Dr. Panigrahy is Chief Resident in the Department of Radiological Sciences at University of California, Los Angeles (UCLA). He received his medical degree from Boston University in 1998. His graduate training included a Howard Hughes Research Fellowship, 1994­1996, at Boston Children's Hospital. He is planning fellowship training in Pediatric Radiology and both Adult and Pediatric Neuroradiology.

Drs. Ratib, Boechat, and Gomes are senior faculty with expertise in MR imaging of pediatric congenital heart disease in the Department of Radiological Sciences at UCLA.

Two of the most exciting advances in magnetic resonance imaging (MRI) of pediatric congenital heart disease have been the development of fast cine pulse sequences and the postprocessing of gadolinium chelate (Gd)-enhanced MR angiographic data. Fast cine pulse sequences can provide anatomic and physiologic information to aid in the diagnosis of simple and complex congenital heart lesions. Three-dimensional volume rendering of Gd-enhanced MR angiographic data provides new and essential information about both stenoses and the anatomic relationships of blood vessels. This article provides an overview of these technical advances in the use of MRI in pediatric congenital heart disease.

Over the last decade there have been significant advances in the role of magnetic resonance imaging (MRI) in pediatric congenital heart disease. Advantages of MRI over echocardiography and cineangiography--heretofore the standard techniques for evaluation--include: 1) the ability to obtain both high-resolution anatomic images in any orientation and dynamic (cine) images of the heart to evaluate its mechanical and functional performance; 2) the fact that the information represents a continuous three-dimensional (3D) set obtained or reconstructed in any imaging plane; and 3) superior evaluation of cardiac morphology, anatomy, and topography of large thoracic blood vessels, including the aorta, central pulmonary arteries, and both systemic and pulmonary veins.

Recent advances in MRI include the development of fast-pulse sequences, that can provide anatomic and functional information, and advanced postprocessing of MR gadolinium chelate (Gd)-enhanced angiographic data. The radiologist must be familiar with the MRI techniques applicable to the evaluation of pediatric congenital heart disease lesions in order to properly tailor the examination to the diagnostic problem. This paper will review the technical aspects of MRI in respect to the more common types of pediatric congenital heart disease, emphasizing recent technical advances.

Getting started and patient preparation

Given the vast number of imaging techniques available, the radiologist must choose the appropriate set of MRI protocols to evaluate the specific clinical problem. If the patient has undergone cardiac catheterization or echocardiography, this information should be reviewed before determining the protocols for the examination. It is also important that the radiologist be familiar with the patient's clinical history because certain abnormalities such as arrhythmias may affect the examination in terms of proper electrocardiographic (ECG) gating. The radiologist should discuss with the referring physician the type of sedation (general anesthetic sedation, deep sedation, or conscious sedation) needed for the study. Proper positioning of ECG leads is critical for obtaining adequate ECG gating, which may be required to overcome the artifact associated with cardiac contractions.

The MRI protocol for pediatric congenital heart disease is made up of anatomic and physiologic components. ECG-gated spin-echo imaging and MR angiography is needed for anatomic information, while functional information may be obtained from dynamic imaging protocols including cine imaging and velocity tagging.

Traditional pulse sequences

The traditional pulse sequences for the MRI of pediatric congenital heart disease are spin-echo and gradient-echo imaging sequences (figure 1). ^1,2 For the spin-echo pulse sequence, the repetition time is set by the R-R interval. For T1-wieghted images, the blood exposed to the 90š pulse has moved out of the imaging section prior to receiving the 180š pulse, which leads to the blood being "black." However, if blood is exposed to both the 90š and 180š pulse, as in diastole, the blood will be of higher signal intensity, which should not be mistaken for a thrombus.

A typical cardiac image consists of a 256 * 128 matrix and a section thickness of 4 to 6 mm. The field of view will depend on the size of the patient. The planes frequently used in spin-echo imaging are the axial, coronal, and sagittal oblique views. The axial plane is ideal for demonstrating ventricular and atrial septal defects. Other planes may be tailored to the disease. For a patient with a double aortic arch, the coronal view may be useful in comparing the relationship between the right and left arches.

Gradient-echo imaging employs a sequence of multiple images at different phases of the cardiac cycle for a given imaging plane. ³ The individual images are viewed sequentially in a cine loop that gives a visual effect of the beating heart. In the gradient-echo sequence, flowing blood appears white, rather than black as it does when using the spin-echo sequence. With gradient-echo sequence, turbulent blood flow resulting from valvular or vascular stenosis leads to a dark signal void.

