Optimal contrast administration is critical to the successful performance of CTA (10–18). Considerations include volume, route and rate of injection, iodine concentration and osmolality, use of a saline flush, timing of scan acquisition, and methods to diminish scan artifacts related to contrast administration.

Although adequate vascular opacification can be obtained with as little as 50 mL of 300 mg I/mL contrast media followed by a saline flush (10), most investigators use a considerably higher volume of contrast for CT angiographic applications. Typically, 75 to 100 mL of nonionic contrast material (averaging 300 mg I/mL) is injected through an antecubital vein using an 18- to 20-gauge catheter, at rates varying between 4 and 5 mL/second, with images obtained following a test bolus of contrast or an automated triggering approach. A right arm approach is preferred to minimize streak artifact across the aortic arch and proximal vessels and to eliminate the possibility of contrast obstruction as the left innominate vein crosses the sternum. Nonionic contrast agents are used to minimize the effects of nausea and vomiting as well as complications arising from extravasations caused by the rapid infusion of ionic contrast agents. Patients with renal insufficiency can be studied with a reduced dose (50 mL) of a low-osmolar agent or with gadolinium diethylenetriaminepenta-acetic acid (DTPA).

The use of injectable saline solution via a power injector as a means to deliver contrast material is widely employed in coronary CTA to clear the right ventricle of residual contrast, allowing better visualization of the right coronary
artery. Before the advent of dual-head power injectors, Hopper et al. (18) first loaded 75 mL of 60% nonionic contrast medium into a power injector in a vertical position, followed by 50 mL of normal saline solution loaded so as to lie on top of the denser nonionic contrast material. Meant to decrease the volume and cost of nonionic contrast material utilization, while using the saline to clear the syringe and intravenous catheter of contrast material following the injection, this approach has the added benefit of significantly decreasing the number of streak artifacts from adjacent venous structures. Newer power injectors contain multiple ports to provide an automated saline flush following the bolus of iodinated contrast with freely selectable flow rates.

In addition to considerations of flow rates and volume, it is also essential to optimize scan delay to ensure adequate opacification of targeted vessels. Although a 20- to 30-second scan delay most often is sufficient to ensure adequate opacification of the thoracic aorta, large variations between patients will be encountered. As reported by Van Hoe et al. (19), in their evaluation of scan delay times in spiral CTA using a test bolus injection, the time to peak attenuation within the aorta varied between 11 and 32 seconds (mean 20, standard deviation 6.1 seconds). This problem is easily resolved by administering 20 mL of contrast media injected at the same rate as the planned injection, with images obtained at the same preselected level (usually the middle ascending aorta or arch) using 80 to 90 mAs. As discussed in Chapter 3, a similar approach may be used to optimize evaluation of the pulmonary arteries. Alternatively, it is now possible to obtain equivalent data using automated scan acquisitions to monitor vascular opacification without the use of a test dose [C.A.R.E bolus (Siemens) or Smartprep (GE Medical Systems, Milwaukee, WI)]. Although administration of a test dose of contrast does have the theoretical deleterious effect of increasing the background attenuation of abdominal organs, rendering visualization of small vessels more difficult with CTA, this consideration is of much less concern within the thorax. An advantage of using a test dose (timing bolus) over an automated triggering technique is the ability to test the integrity of the intravenous line without having to give the entire bolus of contrast. Furthermore, it is not uncommon for an oncologic patient to have an occlusion of a great vein, resulting in collateral pathways of contrast to the aorta with subsequent suboptimal aortic enhancement that may preclude accurate triggering.

