One of the initial steps in performing thoracic CT angiography is to select the anatomic coverage needed to address the clinical question. At a minimum, the anatomic coverage should extend from just above the thoracic inlet so that the proximal portions of the vascular structures in the neck are included in the CT scan. Caudally, the scan should extend at least 1 to 2 cm below the diaphragm. A greater degree of coverage may be needed in some clinical settings. For instance, in the evaluation of sequestration, imaging through the upper abdomen is often necessary because the anomalous artery may rise from the upper abdominal aorta. Similarly, a greater degree of abdominal coverage is needed when assessing hemiazygous continuation of the inferior vena cava.

To ensure that all important anatomic structures are included in the CT angiogram, the study begins with a frontal topogram of the chest and upper abdomen. Nonenhanced axial scans are not routinely acquired. However, in some clinical scenarios, nonenhanced scans should be acquired. Specifically, when a bicuspid aortic valve is a diagnostic consideration, nonenhanced views can be valuable for identifying valvular calcification. Nonenhanced scans are also useful in the evaluation of surgical shunts to assess for contrast leakage and calcification.

The goal in optimizing contrast delivery is to achieve the highest contrast enhancement with the least amount of contrast material. However, no single approach to contrast administration is effective in all patients, as there are several factors that influence the degree of contrast enhancement. These factors can be divided into three major groups: the patient, the contrast injection, and the scan factors.

Patient factors that influence contrast enhancement are the patient’s body size, cardiac output, renal function, and route of vascular access. Factors related to the contrast injection include the volume and concentration of the contrast material, the type of injection technique (power injector or manual injection), and the flow rate. Factors related to scanning are the delay between the start of the contrast injection and the initiation of scanning, the method of triggering the examination (empiric, bolus tracking, or a test injection), and whether scanning should be performed during multiple phases of contrast enhancement.

Optimal vascular enhancement for thoracic CT angiography requires a fast injection of contrast medium via a power injector at a high flow rate of 3 to 4 mL per second. A contrast volume of 100 to 150 mL with concentrations of 280 to 320 mg I/mL usually results in an excellent degree of vascular opacification.

The contrast material is given through a 20- or 22-gauge antecubital vein. Because CT angiography requires high flow rates and large volumes of contrast, the iodinated contrast medium should be nonionic. The use of nonionic agents minimizes gastrointestinal side effects (nausea and vomiting), discomfort at the site of injection, patient motion during intravenous contrast administration, and complications arising from contrast extravasation (7,8).

The delay time between the start of the contrast injection and the initiation of data acquisition is one of the most important factors in optimizing contrast delivery in CT. To time image acquisition, we use real time bolus tracking (Fig. 15-2). This method monitors the attenuation value of a target vessel during the contrast injection and displays the attenuation graphically in a real time fashion. Once the desired attenuation value, usually 120 HU, is reached, the scan sequence is automatically triggered. An alternative method of initiating the CT angiogram is the use of an empiric delay of 25 to 30 seconds after the start of the injection. Empiric timing is used mainly as a default method if the bolus tracking fails to trigger. A third method to time image acquisition is the use of a 15- to 20-mL test bolus of contrast material administered at a rate of 3 to 4 mL
per second. After an 8- to 10-second delay, images are acquired every 2 seconds for a total of 30 seconds. A time-attenuation curve is generated, and the image with the greatest contrast density is used to select the time delay. Although a test bolus is effective, it is more time consuming than the alternative methods and has the added disadvantage of an increased radiation exposure and a slight increase in background attenuation value because of the test-bolus injection. In our experience, bolus tracking has proved most reliable for timing scan initiation.

Thoracic CT angiography should be performed with a 0.5 to 1.5 collimation thickness and a fast table speed (24 to 36 mm) per rotation. The gantry rotation time should be the lowest possible to minimize the duration of the CT acquisition. CT angiography should be performed with a single breath hold.

Reconstruction thicknesses for axial viewing are 1 to 5 mm, depending on the clinical scenario. For example, if arch positioning is of clinical interest, then images at 5-mm thickness usually suffice for diagnosis. If thrombus within a surgical systemic-to-arterial shunt or the presence of a small anomalous vessel is of clinical concern, then 1-mm reconstructions may be more appropriate.

