup and processing. Advantages of 3D CE-MRA include greater anatomical coverage (Fig. 33B.11). For high-grade stenosis, which can cause intravascular flow gaps on TOF MIP images, the addition of CE-MRA to the imaging protocol provides sensitivity and specificity equivalent to CTA in determining the severity of stenosis (relative to DSA as the reference standard).

Fig. 33B.10 Carotid stenosis. Comparison of contrast-enhanced magnetic resonance angiography (CE-MRA) with conventional catheter angiography. A, CE-MRA reveals flow gap in proximal left internal carotid artery (ICA) suggestive of high-grade stenosis. B, Digital subtraction angiography (DSA) of the left carotid system confirms greater than 70% stenosis.

Fig. 33B.11 Similar appearance of mild stenosis (arrow) of left internal carotid artery on oblique maximum-intensity projection (MIP) images. A, Three-dimensional (3D) contrast-enhanced magnetic resonance angiography (CE-MRA). B, 3D time-of-flight (TOF) MRA. Note the greater coverage of the carotids afforded by CE-MRA compared with TOF-MRA.

(From Bowen, B.C., 2007. MR angiography versus CT angiography in the evaluation of neurovascular disease. Radiology 245, 357-361.)

TOF-MRA and CE-MRA have been found to achieve high accuracy for the detection of high-grade ICA stenoses and occlusions, with CE-MRA having slight benefit over TOF-MRA. A systematic review and meta-analysis of 37 TOF-MRA studies and 21 CE-MRA studies was performed by Debrey et al. and found that for the detection of high-grade (≥70-99%) ICA stenoses, TOF-MRA had an overall sensitivity of 91.2%, with a specificity of 88.3%. The sensitivity of CE-MRA was higher at 94.6%, with a specificity of 91.9%. For the detection of complete ICA occlusions, the sensitivity of TOF-MRA was 94.5%, and the specificity was 99.3%, while CE-MRA demonstrated a sensitivity of 99.4% and specificity of 99.6%. However, for moderately severe ICA stenoses (50%-69%), sensitivity was found to be poor in TOF-MRA and only fair in CE-MRA studies. TOF-MRA had a sensitivity of only 37.9% and a specificity of 92.1%, while CE-MRA had a sensitivity of 65.9% with a specificity of 93.5% (Debrey et al., 2008).

The ability to detect an FMD pattern of stenoses by MRA in the carotid vessels remains uncertain. This disorder may not be as well delineated on TOF-MRA owing to limited resolution, although no large comparative studies with CE-MRA have been performed. In one series evaluating FMD in the renal arteries, Willoteaux found the sensitivity and specificity of CE-MRA to be 97% and 93%, respectively. These findings suggest that using CE-MRA to identify FMD in the cervical vessels may be possible, although further studies are required (Willoteaux et al., 2006).

Atherosclerotic narrowing of the vertebral artery commonly involves the origin or distal intracranial portion. For TOF-MRA evaluation of posterior-circulation cerebrovascular disease, the vertebral origins usually are not evaluated for the same reasons the common carotid origins are not evaluated. Typically, a 3D TOF study covering the vertebrobasilar system from the C2 level to the tip of the basilar artery is done (Fig. 33B.12). However, sequential 2D TOF-MRA of the neck is useful in determining whether proximal occlusion is present and in demonstrating flow direction in the vertebral arteries in patients with suspected subclavian steal. A 2D TOF study obtained with no presaturation band shows flow enhancement in both vertebral arteries, whereas a study obtained with a superiorly located walking presaturation band shows flow only in the vertebral artery with normal anterograde flow. The 3D CE-MRA techniques can display both the origins and distal intracranial portions of the vertebral arteries in a single acquisition and are particularly useful in evaluating vertebral artery segments with partial or complete signal loss caused by slow flow and in-plane saturation effects. The accuracy of 3D CE-MRA measurements of stenosis at the vertebral artery origin has yet to be reported, although the accuracy is unlikely to equal that of carotid bifurcation measurements because of the smaller size of the vertebral origins (Kollias et al., 1999). Nevertheless, an analysis of the elliptical centric encoding technique predicts that it can achieve an isotropic spatial resolution of 1 mm (before zero filling) in a field of view typically used for bilateral carotid and vertebral imaging (Fain et al., 1999). Stenosis or occlusion of the subclavian artery is now routinely evaluated with 3D CE-MRA.