Gadolinium-enhanced 3D MRA examinations are usually performed on a 1.5 or 3 T system with a high performance gradient system that allows a TR of less than 4 ms and a TE of less than 1.5 ms. A 4-, 8-, or 16-channel multiarray torso coil is preferable, as it provides an advantage over the body coil by virtue of increased signal-to-noise and provides for higher spatial resolution or faster imaging using parallel imaging techniques. This benefit, however, is proportional to the patient’s body habitus; the increase in signal-to-noise is substantial in thin patients but may be less in obese patients. There are distinct disadvantages in using the multiarray coil that do not apply to the body coil. The FOV is limited in the craniocaudad dimension, which may result in the need for a second angiogram to evaluate the infrarenal aorta in patients with aortic dissection and diffuse aneurysmal disease. However, recent technology allows for coil coupling to image as many anatomic segments as one desires during the same breath-hold. For example, a 16-channel neurovascular coil can be linked to a multichannel torso
coil, allowing imaging of the thoracic aorta to the circle of Willis with a single injection. The bright signal of the subcutaneous fat may also obscure vessels of interest on MIP images, especially if a strong rectangular FOV is used. However, use of subtraction algorithms can minimize this problem if there is good coregistration between the precontrast and gadolinium-enhanced acquisitions. In addition, portions of the aorta closest to the coil often have more signal than segments deeper in the chest, making the MIP images difficult to window appropriately (Fig. 2-5). This can also be minimized with the use of a postprocessing filter function. Finally, in some patients (especially children), the weight of the anterior component of the coil is uncomfortable and may potentiate claustrophobia. For these reasons, some centers still use the body coil when evaluating the thoracoabdominal aorta (Fig. 2-4). However, the body coil is clearly suboptimal when evaluating small vessels, such as the vertebral arteries, because the small FOV needed to resolve these vessels (25 cm) results in poor signal-to-noise. This is compounded by the fact that the very short echo time intrinsic to these 3D sequences (1 to 2 ms) necessitates the use of high bandwidths. This in turn also results in a signal-to-noise penalty.

When studying infants or neonates, the head or knee coil can be used, and spine coils may be used for young children and adolescents with thin body habitus. Larger children and adolescents can be studied in either the multiarray coil or the body coil.

Regardless of the coil selected, the patient’s arms are positioned at the sides, and the right antecubital fossa is catheterized so that contrast draining the injection site is not present in the left brachiocephalic vein (which could obscure arch vessels on MIP images). Another pertinent reason not to inject from the left side is that some patients with thin body habitus may have physiologic compression of the left innominate vein against the sternum, resulting in restriction of flow past the lesion with filling via collateral vessels (44). In a study by Lee et al. (44) 9.1% of 475 carotid MRA studies were poorly enhanced because of left arm injections. Compression of the left brachiocephalic vein between the sternum and aorta was confirmed in four cases by venography, CT, and MRI.

Although some authors advocate placing the patients’ arms over their heads (to minimize aliasing artifacts), this should not be performed for arch studies, as this position may cause compression of the subclavian artery and vein against the first rib. This compression can be significant in normal patients (without pathologic thoracic outlet syndrome) and can lead to a false-positive diagnosis of occlusive disease in the subclavian arteries as well as susceptibility artifacts from concentrated gadolinium in the axillary vein draining the injection site (Fig. 2-6). For this reason, when specifically evaluating the subclavian, axillary, or brachial arteries, the injection should be administered in the side contralateral to the expected pathology so that concentrated gadolinium in the draining vein will not result in artifactual signal loss from susceptibility artifacts. In addition, the use of a large-volume saline flush (20 mL) will help minimize these artifacts. If bilateral disease is anticipated, there are two options: (a) place the intravenous site in a lower extremity or (b) dilute the gadolinium to a 33% solution of normal saline, increase the flow rate to 3 to 4 mL/second, and use the shortest possible echo time the scanner will allow (45). One important point in differentiating real from artifactual stenosis is the absence of collateral vessels seen in pseudolesions.

Given the wide range of variables currently available for performing MRA, we recommend a 3D GRE sequence using the following parameters: [2–4/1–2/30–50 (TR/TE/flip angle)], matrix of 256 × 512 with frequency encoding superiorly to inferiorly, a 4/8 to 7/8 rectangular FOV with a maximum dimension of 25 to 45 cm (dedicated arch vessel studies require a smaller FOV than thoracic aorta examinations). A coronal or oblique sagittal/left anterior oblique equivalent slab is used (thickness 9 to 12 cm), 60 to 100 partitions,
effective interpolated section thickness of 1 to 2 mm, 1 acquisition (acquisition time of 5 to 20 seconds). Saturation pulses and gradient moment nulling are not used. Partial Fourier and parallel imaging can be used to decrease acquisition time, as can TRICKS or TREAT.