The reconstruction field of view should be minimized to include only the relevant anatomic structures. A standard reconstruction algorithm is used to reconstruct the enhanced data set. This algorithm minimizes noise in the data set and therefore can optimize the creation of 3D images. Images should also be viewed with a high-resolution
algorithm if associated abnormalities of the airways are suspected.

Axial images are usually sufficient for diagnosis, but multiplanar or 3D reconstructions, particularly when viewed in a cine mode, provide better anatomic detail about anatomic relationships between the great vessels and adjacent soft tissue structures and the tracheobronchial tree. There are a number of 3D reconstruction techniques available to reconstruct the volume data (9,10,11,12,13). A detailed description of the various rendering techniques is beyond the scope of this chapter and has been described elsewhere in this book. However, a brief review of the various reconstruction techniques is presented below.

Pediatric patients have several inherent problems that are not present in adults, in particular, patient motion, small body size, lack of perivisceral fat, and increased sensitivity to radiation exposure. These problems can be minimized or eliminated by the appropriate use of sedation and intravenous contrast medium, and by the selection of optimal technical factors for performing the CT examination (14,15,16,17).

There are several considerations that need to be addressed for pediatric patients prior to start of an MR scan. Choice of imaging coils can have significant impact on image quality. Because small children and infants require very small field-of-view to obtain adequate spatial resolution, local or phased array coils are required for the study. Infants can often be scanned within the head coil. This provides sufficient space for the child while guaranteeing satisfactory signal-to-noise ratios (SNR) for the examination. Larger children are usually imaged within the torso or cardiac coils. Use of the body coil, especially in children, should be avoided because of lower SNRs.

Sedation is still required for the majority of MR studies of infants and children 5 years of age and younger to prevent motion artifacts during scanning. Older children can usually be scanned successfully as long as there is continued verbal reassurance and explanation of the procedure. Some developmentally delayed older children may still require sedation for MR studies, and they need to be evaluated on an individual basis. Children undergoing conscious sedation should have no liquids by mouth for 3 to 6 hours and no solid foods for 6 hours prior to their study. Conscious sedation requires the presence of health care providers trained in maintaining adequate cardiorespiratory support during and after the examination. Patients’ vital signs are continuously monitored and recorded throughout the procedure and through the recovery period.

CEMRA is a fast, accurate, and flexible method for noninvasive imaging of the arterial and venous system. In the 10 years since its development, the method has been continually refined (28,29). Because of its great clinical utility, it migrated from academic centers into the community within a short period of time and is now the workhorse of non-neurological vascular MR imaging. In patients with complex anomalies involving the thoracic vessels, CEMRA can be useful for simultaneous evaluation of both arterial and venous anomalies as well as complications of surgical intervention (Fig. 15-3).

CEMRA is performed by acquiring a 3D gradient echo sequence following a rapid bolus of gadolinium. It is both reliable and relatively easy to perform given the automation and speed of current MR systems. Unlike iodinated contrast agents, gadolinium chelates can be used safely, even at high doses, in patients with renal failure (5). CEMRA
depicts the relevant arteries and, when desired, venous structures in a 10- to 30-second acquisition, depending on the strength and other technical factors of the MR system used for the study. This can be performed during a breath hold if the patient is nonsedated and able to comply with breath-holding instructions.

For most thoracic cardiovascular applications, a 3D spoiled gradient echo (SPGR) volume that includes the heart, pulmonary arteries, thoracic aorta, and proximal great vessels is used. The parameters for the 3D SPGR sequence are optimized to attain the highest-quality images. In general, faster is better for data acquisition with 3D CEMRA, and low temporal resolution (TR) and echo time (TE) values are usually selected. Too much spin dephasing can cause signal loss at stenoses, although this can be reduced or eliminated by selection of a TE less than 3 msec. Faster data acquisitions allow the gadolinium contrast material to be injected with a higher injection rate, producing a higher arterial gadolinium concentration and optimized enhancement of the relevant anatomy. The high arterial signal to noise (S/N) may then compensate for reduced T1 relaxation and signal averaging.

Fast data acquisition minimizes motion artifacts and makes it easier for patients to successfully suspend breathing for the entire data acquisition window. In addition to
minimizing TR and TE, one needs to select the smallest number of sections sufficient to cover the arterial anatomy and to keep the acquisition time to a minimum. Widening the bandwidth also makes the acquisition faster, but it may also reduce S/N, especially if a very small field-of-view is used, as required for small children and infants. The signal of background tissue is also reduced or eliminated by obtaining a 3D data set “mask” before gadolinium administration to use for digital subtraction (30).