Fig. 33B.12 Basilar artery stenosis seen on both magnetic resonance angiography (MRA) and computed tomographic angiography (CTA). A, A 3T time-of-flight (TOF)-MRA reveals moderate luminal narrowing in mid to distal portion of basilar artery. B, Three-dimensional reconstruction of CTA images reveals a 1 cm area of irregular narrowing in the basilar artery, with 65% stenosis.

In patients with carotid or vertebral artery dissection, MRA is complemented by MRI with fat-saturation sequences that aid in detection and characterization of dissecting hematoma, associated dissecting aneurysm, and the length and caliber of the residual patent lumen, especially at the skull base (Fig. 33B.13). Subacute hyperintense thrombus is better seen if fat suppression is implemented on thin T1-weighted images to eliminate the high signal intensity from perivascular adipose tissue. When the evaluation includes a 2D or 3D TOF combined study or 3D CE-MRA (Kollias et al., 1999), the detection of stenosis, pseudoaneurysm, or occlusion is improved. Also, the presence of a thrombosed false lumen is more convincingly demonstrated when spin-echo images are supplemented with phase-contrast or TOF images showing absence of flow in the false lumen. The improved recognition of these features afforded by MRA impacts on treatment decisions, such as deployment of stents and other endovascular devices. Serial MRA examinations are required to evaluate for recanalization of the vessel following the dissection (Fig. 33B.14) and also to evaluate for dissecting aneurysms that may occasionally develop as the hematoma resolves.

Fig. 33B.13 Right internal carotid artery dissection with intramural hematoma causing severe narrowing of the residual patent lumen as well as pseudoaneurysm at the skull base. A, Three-dimensional (3D) time-of-flight (TOF) axial source image. B, Fat-suppression T1-weighted axial image at C1-C1 shows the narrowed cervical carotid lumen with flow and the thickened wall with unsuppressed hyperintensity consistent with subacute hematoma (arrows). C, 3D TOF axial source image at the level of the carotid canal entrance shows outpouching of pseudoaneurysm (arrow). D, Oblique maximum-intensity projection image displays the length of the narrowed lumen (between arrows), ending at the carotid canal and the pseudoaneurysm (large arrow).

(From Bowen, B.C., 2007. MR angiography versus CT angiography in the evaluation of neurovascular disease. Radiology 245, 357-361.)

Fig. 33B.14 Resolution of carotid dissection followed with magnetic resonance angiography (MRA). A, MRA produced by the three-dimensional time-of-flight technique shows a segmental stenosis (arrows) involving the distal cervical portion of the left internal carotid artery. B, Repeat study obtained after anticoagulant therapy demonstrates return to normal caliber of the internal carotid artery segment (arrows).

The MRA assessment of vascular stenosis after placement of a metallic stent is limited by turbulence and susceptibility effects. Stent geometry, the relative orientation of the magnetic field, and alloy composition contribute to signal intensity alterations within the stent lumen (Lenhart et al., 2000) (Fig. 33B.15). CTA is also limited in determining stenosis of the lumen of a stented vessel. The limitation is primarily due to streak artifacts from the stent, which obscure the residual lumen. Also, widened window settings and volume averaging often result in overestimation of stenosis on CTA (see Fig. 33B.15).

Fig. 33B.15 Stent device in the distal left vertebral artery. A, Coronal time-of-flight magnetic resonance image demonstrates loss of enhancement in the distal portion of the stent placement, suggesting a severe stenosis. B, Axial images of the neck after contrast administration is unable to accurately determine the degree of residual luminal narrowing. Widening the window settings results in overestimation of stenosis, and a later digital subtraction angiography demonstrated only mild stenosis.

Subclavian Steal Syndrome

Subclavian steal syndrome describes the reversal of normal direction of flow in the vertebral artery ipsilateral to a severe stenosis or occlusion occurring between the aortic arch and vertebral artery origin. DSA remains the standard in visualizing disease in the great vessels, along with abnormal retrograde filling of the affected vertebral artery. Given the invasive nature of DSA, Doppler sonography is often used, but this study may be limited by lack of visualization of the relevant pathology in the subclavian artery. Therefore, MRA offers a reliable comprehensive means to test patients with suspected subclavian steal syndrome. PC-MRA methods encode direction of flow and can accurately depict subclavian stenosis along with reversal of flow in the vertebral artery. Although TOF-MRA does not possess true flow-encoded information, flow direction can be deduced with suppression of flow from a single direction by a saturation pulse that allows for selective arterial or venous MRI, with reversal of flow presenting as a flow void. This finding may also be seen with severe stenosis or occlusion but may be distinguished by anatomical imaging of vessel patency, such as with 3D CE-MRA. A 3D CE-MRA has a potential disadvantage in the lone evaluation of subclavian steal syndrome, because it does not possess inherent flow-encoded information. However, the low-resolution 2D TOF localizer acquisition that is often performed beforehand has been shown to provide the same information as a formal TOF-MRA sequence (Sheehy et al., 2005).