It is important to position the slab from axial images at multiple levels to avoid excluding portions of extremely ectatic aortas. An unenhanced acquisition is initially performed to ensure that the vessels of interest are included in the imaging volume, to optimize the rectangular FOV that can be used without aliasing artifacts, to determine the patient’s ability to suspend respiration, and for use as a mask for subsequent image subtraction. This subtraction technique is helpful in eliminating the bright, subcutaneous fat, which may degrade the quality of MIP images (46).

To precisely match peak arterial enhancement to the acquisition of the central lines of k-space, a timing examination can be performed to measure circulation time to the region of interest (antecubital vein to aortic arch). This technique employs a 1-mL bolus of Gd-DTPA followed by 20 mL of saline, both infused at 2 to 3 mL/second while 1-cm axial magnetization-prepared GRE images (TR/TE/FA/inversion time = 11 ms/4.2 ms/15 degrees/300 ms/), one every 2 seconds for 60 seconds, are obtained (46). If a dark blood GRE pulse sequence is not available, one can use an ordinary GRE pulse sequence performed in the sagittal plane. Under these circumstances, because of in-plane saturation effects, a test bolus of gadolinium will result in transient aortic enhancement significantly greater than the precontrast image.

The time to peak arterial enhancement is determined by selecting the slice with the greatest qualitative enhancement. This is less cumbersome than drawing region-of-interest cursors over the desired vessel.

Using the circulation time, an imaging delay between the start of the infusion of contrast and the start of the 3D acquisition is calculated with the following equation (32):

Imaging delay = circulation time – (infusion time/2) – (imaging time/2)

This delay time is optimized for sequential phase-encoding pulse sequences, whereby the important low spatial frequency lines that determine image contrast (center of k-space) are sampled in the center of the sequence acquisition. When using centric or elliptical centric phase-encoding MRA pulse sequences, these important lines are acquired at the beginning of the acquisition, and the formula may need to be adjusted by adding an extra imaging delay of 2 to 4 seconds.

Once the imaging delay time is calculated, the identical 3D gradient-echo sequence is repeated with injection of 20 to 30 mL of gadolinium at 2 to 3 mL/second with an MR-compatible power injector followed by a 20-mL saline flush; an inadequate flush may result in artifactual signal loss from concentrated gadolinium, as previously described (Fig. 2-6). The examination is performed at end-inspiration, as most patients find this more comfortable than breath holding at end-expiration, although the latter usually provides for better subtraction imaging with less misregistration. A second acquisition is performed either immediately after the first or after a 10- to 15-second respite during which patients can catch their breath. The purpose of this second acquisition is to better opacify the false lumen in patients with communicating dissection and as a “bail-out” if the first acquisition is suboptimal from acquiring data too early.

Following gadolinium-enhanced 3D MRA breath-hold T1-weighted 3D axial gradient-echo images can be acquired through the chest to evaluate for extraluminal disease and as a general anatomic survey (47). These images are easy to interpret, as they are very similar in appearance to contrast-enhanced CT and can readily detect pulmonary nodules as small as 1 cm as well as mediastinal lymphadenopathy (48). In active vasculitis and arteritis, these delayed 3D axial MR images provide the best contrast to demonstrate mural enhancement, often seen in early disease prior to the development of stenotic lesions.

Prior to gadolinium-enhanced MRA, axial and either oblique sagittal or coronal ECG-triggered black blood T1-weighted SE images were acquired through the chest. This technique provides excellent anatomic images and is extremely useful to evaluate for mural pathology, such as IMH and the presence of pleural or pericardial effusions. Although it is not always possible to achieve high-quality black blood luminal images because of poor ECG triggering or flow artifacts, the use of longer echo times (<15 ms) and specialized dephasing gradients may be helpful. Some authors believe that aortic diameter is more reproducibly measured using black blood techniques, because of the excellent contrast between the aortic wall and the adjacent lung, unless the aorta is extremely tortuous. In this situation, true short axis aortic measurements can be achieved with reformatted images from the gadolinium-enhanced 3D dataset. Although many centers still routinely perform ECG-triggered T1-weighted SE MR for black blood imaging, some institutions have replaced this time consuming technique with dark blood half-Fourier single shot turbo spin-echo (HASTE) imaging (26, 49). This technique is insensitive to patient motion, can provide excellent image quality even with poor ECG gating, does not require breath holding, and takes less than 1 minute to acquire 20 images. Dark blood images are obtained by using two separate inversion pulses (double IR) so the signal from incoming blood is nulled. Because of the limited signal-to-noise inherent in single shot imaging, a phased-array coil must be used. Contrast is also different with this technique and reflects the echo time of the pulse sequence used.