Gradient-echo imaging usually requires less time than spin-echo imaging. Typically the matrix is 250 * 128 with only two signals acquired. In order to improve image quality for non­breath-hold acquisition, 16 to 32 phases of the cardiac cycle are obtained and played in a cine loop. The number of phases of the cardiac cycle obtained depends on the heart rate and the memory capability of the MRI system. Other variations of cine imaging include velocity-encoded cine MRI, myocardial tagging, and fast imaging with steady-state precession (FISP) technique for acquisition of high temporal and high spatial resolution images.

Velocity-encoded or phase-mapping cine MRI is a modified gradient-echo technique that not only determines the direction of blood flow, but also quantifies blood velocity and blood flow. ⁴ For myocardial tagging cine imaging, saturation bands are placed across the myocardium at the end of diastole. ⁵ The tags are attached to tissue throughout the cardiac cycle. Tagging techniques are used to quantitatively analyze regional myocardial wall motion, especially for ischemic heart disease and cardiomyopathy. In pediatric congenital heart disease, tagging techniques have been used to evaluate right ventricular dysfunction of patients with right ventricular outflow obstruction, and to study the left ventricle of patients with transposition of the great arteries. ⁵

Segmented k-space gradient-echo imaging techniques

Breath-hold cardiac imaging makes use of segmented k-space techniques and requires stronger gradient systems with faster switching capabilities. With conventional spin-echo and cine imaging, a single line of k space is acquired per R-R interval. Segmented k-space techniques acquire multiple lines (eg, segments) of k space per R-R interval, which decreases the imaging time (figure 2). If the number of k space lines acquired per R-R interval is high enough, the acquisition time can be reduced to a single breath hold. ⁶ The segmented-k techniques use a very short repetition time, approximately 4 to 10 msec. Lower repetition times require smaller flip angles, which result in a lower signal-to-noise ratio compared with conventional cine studies. The different types of segmented k-space gradient-echo techniques include FAST-CARD, turbo-FLASH (fast low angle shot), fast GRASS (gradient recalled acquisition in the steady state) and true-FISP (figure 3). Pereles et al ⁷ showed that true-FISP imaging depicts morphologic and functional abnormalities of congenital heart disease with great clarity and more rapidly than FLASH images, allowing for greater diagnostic confidence. The disadvantage of both the true-FISP and the FLASH technique is that they are extremely susceptible to metal, due to their sensitivity to magnetic field inhomogeneities, leading to artifact (figure 4). A HASTE sequence is less susceptible to metal, but cannot provide cine information (figure 4).

MR angiography

The three major types of MR angiography include time-of-flight (TOF), phase contrast, and contrast-enhanced techniques. ^8,9 Time-of-flight angiography can differentiate blood from stationary tissue by means of flow-related enhancement. This technique combines gradient-echo imaging and flow compensation. By using gradient-echo imaging, the blood flow into the plane of the section is fully magnetized, thus appearing bright in relation to the surrounding stationary tissue. Images of the anatomic vascular structures are created by combining contiguous sections in two-dimensional (2D) or 3D TOF sequences. A maximum intensity projection (MIP) algorithm is commonly used in the creation of TOF angiography images.

The second type of MR angiography is phase-contrast MR angiography that measures the phase shift of blood relative to flow-related enhancement in TOF angiography. This technique is similar to velocity-encoding cine imaging except that multiple contiguous images are acquired, as opposed to multiple images of the same section. The advantage of phase-contrast MR angiography is that it permits calculation of blood velocity through determination of the blood proton shift over a period of time.

The third type, contrast-enhanced MR angiography, is the method of choice at our institution for the evaluation of pediatric congenital heart disease. Because Gd-based contrast agents are paramagnetic, they shorten the spin-lattice relaxation time of blood, giving it a high signal intensity on T1-weighted images compared with the surrounding tissue. Unlike TOF and phase-contrast MR angiography, contrast-enhanced MR angiography is not dependent on the inflow of saturated blood. It allows the repetition time to be shortened without affecting signal-to-noise ratio. Contrast-enhanced MR angiography is not cardiac gated, requires a shorter time of acquisition, and allows 3D images to be obtained during a breath hold. There are two possible strategies for performing 3D contrast-enhanced MR angiography: the breath-hold strategy and the non­breath-hold strategy. Contrast-enhanced MR angiography data may be postprocessed in multiple ways.