In young children who are either sedated or are unable to perform a successful breath hold, we have altered our CEMRA technique to help minimize artifacts associated with respiration. Rather than reduce acquisition time by using a partial Fourier technique, as done with older children and adults, we have effectively lengthened the acquisition time by sampling all of the k-space. This allows for effective signal averaging, reduces motion related artifacts, and increases the SNR. Because children have rapid circulation times, it is difficult or impossible to obtain a pure arterial phase acquisition with this approach. The veins are often enhanced to the same degree as the arteries. To generate only arterial-phase images, the veins can be removed by postprocessing techniques. Often, especially in cases of complex congenital heart disease, it is very useful to have the veins and arteries enhanced simultaneously because it helps depict the anatomical relationship between the structures.

The flip angle of the 3D sequence is optimized for T1 contrast on the basis of the repetition time and expected blood gadolinium concentration. In practice, a flip angle of 30 to 45 degrees works well in nearly all cases. The flip angle may need to be larger for higher doses of contrast material and longer repetition time or smaller for lower doses of contrast material and shorter repetition time. Zero filling in the section direction is useful because it doubles the number of sections, which is useful for image reconstruction, without increasing imaging time.

The imaging volume is prescribed to depict all of the relevant anatomy while minimizing the actual acquisition time. In almost all cases, a coronal imaging volume is used, the exception being larger children or adults in whom the thoracic aorta is the only region of interest. In the latter scenario, a sagittal imaging volume is used.

In larger patients, correct timing coordination of the bolus injection with peak vascular enhancement during acquisition of central k-space data is essential for good quality studies. There are several ways to time contrast delivery. Each has its own advantages—the best guess method is simplest, the automated detection method is the most operator independent (31), and the timing run method is the most reliable (32). Fluoroscopic triggering, although not yet widely available, is reliable and fast for achieving optimal enhancement in every case (33,34,35). As MR systems become faster, 4D MRA will likely become the dominant method. This will acquire multiple complete 3D acquisitions, and the optimally enhanced acquisition will be chosen retrospectively. Although some current scanners have this capability, the resolution is limited compared with other methods.

In infants and small children, it is difficult to achieve a well-timed examination that results in a pure arterial- or venous-phase acquisition. Because the circulation time is so rapid, both the arteries and veins enhance within a short period of time following contrast infusion. Reducing the acquisition time to a point where a timed acquisition can be effective may sacrifice image resolution and reduce SNR. We have found that simultaneous initiation of contrast infusion and data acquisition result in diagnostic CEMRA studies that have adequate resolution and SNRs, although both venous and arterial structures are enhanced.

The dose of gadolinium is an important determinant of image quality, and widely ranging doses have been advocated. Three-dimensional CEMRA studies have used doses of gadolinium ranging from 0.5 mmol/kg (“half dose”) to 0.3 mmol/kg (“triple dose”). For adult patients, we routinely use 30 mL of gadolinium per patient. For most adults in the United States, this dose is equivalent to approximately 0.15 to 0.25 mmol/kg. High-dose bolus injection of gadolinium for MR angiography is still not approved by the U.S. Food and Drug Administration. However, given the preponderance of studies supporting the safety and efficacy of these doses in the literature, we feel that this dose is medically justifiable. We do not exceed a total weight-based dose of 0.3 mmol/kg for any study.

In children and infants, careful attention must be given to the volume of infused gadolinium chelate in order to not exceed the clinically accepted upper limit of 0.3 mmol/kg. In these patients, unlike adults, we use a weight-based dosing of 0.2 mmol/kg for CEMRA studies. In some small infants, only 1 or 2 mL of gadolinium may be needed. This volume is so small that it can often be preloaded into the intravenous line tubing prior to infusion, simplifying the injection process. After the gadolinium has been injected, a saline flush of approximately twice the gadolinium volume is infused to clear the line and to ensure that the patient receives the entire gadolinium dose.