Alternatively, an ECG-triggered double-inversion, single-slice breath-hold technique can be used for higher spatial resolution techniques, such as detailed evaluation of thoracic aortic atheromas (50). By applying an additional inversion pulse, a triple IR pulse sequence can be obtained with fat suppression. It is important to point out that this fat suppression technique is based on inversion time and is not specific for fat. For example, a bronchogenic cyst with the same T1 of fat will also lose signal on a triple IR pulse sequence.

ECG-triggered turbo SE T2-weighted or turbo short T1 inversion recovery (STIR) images have a limited role in the evaluation of aortic disease with the exception of patients suspected of having graft infections or aortitis. In the latter clinical setting, high signal intensity within the aortic wall may correlate with disease activity.

Contrast-enhanced MRA provides excellent anatomic images but cannot quantify flow or pressure gradients. Under these circumstances, cine phase-contrast MR with velocity encoding can supply additional information, including measurement of pressure gradients across stenotic valves and areas of aortic narrowing (coarctation) (51, 52). These sequences can also be used to differentiate the true from false lumen in aortic dissection and to determine if flow in the false lumen is antegrade or retrograde.

MPVRs are a variant of MPRs that allow a user to obtain 2D representations from a 3D volume (or slab) at any angle using average, ray sum, MIP, or even minimum intensity projections (MINIPs). Familiar as a common method for visualizing both MRA and CTA data, MPVRs may be obtained in any direction, including the axial plane, thereby simulating true cross-sectional images (57). Of these, the most commonly used for imaging vessels are MIPs (58). MIPs are derived by projecting onto an imaging plane the brightest voxel encountered in a ray through the scan volume. The result is a 2D image in which blood vessels are typically highlighted (Fig. 2-7A). Alternatively, voxel values along individual rays may be either averaged or summed; averaged images mimic the appearance of routine MPRs, whereas summed vessels appear translucent. MIPs encode variations in voxel attenuation or intensity, allowing differentiation between calcium and thrombus; they are not threshold dependent. As a result,
they also may be of value by depicting differences in flow rates, as, for example, occurs in aortic dissection or even within single vessels as a result of antegrade flow. However, unlike other volume rendering techniques, those derived using MPVRs have no depth cues. As a consequence, MIPs are best suited for displaying anatomy in which superimposition of structures is minimized. This limitation may be overcome in part by multiple projections viewed in a cine mode as well as volume editing, restricting MIPs to only a few transaxial images—so-called thin slab MIPs—to remove unwanted overlying structures. Nonetheless, MIP images, as all other MPVRs, are best interpreted in conjunction with the original axial source images (58). The MIP algorithm is usually most effective when the objects of interest are bright and there is little gradation in the image. High signal intensity from sources other than vessels can interfere with visualization; however, this can be mitigated on MR imaging by using image subtraction. MIPs are simple and quick to perform, but they retain only approximately 10% of the data from the original image.

Although both MDCT and MR have been successfully used to diagnose coarctation, in our opinion, MR has proved of greater value, especially in the pediatric age group as well as for routine postsurgical evaluation, as pressure gradients across the stenotic lesions can be determined (107).

Various traditional MR imaging techniques, performed in the axial and oblique sagittal plane, have been successfully used in the evaluation of coarctation, to assess the location, the degree of anatomic narrowing, and the extent of collateral circulation. These include multiplanar ECG- triggered TI-weighted SE techniques (108) supplemented by phase-contrast (109) or cine MRA (110, 111). The latter technique, when performed in the oblique sagittal plane, can readily demonstrate the dephasing that occurs from turbulent flow across the site of narrowing. Cine phase-contrast techniques, including velocity-encoded cine MR, are used to measure the peak velocity across the site of coarctation (107). With this information, the pressure gradient across the coarctation can be estimated. One study demonstrated that peak coarctation jet velocity measured by MR velocity mapping was comparable to results obtained with continuous wave Doppler sonography (51), and a more recent study showed it was similar but added additional morphologic evaluation not assessable by sonography (112). This technique can also be used to determine the extent of collateral circulation by measuring flow immediately distal to the coarctation and at the level of the diaphragm. In normal aortas, flow will
decrease by an average of 8% as it reaches the diaphragm, whereas in patients with hemodynamically significant coarctation, the flow will be augmented by an average of 80% (113).

