Guidelines and protocols for cardiovascular magnetic resonance in children and adults with congenital heart disease: SCMR expert consensus group on congenital heart disease

Abstract

Cardiovascular magnetic resonance (CMR) has taken on an increasingly important role in the diagnostic evaluation and pre-procedural planning for patients with congenital heart disease. This article provides guidelines for the performance of CMR in children and adults with congenital heart disease. The first portion addresses preparation for the examination and safety issues, the second describes the primary techniques used in an examination, and the third provides disease-specific protocols. Variations in practice are highlighted and expert consensus recommendations are provided. Indications and appropriate use criteria for CMR examination are not specifically addressed.

Figure 1

Spin echo and gradient echo cine. 63-year-old patient who has undergone surgical repair of coarctation of the aorta. The images are oriented parallel to the long-axis of the aortic arch and shown in diastole. A. ECG-triggered fast spin echo sequence acquired with a double inversion preparation pulse to suppress signal from flowing blood. Note that the resulting signal from blood is dark. B. ECG-gated steady-state free precession cine sequence. Note the signal from blood is bright.

Figure 2

Stainless steel coil artifact. 17-year-old patient with dextrocardia, double-outlet right ventricle, and pulmonary stenosis who has undergone a Rastelli operation and catheter implantation of a single stainless steel vascular coil to occlude a small left superior vena cava draining to the left atrium. ECG-gated steady-state free precession sequence in a coronal plane (A) and axial plane (B) demonstrating a region of signal loss in the upper left chest (arrow) resulting from the coil. The area of signal loss is several times larger than the coil and obscures a portion of the left pulmonary artery. (C) shows an ECG-triggered fast spin echo sequence with a double inversion preparation pulse acquired in diastole with a similar orientation as in (B) With this sequence, the extent of signal loss artifact is reduced, and the left pulmonary artery is more clearly visualized.

Figure 3

Contrast-enhanced magnetic resonance angiography. 9-year-old patient with partially anomalous pulmonary venous return of the left upper pulmonary vein (arrow) to the leftward aspect of the left innominate vein. Contrast-enhanced magnetic resonance angiogram shown in a coronal plane using a sub-volume maximal intensity projection (A) and volume rendering (B).

Figure 4

3D steady-state free precession. A patient with transposition of the great arteries who has undergone a Senning operation. An ECG and respiratory navigator-gated 3D SSFP sequence was utilized to generate a 3D volume with 1.5 mm isotropic resolution timed to mid-diastole. The navigator efficiency was 45% and the acquisition time was 6 minutes. Multiplanar reformatting of this volume allows a comprehensive morphologic evaluation of the heart and great vessels including the Senning pathways (A, B, and C).

Figure 5

Coronary MR angiography. Patient with anomalous origin of the right coronary artery from the left aortic sinus of Valsalva. Coronary angiography was performed using an ECG and respiratory navigator-gated 3D SSFP sequence with data acquisition timed to the diastolic rest period of the cardiac cycle. Multiplanar reformatting oriented in short-axis to the aortic root yielded this image.

Figure 6

Ventricular views. The diagram illustrates one approach to planning standard ventricular views (right column) based on adjusting the slice locationon two other views (left and middle columns). Note that for the 4C, LV 2C, and LV 3C, the imaging plane is carefully positioned to pass through the apex of the LV and bisect the mitral valve plane. LV, left ventricle; RV, right ventricle; RVOT, right ventricular outflow tract; SA, short-axis; 2C, 2-chamber; 3C, 3-chamber; 4C, 4-chamber.

Figure 7

Planning for ventriculography. An axial stack of cine images for ventriculography is planned by adjusting the slice locations on both coronal and sagittal images (top row). A short-axis stack of cine images for ventriculography is planned by adjusting the slice locations on 4-chamber (4C) and left ventricular 2-chamber (LV 2C) images in diastole (bottom row). Note that both the axial and short-axis stacks are prescribed to ensure complete coverage of the left and right ventricles. In this short-axis example, the slices are oriented perpendicular to the ventricular septum on the 4C view, and care is taken to ensure that coverage includes the anterior portion of the dilated right ventricle which extends above the tricuspid valve plane. An alternative short-axis planning approach is to orient the slices parallel to the atrioventricular valve plane on the 4C view (not shown).

Figure 8

Tracing ventricular borders. Diagram demonstrating the drawing of left and right ventricular endocardial contours in end-diastole. Images may be acquired in a ventricular short-axis orientation (A and B) or in an axial orientation (C and D). There is practice variation regarding whether to trace the papillary muscles and right ventricular trabeculations to exclude them from (A and C) versus include them in (B and D) the blood pool.