Contrast may be administered either by hand or by a MR compatible power injector. For larger children or adults, a 20- or 22-gauge angiocatheter is routinely inserted in an antecubital vein prior to the start of the study. In smaller children or infants, a smaller angiocatheter can be used if needed, and this can be positioned elsewhere as long as it can handle a rapid bolus of 1 to 2 mL per second. A long extension tube is used to connect to hand-held syringes or a power injector located outside of the MR scanning room. Since CEMRA is a first-pass technique, the contrast must be infused while the patient is in the magnet, thus the need for the long extension tubing. The power injector ensures a reliable rate of contrast delivery and eliminates the need for a second technologist during the study. Contrast is usually infused as a bolus at a rate of 1 to 2 mL per second and is followed immediately by a saline flush.

As discussed in the CT section above, accurate interpretation of CEMRA requires interactive manipulation of the 3D data sets. It is not possible to rely solely on the source images because they are subject to partial volume effects. The reconstructed images greatly enhance diagnostic confidence. In the past, the most widely used postprocessing technique for CEMRA was maximum-intensity projection. The diagnostic accuracy of contrast-enhanced MR angiography using the MIP technique is well described and accepted clinically (36,37,38,39,40). In the thorax, subvolume MIPs, made from only a selected portion of the original data set, are useful to exclude nonrelevant anatomy.

Multiplanar reconstructions are useful in the evaluation of thoracic vascular anomalies. Because both MIPs and volume-rendered images are projectional images, adjacent structures may overlap and may obscure the relevant anatomy. MPRs are generated using any angle through the original data set. Orientations that depict the blood vessel(s) of interest in an optimal manner are selected interactively. This is a very practical method to accurately determine the size of a vessel or its angulation and relationship to other structures.

Volume rendering is another useful method for depicting thoracic vessels. Recent studies have shown that the VR technique has significant advantages for CEMRA (39,40). Mallouhi et al. (39) found that VR performed slightly better than MIP for quantification of renal stenoses greater than 50% and significantly better for severe stenoses. VR also had a substantial improvement in positive predictive value, and renal vascular delineation on VR images was significantly better. In another series, MIPs were statistically less reliable for determining renal artery stenoses compared with digital subtraction angiography (40).

The MIP algorithm selects only the voxel with the highest attenuation along a ray projected through the data set; volume-averaged voxels may be erroneously excluded from the final image, resulting in overestimation of stenosis. VR is based on the percentage classification technique, which is used to estimate the probability of a material being homogeneously present in a voxel (39). This provides an accurate determination of the amounts of materials when the voxel consists of two or more materials, which are volume averaged. With VR, the volume-averaged voxels are included in the final image because VR calculates a weighted sum of data from all voxels along a ray projected through the data set.

Spin echo (SE) was the first sequence used for evaluating cardiac and thoracic vascular morphology. The development of ECG-gating made SE technique especially useful by substantially reducing motion artifacts (41,42,43). Spin echo sequences generally provide good contrast between the vessel wall and blood. These are called black blood images because of the signal void created by flowing blood. Blood signal may appear brighter in slowly flowing areas, such as areas immediately adjacent to the vessel wall. Presaturation with radio frequency and reduction of the echo time minimizes blood signal and increases contrast on gated SE images (43). Although widely available, SE imaging has limited temporal resolution and is degraded by respiratory and other motion-related artifacts.

Shorter acquisition times are achieved with fast (or turbo) spin echo (FSE) pulse sequences, also known as rapid-acquisition relaxation enhancement (RARE) (42). Although faster, soft tissue contrast can be less than that with SE techniques because of the wide range of acquired TEs inherent in FSE technique (43). Numerous modifications to the basic FSE sequence have been made, including the use of one or more inversion pulses, increased echo train length, half-Fourier reconstruction, and echo planar techniques.

Single-shot FSE (SSFSE) sequences use a very long echo train in tandem with half-Fourier reconstruction (44). The center of k-space is acquired in a short time, minimizing motion blurring. The rapid acquisition of multiple phase lines for a single TR allows for coverage of the entire heart and thorax in the time frame of one or two breath holds. SSFSE technique has the advantage of being faster than the FSE technique in the evaluation of thoracic aortic disease (45,46). The SSFSE sequence can be modified for better cardiac results by reducing the echo train length, lowering the effective TE, and using a blood-suppressed preparation method (47,48).