Four types of postprocessing methods are available, including multiplanar reconstruction (MPR), shaded-surface rendering, maximal intensity projection (MIP), and volume rendering (figure 3). In a recent study, conducted by the authors of this paper, volume rendering was found to provide additional information in 66% of cases in which it was used, and 3D volume rendering provided improved visualization of overlapping structures, which are not separated by conventional MIP techniques. ¹⁰ Three-dimensional volume rendering was useful in the evaluation of vessel stenoses and for demonstrating anatomic relationships in complex congenital heart disease. We also found that 2D multiplanar reformatting allowed improved accuracy in the measurement of vessel diameter, particularly of tortuous vessels because of the ability to properly align the measurement along the long axis of the vessel.

Other researchers have found that surface rendering does not add more information in cases of pediatric congenital heart disease. ¹¹ More research is needed comparing the different types of postprocessing of Gd-enhanced MR angiographic data to cross-sectional imaging (spin-echo imaging) for measuring vessel diameter and stenoses.

Image interpretation

MR studies of pediatric congenital heart disease provide both morphological and functional information. Pediatric congenital heart disease, because of its complexity, requires a systematic, organized approach to achieve accurate interpretation. MR analysis of this complex disease requires a rigorous approach that can be separated into three parts: 1) localization of the three cardiac segments (atria, ventricle, great arteries); 2) the determination of the type of atrioventricular and ventriculoarterial connections; and 3) the detection of associated anomalies.

The indications for MRI in the evaluation of patients with congenital heart disease have been reported by the Task Force of the European Society of Cardiology. ^12,13 The major indications include: 1) segmental description of cardiac anomalies; 2) evaluation of thoracic aortic anomalies; 3) noninvasive detection and quantification of shunts, stenoses, and regurgitations; 4) evaluation of cono-truncal malformations and complex anomalies; 5) identification of pulmonary and systemic venous anomalies; and 6) postoperative studies. ^12,13

The remainder of this article will focus on the role of MR imaging in the diagnosis of some of the more common types of pediatric congenital heart lesions.

Atrial structures

Atrial septal defects (ASD) are the most common congenital cardiac defects. There are three main types: secundum (located in the middle of the atrial septum), sinus venosus (located superiorly at the junction of the superior vena cava and right atrium), and primum (in the inferior portion of the atrial septum, near the atrioventricular valves). Because of left-to-right shunting, right atrial and right ventricular volume overload is common. With MRI, the diagnosis of an ASD can be made by demonstrating the defect on two adjacent axial images or during multiple phases of the cardiac cycle during cine imaging at one anatomic level. ¹⁴ The sensitivity and specificity of MRI for the diagnosis of ASD are both > 90%. ¹⁵

Ventricular structures

MRI also provides valuable information about ventricular septal defects (VSD). The majority of VSDs occur in the perimembranous septum, an anatomic area that is well evaluated with spin-echo or gradient-echo MRI. MRI defines not only the extent of the defect, but its relationship with other ventricular structures as well, resulting in a more complete characterization.

Evaluation of the great arteries

Pulmonary artery

MRI is useful in the evaluation of the pulmonary arteries with respect to origin, presence, size, and confluence. Images in the axial plane are usually adequate to demonstrate the main, right, and left pulmonary arteries. If the left pulmonary artery cannot be evaluated in the standard axial view, an oblique view paralleling the long axis of the left pulmonary artery is useful.

Hypoplastic pulmonary arteries are found in the spectrum of tetralogy of Fallot, the most common type of cyanotic congenital heart disease. The main anatomic features of the tetralogy of Fallot are infundibular pulmonary hypoplasia and stenosis, a large malaligned VSD, an overriding aorta and secondary right ventricular hypertrophy (figure 5). Determination of the morphology of the pulmonary arteries in patients with right ventricular outflow lesions is crucial when either corrective or palliative surgery is being planned.

Cineangiography has been considered the gold standard in the imaging of the pulmonary arteries of patients with right ventricular outflow lesions. Strouse et al ¹⁶ prospectively compared MRI to cineangiography in the evaluation of the central pulmonary arteries of patients with right ventricular outflow obstructive lesions. This study demonstrated that MRI was equivalent to cineangiography in evaluating the size of central pulmonary arteries. In addition, MRI was able to show the appearance of central pulmonary arteries when cineangiography was unable to demonstrate them. ¹⁶

MRI and echocardiography are similar in their ability to demonstrate the abnormalities of the right ventricular outflow tract and main pulmonary artery. However, MRI is superior to echocardiography in showing abnormalities of the central right and left pulmonary arteries. In general, measurements of the size and status of the pulmonary arteries in MRI correlate well with those obtained at conventional angiography. ^17,18 MRI is also excellent for evaluating shunts between the systemic and pulmonary circulations, including Blalock-Taussig, Glenn, and Waterston anastomoses (figure 5), that are commonly used in patients with right ventricular outflow tract lesions.