Acute Ischemic Stroke

As advances are made in rapid MRI techniques, a novel application for MRA has been the use of this technique to potentially guide acute stroke intervention. In a study by Marks et al., patients with isolated MCA occlusion and early recanalization at TOF-MRA demonstrated more favorable clinical response than those with tandem ICA/MCA occlusion and early recanalization, among those treated with IV tissue plasminogen activator (tPA) 3 to 6 hours after symptom onset (Marks et al., 2008). The correlation of early angiographic findings with clinical outcomes in stroke has prompted further investigation into utilizing MRA as predictive instrument. The Boston Acute Stroke Imaging Scale (BASIS) was devised as a simple classification instrument to help predict outcomes in acute stroke. In this algorithm, patients are classified as having major stroke if they either demonstrate a proximal cerebral artery occlusion (distal intracranial ICA, proximal [M1 or M2] MCA, and/or basilar artery) or if large parenchymal acute ischemia is found on diffusion MRI or early CT imaging in the event no proximal cerebral artery occlusion is identified. All other circumstances are classified as a minor stroke by imaging. Using this classification, Torres-Mozqueda et al. found that a total of 71.4% of major stroke survivors and 15.4% of minor stroke survivors were discharged to a rehabilitation facility, whereas 14.3% and 79.2%, respectively, were discharged to home. The mean length of hospitalization was 12.3 days for major stroke and 3.3 days for minor stroke. Prospective validation of the BASIS classification instrument, along with future investigations examining dynamic MRA changes, remain promising potential uses of MRA for guidance of stroke treatment (Torres-Mozqueda et al., 2008).

Cerebral Aneurysms

Three-dimensional TOF-MRA is now readily accepted as a noninvasive screening tool for familial aneurysmal disease. It has also been used as an alternative to DSA for the surgical management of patients with ruptured aneurysms. A review of the relevant literature for 1988 to 1998 found that TOF-MRA (and CTA) depicted aneurysms with an accuracy of about 90% (White et al., 2000). Sensitivity was much greater for detection of aneurysms larger than 3 mm (94%) than for detection of aneurysms 3 mm or smaller (38%). Diagnostic accuracy was similar for anterior and posterior circulation aneurysms (Fig. 33B.17). In general, noninvasive imaging evaluation includes a review of T1- and T2-weighted (fast) spin-echo images and T2*-weighted gradient echo images, in addition to the source images and MIP images from the MRA acquisition.

Fig. 33B.17 Anterior communicating artery aneurysm. A, A three-dimensional time-of-flight magnetic resonance angiogram (3D TOF-MRA) on 1.5T reveals a lobulated, saccular aneurysm arising from the junction of the A1 and A2 segments. B, Digital subtraction angiogram (DSA) prior to coil embolization also demonstrates this anterosuperiorly directed aneurysm.

The role of 3D TOF-MRA in assessing intracranial aneurysms before endovascular treatment is adjunctive to the definitive DSA study (Adams et al., 2000). Based on a composite assessment that included aneurysm detection rate, aneurysm morphology, neck interpretation, and branch vessel relationship to the aneurysm, Adams and colleagues found MRA to be inferior to DSA overall and to have missed aneurysms smaller than 3 mm. Nevertheless, MRA provided complementary information to DSA in anatomically complex areas or in the presence of intramural thrombus. The authors applied the assessment to four different types of image data display: axial source images, multiplanar reconstruction (MPR) of the source image data, MIP images, and 3D isosurface-rendered images. Among these types of images, the MPR and 3D isosurface-rendered images were comparable to the DSA images in all categories of the composite assessment, whereas the MIP images scored poorly in all categories except aneurysm detection. These findings indicate that better noninvasive characterization of aneurysms with TOF-MRA can be achieved by adding MPR or 3D isosurface-rendered images to the source and MIP images that are now routinely reviewed in clinical practice.