T2-weighted inversion recovery imaging is now used as the front-line sequence for depiction of cardiac and thoracic vascular morphology. This technique uses a selective and a nonselective 180-degree inversion pulse followed by a long inversion time to null blood magnetization (49,50). A second selective 180-degree inversion pulse can also be applied to null fat. This is referred to as double (or triple) inversion recovery (DIR and TIR, respectively). The sequence is acquired either with breath hold or a nonbreath-hold technique and provides for excellent delineation of vessel wall or myocardial blood interfaces. It effectively nulls blood and depicts blood vessel interfaces so well that the sequence has even been useful for performing coronary angiography (50).

Bright blood imaging yields both morphologic and functional data. In this sequence, blood has a bright signal intensity. Multiple consecutive images are acquired and can be viewed dynamically to depict cardiac motion. Sequences include gradient echo (GRE), fast GRE, segmented k-space fast GRE, and steady state free precession (SSFP) (true FISP, FIESTA) techniques.

Gradient echo imaging is well suited for cardiac and vascular imaging because of its short echo and repetition times. Blood appears bright compared with adjacent myocardium due to time-of-flight effects as well as the relatively long T2. Markedly turbulent blood loses signal due to intravoxal dephasing, a helpful artifact for assessing areas of stenosis or valvular regurgitation (51).

A segmented k-space approach provides high-resolution dynamic images and can be performed much more rapidly than other techniques (52,53,54). Using short echo times (2 msec) and short TRs (<10 msec), multiple lines (segments) of k-space are acquired during each cardiac cycle. In contradistinction, in the GRE techniques, only a single line of k-space is acquired per cycle. Segmented k-space fast GRE imaging remains a mainstay for dynamic cardiac and vascular imaging and has been improved and adapted for even faster acquisition times (55,56,57). However, the technique is limited by the need to maintain adequate enhancement of inflowing blood. At lower TRs, now available with high-performance gradient systems, inflow enhancement of the cardiac blood pool diminishes and saturation occurs, reducing vessel wall–blood contrast. The inability to further reduce TR effectively limits achievable spatial and temporal resolution.

Steady state free precession is a state-of-the-art approach to the improvement of cine imaging (Fig. 15-4). Image contrast in SSFP depends on the T1/T2 ratio of tissue and, unlike GRE techniques, is less dependent on flow. SSFP uses the available blood signal very efficiently and accurately depicts blood, myocardium, and epicardial fat (58). First described in 1986, it was not used for cardiac or vascular applications for some years (59,60). SSFP is susceptible to magnetic field inhomogeneities and requires very short TRs, limiting its use until recently. With technical improvements in magnetic field homogeneity and the development of higher performance gradient systems, diagnostic SSFP images can now be obtained with limited artifacts (61,62,63,64). This technique is also known as BFFE (balanced fast field echo), FIESTA (fast imaging employing steady state acquisition), FISP (fast imaging with steady precession), and trueFISP.

In the heart, SSFP sequences result in improved contrast between myocardium and ventricular cavities with clearer delineation of trabeculation and papillary muscles as compared with segmented k-space fast GRE techniques. The signal-to-noise and contrast-to-noise (C/N) ratios of SSFP are substantially higher than conventional techniques (65,66,67). Barkhausen et al. found that the mean C/N ratio improved by an average of 46% and 100% in short- and long-axis images, respectively, compared with the standard cine gradient echo sequences (67). The reported C/N ratios are also higher than those obtained with contrast-enhanced gradient echo techniques (68). Pereles et al. found that SSFP depicts morphologic and functional abnormalities with greater precision and provides greater diagnostic confidence than conventional techniques (66). The other advantage of SSFP is improved temporal resolution (67,69). Reduction in acquisition time by a factor of two to three at similar spatial resolutions is possible. Shorter acquisition times also can be exploited to improve spatial resolution.

Phase contrast (PC) techniques allow for depiction of blood flow and quantitative analysis of the velocity and volume of flow through a vessel (Fig. 15-5). Image data regarding the phase and magnitude of spin vectors are routinely acquired in MR imaging, but the phase data are usually discarded. PC MRI allows phase information to be saved and displayed. Pixels corresponding to spins moving across a magnetic gradient, as occurs with active blood flow, are assigned bright or dark signal intensity, while stationary spins are assigned an intermediate (grey) value.

This can be used to determine patency of vessels and can be quantitatively evaluated to determine a profile of spin velocities across a vessel lumen. Spins moving across a magnetic gradient accumulate a phase shift proportional to their velocity. Most recent MR systems can now accurately calculate the velocity profile by analyzing the phase shift in a pixel using an equation that uses the gyromagnetic ration, time interval between gradient lobes, and the area of the bipolar gradient pulse (70).