Gadolinium-enhanced 3D MRA can readily demonstrate the presence of collateral circulation (Fig. 2-33) as well as the degree of associated arch hypoplasia (Fig. 2-29) and concomitant branch vessel abnormalities. It also provides excellent anatomic images of the extent of coarctation without dephasing artifacts from turbulent flow and allows reformation of 3D data into multiple surgically pertinent imaging planes in patients with associated tubular hypoplasia (Fig. 2-29). CT (90) and, more recently, ECG-gated MDCT can also demonstrate the anatomic segment of coarctation and the presence of collateral vessels (114) (Fig. 2-34).

MR may be used to predict the response of patients who are candidates for balloon angioplasty and/or stent placement (115, 116). The best results with balloon angioplasty occur when the stenosis is focal and located just distal to the left subclavian artery. When the stenosis involves the proximal arch or involves a long segment, balloon angioplasty is less successful. As documented by Bank et al. (116) in their study of 12 children with suspected coarctation evaluated both before and after surgical repair, MR proved accurate in identifying those patients who were unlikely to benefit from angioplasty. In addition, MR proved valuable by determining the appropriate balloon size for angioplasty.

All patients with surgically repaired coarctation should undergo surveillance either with MR (111, 112, 117) or with MDCT imaging to detect recurrent coarctation and pseudoaneurysm formation (Fig. 2-35). One study using ECG-triggered 2D TOF MRA with 3D surface renderings demonstrated a sensitivity of 100% for the detection of aneurysms (98). Recurrent coarctation after primary repair may result from multiple causes, including inadequate removal of ductal tissue, failure to repair associated aortic arch hypoplasia, and suture line tension (118). These patients can be treated with transluminal angioplasty and/or stenting with excellent results (102). Alternatively, coarctation resection and/or extra-anatomic bypass grafting from the aortic arch to the descending aorta (Figs. 2-29, 2-30, and 2-35) can be performed with excellent long-term results (103).

Early CT literature using nonhelical scanning demonstrated high accuracy rates for the diagnosis of thoracic aneurysms and their complications (140, 141, 142, 143, 144, 145, 146, 147, 148, 149). Contrast-enhanced MDCT can demonstrate all the gross pathologic features of aortic aneurysms, including dilatation, extraluminal thrombi, displacement or erosion of adjacent structures
(Fig. 2-45), and perianeurysmal thickening and hemorrhage (140, 141, 142, 143, 144, 145). CT has also proven valuable in detecting unusual types of aneurysms, including mycotic (Fig. 2-46) and ductus aneurysms (Fig. 2-47), as well as complications arising from aneurysms, including those secondary to rupture (145, 146, 147, 148, 149). CT also plays an important role in distinguishing TAAs from pulmonary masses adjacent to the mediastinum.

Thoracic aneurysms secondary to atherosclerosis are readily demonstrated as fusiform dilatation of a segment of the aorta, which usually contains mural thrombus, and may demonstrate calcification of the thrombus or the aortic wall (Fig. 2-48). The pattern of mural thrombi and calcifications within aortic aneurysms has been described (146, 147). Calcification within aneurysms is common, occurring in up to 85% of cases, both mural (Fig. 2-45) and within the thrombus itself. The latter type of calcification has been reported to occur in approximately 25% of patients with both thoracic and abdominal aortic calcifications (147). The significance of calcification within thrombi is that this appearance can be mistaken for the displaced intimal calcification seen in aortic dissection and IMH.