Endovascularly coiled intracranial aneurysms are increasingly being followed with MRA imaging to detect aneurysm recurrences, although the data remain mixed regarding its accuracy (Fig. 33B.18). One meta-analysis evaluated 16 studies that compared 1.5T TOF-MRA and 1.5T CE-MRA with DSA in the follow-up of coiled intracranial aneurysms. Pooled sensitivity and specificity of TOF-MRA for the detection of residual flow within the aneurysmal neck or body were 83.3% and 90.6%, respectively. Pooled sensitivity and specificity of CE-MRA for the detection of residual flow were 86.8% and 91.9%, respectively, but were not found to be significantly different. The included studies were of moderate methodological quality, and all pooled estimates were subject to heterogeneity (Kwee and Kwee, 2007). A prospective analysis was performed to compare TOF-MRA and CE-MRA at 1.5T and 3T to a reference standard of DSA in the evaluation of previously coiled intracranial aneurysms. For the detection of any aneurysm remnant, the sensitivity was 90%, 85%, 88%, and 90% for 1.5T TOF, 1.5T CE, 3T TOF, and 3T CE-MRA, respectively. These sensitivities dropped to 50%, 67%, 50%, and 67%, respectively, for the detection of only larger (class 3 and 4) aneurysm remnants, because several of these remnants were underclassified as a smaller remnant by MRA. CE-MRA at 1.5T and 3T had a better sensitivity for larger remnants than TOF-MRA, which may be related to greater flow-related artifacts within larger aneurysm remnants on TOF-MRA compared with the luminal contrast-filling characteristics of aneurysms on CE-MRA. Specificities of these four MRA techniques for detecting any aneurysm remnant were 52%, 65%, 52%, and 64%, respectively. Specificities improved to 85%, 84%, 85%, and 87%, respectively, for the detection of larger (class 3 and 4) aneurysm remnants, reflecting the difficulty in detecting smaller remnants with MRA. Regarding the detection of any aneurysm growth since previous comparison angiograms, sensitivities for these MRA techniques were 28%, 28%, 33%, and 39%, respectively, and specificities were 93%, 95%, 98%, and 95%. Although CE-MRA is more likely than TOF-MRA to classify larger aneurysm remnants appropriately, TOF-MRA better identifies the location of coil masses and may be more advantageous if suboptimal CE-MRA contrast bolus is given. Therefore, the advantage of CE-MRA over TOF-MRA remains uncertain and consideration for both exams may be made in the follow-up of patients with coiled intracranial aneurysms (Kaufmann et al., 2010).

Fig. 33B.18 Right ophthalmic artery aneurysm following coil embolization. A, Computed tomographic angiography source image nondiagnostic for residual lumen due to streak artifacts. B, Three-dimensional time-of-flight magnetic resonance angiogram (3D TOF-MRA) axial source image at level of aneurysm dome reveals central and eccentric hypodensity due to packed coils and peripheral hyperintensity due to flow-related enhancement in residual lumen. C, 3D TOF-MRA axial source image at level of aneurysm neck also shows evidence of flow through patent neck remnant (arrow). D, Coronal maximum-intensity projection image demonstrates continuity of flow into neck and dome remnants of coiled aneurysm (arrows).

(From Bowen, B.C., 2007. MR angiography versus CT angiography in the evaluation of neurovascular disease. Radiology 245, 357-361.)

Venous Disorders

Imaging of the cerebral venous system differs from approaches tailored to the arterial system because of the lower velocity of venous flow and the morphology of the venous sinuses. For suspected dural sinus occlusion resulting from thrombosis or tumor invasion, the primary technique is 2D TOF-MRA in conjunction with spin-echo imaging. To establish the diagnosis of venous thrombosis, lack of visualization of a vein or sinus on the source images and angiogram must be accompanied by identification of the clot on the spin-echo images at the location of the suspected occlusion (Fig. 33B.19). With the 2D TOF technique, optimal flow enhancement in a section is achieved when it is perpendicular to the flow direction. This condition is best approximated using coronal sections to image the sagittal, straight, and transverse sinuses (as well as the internal cerebral veins, basal veins of Rosenthal, and to a lesser degree, the vein of Galen). The acquisition of coronal sections can be augmented by the acquisition of oblique sagittal sections to allow better flow enhancement in the posterior portions of the transverse sinuses and the cortical veins draining into the superior sagittal sinus. With the 2D TOF-MR venography technique, arterial signal is reduced or eliminated by an axial presaturation band placed across the upper neck below the skull base. A common diagnostic pitfall of the technique is the presence of flow gaps in the transverse sinus. Ayanzen and colleagues (2000) observed these gaps in 31% of patients with normal MRI findings. Flow gaps were not observed in the superior sagittal, straight, or dominant transverse sinuses, so gaps occurring in these locations should raise suspicion of venous obstruction. The authors found that the nondominant transverse sinuses (90% of gaps) or codominant transverse sinuses (10%) that demonstrated the flow gaps were hypoplastic yet patent by DSA.