Velocity-encoded (VEC) PC techniques are used to determine flow, peak flow rates, and the volume of flow per unit time. Once the velocity profile is acquired, the area of the vessel lumen of interest is measured, and the volume of blood moving through a vessel is quantitatively determined. ECG-gated velocity-encoded PC is used to calculate cardiac output. A region of interest placed on the ascending aorta can be used to accurately calculate the volume of blood flow per minute. A similar region of interest placed on the pulmonary artery can be used to calculate pulmonary blood flow, which generally is equal to that determined in the aorta. If there is an intra- or extracardiac shunt present, the ratio of pulmonary to systemic flow can be calculated. Peak flow rates across an area of stenosis can also be used to determine the pressure gradient. Using a modified Bernoulli equation, the pressure gradient, in millimeters of mercury (mmHg) across a stenosis is calculated as the product of the cube of the peak velocity as measured in meters per second.

VEC PC imaging is a valuable technique for quantitative assessment of flow dynamics in congenital heart disease. Clinical applications include the measurement of collateral flow and pressure gradients in coarctation of the aorta (Fig. 15-6), differentiation of blood flow in the left and right pulmonary arteries, quantification of shunts, and evaluation of valvular regurgitation and stenosis. After surgical conduit placement, VEC PC is used to monitor blood flow, restenosis, and flow dynamics. With VEC PC imaging, there are some potential pitfalls such as potential underestimation of velocity and flow, aliasing, inadequate depiction of very small vessels, and possible errors in pressure gradient measurements (71). Nevertheless, VEC MR imaging is a valuable tool for preoperative planning and postoperative monitoring in patients with coronary heart disease (CHD).

A new method of spatial encoding referred to as parallel imaging is helpful for MR of the thoracic vessels (72,73). This type of imaging uses arrays of radio frequency detector coils to acquire multiple data points simultaneously rather than sequentially, as has been traditionally performed. This results in faster imaging speeds that can be used to increase either temporal or spatial resolution.

An R factor (parallel factor) is used to describe the order of factor used. For a given R factor, data acquisition time is reduced to approximately 1/R if the imaging matrix is unchanged. A factor of two is currently the most widely available factor, although higher factors are now being utilized by some centers. At a factor of two, a typical acquisition time
will be reduced by 50%. Alternatively, image resolution could be doubled, by acquiring more phase lines or thinner partitions, without changing the study acquisition time.

In the thorax, the scan times of many sequences can be reduced by a factor of two or four given the current technology, and larger reductions in scan time will likely be possible in the future. The increased speed associated with parallel imaging does not come without a price. The SNR of parallel imaging is always reduced compared with the SNR ratio of other sequences obtained using the same coil array. The SNR of accelerated studies is lower by approximately 20% at a parallel factor of two and is further reduced by the square root of the parallel factor at greater levels.

The normal left aortic arch and its branches develop from paired pharyngeal arches, which are bilateral bars of mesoderm, separated by ectodermal clefts. The pharyngeal arches form during the fourth and fifth weeks of development and undergo a process of orderly regression. Each arch has its own cranial nerve as well as its own artery. These arteries are known as aortic arches and arise from the aortic sac, which is the most distal part of the truncus arteriosus. There are five aortic arches in the human embryo, numbered craniocaudally as I, II, III, IV, and VI. Arch V either never forms or forms incompletely and then regresses. The aortic arches course dorsally from the aortic sac to terminate in the right and left dorsal aortas. The ventral portion of the
aortic sac develops into the truncus arteriosus, which is divided by the aorticopulmonary septum into the proximal aortic root and the main pulmonary artery.

By the sixth week of intrauterine life, most of the first and second aortic arches have atrophied, except for small portions that persist to form the maxillary artery and stapedial arteries, respectively. The third, fourth, and sixth arches remain and develop into the great arteries (Figs. 15-7, 15-8).

The third aortic arch forms the common carotid arteries and the first part of the internal carotid arteries. The fourth aortic arches form asymmetric structures. The right fourth arch forms the most proximal segment of the right subclavian artery. The distal part of the subclavian artery forms from a portion of the right dorsal aorta and the seventh intersegmental artery. The left fourth arch forms the portion of the aortic arch between the left common carotid and the left subclavian arteries. The sixth aortic arches also develop asymmetrically. The right arch forms the right pulmonary artery, and the left arch forms the left pulmonary artery and the ductus arteriosus.