Figure 9

Effect of aliasing on phase-contrast cine CMR (PC CMR) flow measurements. Sixteen-year-old patient with surgically repaired tetralogy of Fallot and mild pulmonary valve stenosis. PC CMR was performed in the main pulmonary artery with the velocity range (v_enc) set incorrectly at 200 cm/sec (top row) and then with the v_enc set correctly at 300 cm/sec (bottom row). Magnitude (A, D) and phase images (B, E) in systole, and the resulting flow curves (C, F) generated from analyzing the region of interest (yellow contour) are shown. Because the peak velocity is 260 cm/sec, aliasing (B) and flow underestimation (C) are seen with a v_enc of 200 cm/sec but not with a v_enc of 300 cm/sec (E and F).

Figure 10

Planning a CMR phase-contrast acquisition to measure flow in the main pulmonary artery. The PC CMR imaging plane is simultaneously viewed and adjusted on orthogonal views of the main pulmonary artery (top row) to ensure that it is oriented perpendicular to the blood vessel. The resulting magnitude and phase images are shown (bottom row).

Figure 11

Proper positioning of the imaging plane for main pulmonary artery blood flow measurement. PC CMR velocity and flow measurements are most accurate when the location of interest is at the isocenter of the scanner during the acquisition. Most MR scanners will slide the patient table so that the center of the imaging plane (yellow circle) is at the scanner’s isocenter (vertical red line). Prescribing the imaging plane so that the center of the image is at the same level in the superoinferior dimension as the location of interest is therefore recommended.

Figure 12

First-pass perfusion. Patient with anomalous origin of the left coronary artery from the pulmonary artery who underwent left coronary artery re-implantation and subsequently developed severe stenosis of the re-implanted coronary artery. First-pass perfusion images at a mid-ventricular level with adenosine stress (A) and at rest (B). Note the extensive sub-endocardial hypoperfusion of the left ventricle at stress but not at rest which indicates inducible ischemia.

Figure 13

Schematic diagram illustrating data acquisition timing in a first-pass perfusion sequence. Each rectangle represents data acquisition to form one complete image and the numbers inside them correspond to different slice locations. At a heart rate of 60 bpm (A), the cardiac cycle length is long enough to allow acquisition of 4 slice locations during each beat. The slice locations are timed to different phases of the cardiac cycle, but each location is acquired repeatedly at the same phase in subsequent cycles. At a heart rate of 120 bpm (B), the cardiac cycle length is shorter so only two slice locations can be acquired over each beat; the other 2 locations are acquired the following beat. Note that the temporal resolution (images per unit time) is the same in A and B.

Figure 14

Late gadolinium enhancement following a Fontan procedure. Patient with tricuspid atresia and normally related great arteries who underwent a Fontan procedure. Late gadolinium enhancement image in a ventricular short-axis view showing enhancement and wall thinning of the inferior septum consistent with a chronic myocardial infarction.

Figure 15

Late gadolinium enhancement in endocardial fibroelastosis. Patient with a history of severe congenital aortic valve stenosis who underwent balloon valvuloplasty. Late gadolinium enhancement images in 4-chamber (A) and short-axis views (B). Note the extensive subendocardial late gadolinium enhancement consistent with endocardial fibroelastosis.

Figure 16

Schematic diagram illustrating data acquisition timing in a late gadolinium enhancement sequence. Each rectangle represents data acquisition timed to coincide with the cardiac rest period and during which a user-defined number of k-space lines are filled. Multiple data acquisitions and thus cardiac cycles are required to fill k-space and produce an image. At a heart rate of 80 bpm (A), data acquisition occurs every second cardiac cycle in order to allow sufficient time for recovery of longitudinal signal. At a heart rate of 120 bpm (B), the cardiac cycle length is shorter so the sequence is modified to acquire data every third cycle in order to maintain the same time for the recovery of longitudinal signal. In addition, the data acquisition duration is shortened to compensate for the briefer cardiac rest period associated with a faster heart rate.

Figure 17

Secundum atrial septal defect (ASD) imaging protocol. A. Axial steady-state free precession (SSFP) image in a 10-year-old with a large secundum ASD (white arrow). B. The axial SSFP image is used to plan a stack of oblique sagittal SSFP images to visualize the ASD and the superior and inferior defect margins. C. The axial and oblique sagittal images are used together to plan a stack of phase-contrast (PC) cine images to visualize the ASD flow en face. This provides insight into the oval shape of the defect and may demonstrate additional ASDs.

Figure 18

Sinus venosus septal defect imaging protocol. A. Axial steady-state free precession (SSFP) cine image in a 22-year-old with a large sinus venosus septal defect (white arrow). B. The axial SSFP image is used to plan a stack of oblique sagittal SSFP cine images to visualize the defect in an orthogonal view and assess its superoinferior dimension.

Figure 19

Ebstein anomaly imaging protocol. A. Right ventricular 3-chamber view (RV 3C) steady-state free precession cine image in a 22-year-old with Ebstein anomaly. In this example, the functional tricuspid valve plane is displaced and inflow is directed into the right ventricular outflow tract (white arrow). B. The RV 3C image is used to plan a stack of cine images to visualize the displaced tricuspid valve orifice en face for anatomic assessment or flow quantification.