Fig. 33B.19 Acute superior sagittal sinus thrombosis. A, Magnetic resonance angiogram (lateral view) produced from 80 sequential two-dimensional time-of-flight coronal sections covering posterior half of the head. Flow-related signal is observed in transverse (open arrows) and sigmoid sinuses but not in superior sagittal sinus (closed arrows) or straight sinus. Short segments of vessels projecting over posterior course of superior sagittal sinus represent patent superficial cerebral veins lateral to the sinus. B, Postgadolinium midsagittal T1-weighted image shows hypointense signal in superior sagittal sinus and enhancing margins. Combined with the results in A, findings are consistent with the presence of intraluminal thrombus.

Alternative techniques for demonstrating cerebral veins and dural venous sinuses lack the robustness of the 2D TOF technique. The 3D TOF technique suffers from saturation effects and hence frequent signal loss in the veins and dural sinuses. The 2D phase-contrast technique is limited by gradient heterogeneity, eddy currents, aliasing artifacts, and lower spatial resolution. Although the phase-contrast technique can be useful in differentiating very slow flow in the dural sinuses from thrombosis, it was recently found to be inferior to the 2D TOF technique and a contrast-enhanced 3D FLASH (fast low-angle shot) MRA technique in displaying the normal septal veins, internal cerebral veins, and the basal veins (Kirchhof et al., 2002). Both the 3D FLASH technique and a 3D magnetization-prepared, rapid-acquisition gradient echo MRI technique (Liang et al., 2001) are contrast-enhanced methods and reportedly are superior to the 2D TOF technique in depicting normal venous structures, especially in overcoming the flow gap artifact. These contrast-enhanced techniques, though, have two potential limitations: (1) both techniques involve rapid acquisition (1-2 minutes per 3D volume), so the intensity of the intravascular signal depends on the timing of the contrast infusion relative to data acquisition, and (2) chronic thrombus may enhance with gadolinium and mimic a patent lumen. Consequently, these two contrast-enhanced techniques should be viewed as adjuncts to the 2D TOF technique.

Vascular Malformations

Traditional MRA methods (2D and 3D TOF and PC-MRA) have played a secondary role to DSA in evaluating intracranial arteriovenous malformations (AVMs) because of a lack of consistent and complete demonstration of all components of an AVM: feeding arteries, nidus, and draining veins. For this reason, and because of the general impression that TOF-MRA adds little to the spin-echo MRI findings useful for preliminary staging of an AVM (nidus size and location and central versus peripheral pattern of venous drainage in the Spetzler-Martin Grading Scale) before definitive DSA, many investigators have considered TOF-MRA of AVMs superfluous (Table 33B.1). Phase-contrast techniques have been used by some investigators to estimate blood flow velocities and volume flow rates in the largest arteries supplying the AVM. A high-resolution 3D gradient echo technique based on the paramagnetic property of deoxyhemoglobin has been used to detect cerebral veins with submillimeter resolution, resulting in greater sensitivity in identifying the presence of small AVMs. But compared with TOF-MRA, the technique provides poorer detection of feeding arteries and is markedly limited in its delineation of nidus size and shape when there are susceptibility artifacts from nearby bone, air, or blood products (hemosiderin) (Essig et al., 1999). High-resolution, real-time, auto-triggered, elliptic, centric-ordered 3D CE-MRA (Farb et al., 2001) and lower-resolution time-resolved 2D CE-MRA (Klisch et al., 2000) have also been applied to the evaluation of AVMs and the results compared with those of TOF-MRA and intraarterial DSA. In an initial investigation, Farb and colleagues found that their 3D CE-MRA technique was superior to 3D TOF-MRA, particularly in depicting nidus and draining veins. The 3D CE-MRA technique consistently showed AVM components and their spatial relationships on MIP images and was equivalent to DSA in depicting AVM components in 70% to 90% of cases (total patients = 10), based on blinded independent assessments by two experienced neuroradiologists. Several small studies prospectively examined 4D CE-MRA at 3T for evaluating AVM and found that 4D CE-MRA yielded 100% accuracy in classifying Spetzler-Martin classification (nidus size, venous drainage, eloquence) of cerebral AVMs when compared to DSA. One study demonstrated a 93% rate of detection of feeding arteries using 4D MRA alone. However, the combined use of selective ASL provided additional functional or anatomical information in 4 of 16 cases (25%), enabling the detection of a cross-filling feeding artery that was not identified by 4D MRA without selective arterial spin labeling (ASL) technique, and improved sensitivity to 96% (Kukuk et al., 2010). These novel techniques may prove to be a noninvasive alternative to DSA in the assessment of patients with cerebral AVMs (Hadizadeh et al., 2008).