Aortic thrombi may be crescentic (Fig. 2-49) or circumferential, in which case the residual lumen appears circular. As a rule, the residual aortic lumen is smooth, although the aortic lumen may develop a very irregular contour, especially in large aneurysms, depending on the extent of thrombus (Fig. 2-48). Unlike IMHs (which are discussed later), thrombus is often circumferential and does not displace a calcified intima. When the thrombus forms a sharp margin with the enhanced lumen, and no displaced intimal calcifications are present, it may be difficult to distinguish a thoracic aneurysm from a chronic aortic dissection with a thrombosed false lumen.

CT can be used to follow up aneurysms, although this involves repeated use of intravenous contrast and ionizing radiation not inherent in MR. An increasing diameter is readily detected with serial scans and helps prompt a surgical decision. CT may also have value in assessing patients with iatrogenic pseudoaneurysms, which may occur following aortic valve replacement, cardiopulmonary bypass, or coronary artery bypass grafting.

With the introduction of volumetric CT, interest has focused on the potential use of retrospective reconstruction techniques to evaluate aortic aneurysms. Quint et al. (150) evaluated the diagnostic contribution of helical CT with multiplanar reformations in the evaluation of 49 patients with thoracic aortic disease (36 aneurysms, 6 penetrating ulcers, 5 dissections, and 2 pseudoaneurysms). Overall diagnostic accuracy was 92%, both with and without multiplanar reconstructions. Although MPRs were of value by displaying pathology in a format more accessible to nonradiologists, their addition failed to change the diagnosis in any patient (150). Furthermore, whereas in two patients

extension of an aneurysm into the aortic arch necessitating the use of hypothermic circulatory arrest was shown more clearly using MPRs, in a third patient multiplanar reconstructions incorrectly predicted this need.

In distinction, Kimura et al. (151) have shown that both SSD and especially CT angioscopy are superior to axial images for assessing the relationship between distal arch aneurysms and the origin of the great vessels. Comparing
the accuracy of 2D axial source images with SSDs and endoscopic renderings in 12 patients with distal aortic arch aneurysms, these authors found that depiction of the relationship between the aneurysm and the orifice of the left subclavian artery was correct in only 5 (42%) of 12 patients, using axial images; in distinction, using 3D display only, this relationship could be discerned in 9 of the 12 (75%), whereas use of 3D endoscopic renderings allowed accurate assessment in 11 of 12 (92%) patients (151). The reasons for failure of axial images to accurately identify the anatomic relationship between aneurysms and the left subclavian artery included the presence of atheromas obscuring continuity and gradual sloping of the proximal portion of the aneurysm obscuring its precise margin. Of the three renderings, CT angioscopy most clearly delineated the distance between the arterial orifice and the border of the aneurysm.

An even more compelling use of CT for evaluating aortic aneurysms has been reported by Rubin et al. (152, 153). Using helical CT angiographic data, these authors demonstrated that it is possible to automatically derive accurate quantitation of arterial cross sections, segment lengths, and absolute curvature by continuously describing variations in cross-sectional dimensions and as a function of position along the median vessel path. Previously limited, especially in patients with markedly tortuous aortas, use of quantitative vascular CT should now allow more accurate assessment of longitudinal changes in aortic dimensions with time, as well as more accurate treatment planning, especially in those cases for which endoluminal stent-grafts are planned. This is in addition to the already well established role of CT in pre- and postoperative evaluation of patients with both surgical and endovascular stent-graft repair of TAAs (154, 155, 156, 157, 158).

In our experience, the extent of aortic aneurysms, and in particular their relationship to adjacent structures, are better visualized using either MPRs, curved MPRs, or 3D renderings. In this regard, the use of MIPs to delineate the exact location of calcification may be of particular value in assessing the necks of aneurysms.

CT does have some recognized limitations in the evaluation of aneurysmal disease, aside from the risks of iodinated contrast and radiation. Evaluation of the aortic root is often difficult secondary to motion artifacts unless ECG gating is performed (Fig. 2-39). Many patients with advanced aneurysmal disease often have concomitant renal insufficiency; however, with a 64-detector CT, optimized timing, and judicious use of a saline flush, adequate studies can be performed with as little as 50 mL of contrast. Gadolinium chelates have been successfully used for CTA as an alternative to iodinated contrast (159, 160), but the high dose needed for a diagnostic quality scan (0.4 mg/kg) is much more expensive than nonionic iodinated contrast and has recently been linked to a rare kidney disease.