Aorta

Coarctation of the aorta is defined as a congenital narrowing of the aorta. The most common site of localized coarctation is just distal to the left subclavian artery, near the level of the ductus arteriosus (figure 6). MRI is useful in diagnosis of the lesion, presurgical planning, and postsurgical follow-up. ¹⁹ Characteristics of the coarctation including its site, length, severity, relationship to the left subclavian artery, and the extent of collateral blood flow, should be evaluated preoperatively. The coarctation should be evaluated postoperatively to determine if restenosis or aneurismal dilatation has occurred. MRI has been shown to be comparable to angiography in detecting postoperative residual stenosis and aneurysms at the site of the coarctation. ²⁰ We have found that volume-rendering of Gd-enhanced MR angiographic data is useful in demonstrating the anatomic distribution of collateral flow in cases of coarctation (figure 6).

Vascular rings are also well evaluated by MRI. ¹⁹ The most common symptomatic vascular rings are the double aortic arch and the right aortic arch with an aberrant left subclavian artery. In patients with a double aortic arch, a right and left arch arise from a single ascending aorta, with the right arch usually higher and larger than the left arch. MR images in the axial and coronal planes are the most useful for demonstrating the vascular anatomy (figure 7).

Evaluation of systemic and pulmonary veins

Systemic venous return

MRI is useful in evaluating systemic venous anomalies associated with congenital heart disease and heterotaxic syndromes. Polysplenia and asplenia are heterotaxy syndromes in which there is bilateral left (polysplenia) or right (asplenia) sideness. Asplenia is associated with complex cyanotic disease, while polysplenia may be associated with less complex acyanotic heart disease (left-to-right shunts). Other types of systemic venous anomalies that can be diagnosed by MRI include persistence of the left superior vena cava, anomalous left innominate vein, left inferior vena cava, and duplication of the inferior vena cava.

Pulmonary venous return

In total anomalous pulmonary venous return, the pulmonary venous return is connected to venous structures such as the superior vena cava, right atrium, or portal vein. The three types are supracardiac, cardiac, and infracardiac, with the supracardiac type being the most common type. MRI has been shown to be superior to echocardiography and cineangiography for the diagnosis of anomalous pulmonary venous connections. ²¹ MRI has also been shown to be excellent in showing obstruction or stenosis of pulmonary venous structures after correction of total anomalous pulmonary venous connections. ²²

Conotruncal rotational abnormalities

A group of congenital heart lesions arise from abnormal rotation of the conotruncus. Donnelly et al ²³ demonstrated that axial MRI is an excellent method to show the anatomic relationship between the aorta and pulmonary artery at the level of the semilunar valves. These abnormalities include D-transposition of the great arteries, L-transposition of the great vessels, truncus arteriosus, double-outlet right ventricle, and situs inversus (figure 8). In normal development, the primitive truncus divides into the aorta and the pulmonary artery, which then rotate 150š counterclockwise, so that the pulmonary artery lies anterior and to the left of the aorta. The conotruncal abnormalities represent different degrees of this rotation. ²³

Patent ductus arteriosus

The ductus arteriosus is the result of the persistence of a normal embryologic structure, the sixth aortic arch. The ductus arteriosus extends from the origin of the left pulmonary artery to the descending aorta just beyond the origin of the left subclavian artery. Functional closure of the ductus arteriosus typically occurs within 24 hours of birth. In some patients, the mechanism for the closure of the ductus arteriosus is abnormal, resulting in a persistent communication between the aorta and pulmonary artery. MRI is a useful tool is demonstrating the presence of a patent ductus arteriosus, both in the setting of other cardiac lesions (figure 8D) and in isolation (figure 9).

Conclusion

Recent advances in MRI of pediatric congenital heart disease complement echocardiography and cineangiography. Two of these advances include the development of faster segmented k-space gradient-echo imaging and postprocessing of Gd-enhanced MR angiographic data including 3D volume rendering. MRI has a definitive role in the evaluation of atria, ventricles, pulmonary arteries, aorta, systemic veins, and pulmonary veins with respect to congenital heart disease lesions. In order for MRI of pediatric congenital heart disease to be effective and efficient, a basic understanding of the technical aspects of the examination and of the disease processes is essential.