Dural arteriovenous fistulas (AVFs) most commonly involve the cavernous, transverse, and sigmoid sinuses along the skull base. Arterial feeders not seen on spin-echo MR sometimes are detected on 3D TOF-MRA but much less often than on catheter angiography. Transverse and sigmoid sinus occlusion and dilated cortical veins are detected better by MRA than spin-echo imaging, yet neither technique achieves the accuracy of catheter angiography. Traditional 3D TOF-MRA is useful in detecting cavernous sinus fistulas, because flow enhancement in the cavernous sinus and contiguous veins can provide evidence of the fistula (Fig. 33B.20). This finding must be regarded with caution because venous flow signal has been observed in the cavernous sinus and inferior petrosal sinus in a variable percentage (from 4% to 36%, depending on unspecified technical differences between MR scanners) of patients without clinical evidence of carotid-cavernous fistula (Ouanounou et al., 1999). Farb et al. compared the use of 3T time-resolved MRA for the diagnosis and classification of a cranial dural AVF with DSA. Patients underwent a commercially available contrast-enhanced time-resolved imaging of contrast kinetics (TRICKS) MRA (trMRA) technique that combines k-space segmentation into central, mid, and peripheral zones with superimposed elliptic centric-view ordering. High temporal resolution is obtained by sampling the central zones of the k-space at a more frequent rate than peripheral zones. The high spatial and temporal resolution allows for identification of dural AVFs by identifying both early opacification of a dural sinus and reflux into the connecting cortical veins if present. In 93% (39/42) of dural AVF cases, investigators had unanimously correct interpretations of the trMRA to correctly identify or exclude all fistulas and accurately classify them when found. The small series suggests that trMRA may be a reliable technique in the screening and surveillance of DAVFs in certain clinical situations, but further validation studies are required (Farb et al., 2009).

Fig. 33B.20 Spontaneous resolution of a left carotid-cavernous dural arteriovenous fistula (AVF). Magnetic resonance angiograms (axial maximum-intensity projection images) of sellar region and circle of Willis were acquired with three-dimensional time-of-flight technique (no gadolinium enhancement). The studies, performed at the time of clinical presentation (A), 3 months later (B), and 3 years later (C), show progressive resolution of venous drainage from the AVF. Flow-related signal in left cavernous sinus (closed arrow), sphenoparietal sinus (open arrow), and cerebral veins results from shunting of high-flow-rate arterial blood through fistula. Note progressive decrease in signal in the sphenoparietal sinus. Flow-related signal in the left orbit is caused by the ophthalmic artery (arrowheads), not the superior ophthalmic vein, which was found by catheter angiography to be thrombosed at the time of presentation.

The diagnostic imaging features of venous malformations (angiomas, developmental venous anomalies) are well shown on postgadolinium T1-weighted spin-echo images. These features include the radially oriented collection of small vessels (medullary veins) that produce a caput medusa or spoke-wheel configuration. This is contiguous with a large trunk vein that drains into either subependymal or superficial cerebral veins or a dural sinus. The 2D TOF technique and the phase-contrast method with low anticipated maximum blood flow velocity often display these slow-flow malformations without the use of gadolinium and allow determination of flow direction. However, the 3D TOF technique, which provides greater spatial resolution, requires gadolinium to avoid saturation effects. Cavernous malformations and capillary telangiectasias do not show flow enhancement on MRA studies, and the former usually are identified on spin-echo and gradient echo images by the heterogeneous signal intensity caused by blood products from prior hemorrhage.