In the fifth week of intrauterine life, the right dorsal aorta disappears between the origin of the seventh intersegmental artery and the junction with the left dorsal aorta. The two aortas fuse caudally below T4 to form a single aorta in the distal thorax and abdomen.

Regression of a segment of arch that should have persisted or persistence of a segment of arch that should have regressed explains the development of most arch anomalies (74,75,76). These anomalies include left aortic arch with aberrant left subclavian artery, double aortic arch, right aortic arch with mirror image branching, right aortic arch with aberrant left subclavian artery, and cervical aortic arch. Aortic arch anomalies are of clinical importance when they produce symptoms.

The three major venous systems of the thorax—cardinal, subcardinal, and supracardinal veins—develop during the early embryonic period (Figs. 15-9, 15-10). The cardinal veins form the main venous drainage system of the embryo. This system consists of the anterior cardinal veins, which drain the cephalic part of the embryo, and the posterior cardinal veins, which drain the remaining part of the fetal body. The intersegmental veins drain into the anterior and posterior cardinal veins. Through anastomoses, the cardinal venous system becomes the superior vena cava and brachiocephalic, subclavian, internal jugular, and superior intercostal veins.

The subcardinal system is a paired longitudinal venous plexus located in the developing fetal abdomen. This network mainly drains the kidneys and gonads in the urogenital ridge. Numerous ventrodorsal anastomoses unite the subcardinal with the supracardinal venous system.

The supracardinal veins lie lateral to the dorsal aorta and drain the body wall via the intercostal veins. By the sixth intrauterine week, they take over the functions of the posterior cardinal veins. The supracardinal veins are the origin of the adult azygos venous system. The fourth to 11th right intercostal veins empty into the right supracardinal vein, which, together with a portion of the posterior cardinal vein, forms the azygos vein. On the left side, the fourth to seventh intercostal veins enter into the left supracardinal vein, and
the left supracardinal vein empties into the azygos vein and is then known as the hemiazygous vein.

Variations in venous anatomy are the result of regression of a channel that should have persisted or persistence of a channel that should have regressed. The common deviations from the normal developmental pattern are the left superior vena cava, which is caused by persistence of the left anterior cardinal vein and obliteration of the common cardinal and proximal part of the right anterior cardinal vein, and the absence of the inferior vena cava, which occurs when the right subcardinal vein fails to establish a connection with the liver and shunts its blood directly into the right supracardinal vein (75).

In utero, blood from the placenta, which is 80% to 85% saturated with oxygen, returns to the fetus via the umbilical vein. Most of this blood flows into the ductus venosus and then into the inferior vena cava. In the inferior vena cava, the saturated placental blood mixes with deoxygenated blood (30% saturated) returning from the lower limbs. The mixed blood (70% saturated) from the inferior vena cava enters the right atrium, where it passes through the oval foramen into the left atrium. In the left atrium, it mixes with the pulmonary venous flow.

From the left atrium, the blood enters the left ventricle. Blood from the left ventricle (65% saturated) enters the aortic arch and supplies the head and neck. The desaturated venous blood from the head is drained by the superior vena cava. From the superior vena cava, it flows by way of the right ventricle into the pulmonary trunk. Since pulmonary vascular resistance is high during fetal life, the main portion of this blood passes directly through the ductus arteriosus into the descending aorta, where it mixes with blood from the proximal aorta, before being distributed to the trunk, lower extremities, and placenta.

At birth, there is the cessation of placental blood flow and the beginning of respiration, which results in closure of the ductus arteriosus by contraction of its muscular wall. As a result, blood in the ductus arteriosus no longer flows into the aortic arch. However, a small left-to-right shunt is not unusual during the first few days after birth. Complete anatomical closure by intimal proliferation occurs after 1 to 3 months, resulting in the ligamentum arteriosum (74,75). The return of oxygenated blood from the lungs also increases pressure in the left atrium, which, combined with a decrease in pressure in the right atrium, closes the foramen ovale. Fibrous union occurs in several months, although in approximately 30% of individuals anatomical closure may never be obtained, resulting in a patent foramen ovale.