Gauge
Injection rate (ml/s)
Pressure limits (psi)
24
≤2
≤100
22
≤3
≤300
20
≤4
18
≤5
As shown in Fig. 3.1, different combinations of contrast and saline can be used depending on the specific clinical scenario [6]. For most cases of simple CHD (i.e., evaluation of systemic or pulmonary vessels in the absence of shunts) or adult structural heart disease, a relatively compact bolus of contrast followed by a saline chaser (dual-phase protocol) is normally used (Fig. 3.2a). For many CHD cases where simultaneous opacification of both left and right cardiovascular structures is desired, a triple-phase protocol can be used (Fig. 3.2b). This typically comprises a faster initial injection of contrast, followed by a second slower injection and a saline chaser; alternatively, contrast injection can be followed for a mixture of contrast and saline (i.e., 60:40) at the same rate plus saline. These typically result in lower although sufficient attenuation of the pulmonary chambers compared to the systemic ones. Alternatively, a prolonged injection either of contrast alone at slower rates or contrast:saline mixture at standard rates (single-phase protocols) can result in similar although overall lower biventricular opacification. For cases where simultaneous arterial and venous evaluation is intended, an initial injection of contrast can be administered, and after a 30–60 s pause, a standard dual-phase protocol is performed, with the goal of obtaining sufficient attenuation within the arteries from the second injection and in the veins from the initial contrast bolus (triple-phase venous protocol).
Fig. 3.1
Schematic representation of different contrast injection protocols. The variation in contrast injection protocols is depicted, with relative volumes of contrast versus saline specified
Fig. 3.2
Examples of CTA in CHD. a 35-yo male with dextrotransposition of the great arteries and prior Mustard repair complicated with baffle stenosis treated percutaneously. The CTA demonstrates a widely patent stent between the left atrium and RV (arrow). b 2-yo female with double-outlet RV. The arrowhead indicates the ventricular septal defect. LV Left ventricle; RV right ventricle
Image Acquisition
Timing of acquisition is determined by the arrival of the contrast to the structure(s) of interest. In most cases, a region of interest (ROI) is placed in the target chamber and acquisition begins automatically when attenuation within the ROI raises above a certain threshold (automatic bolus tracking). In cases when the path of the contrast cannot be predicted because of unknown anatomy, a test bolus with a small contrast dose can be first performed to determine the delay to its arrival to the target structure, and subsequent CTA is timed accordingly. Alternatively, contrast can be tracked real time and acquisition initiated manually when the contrast arrives to the anatomic region of interest.
Although in the past images were often obtained without electrocardiographic (ECG) synchronization because ECG gating was associated with markedly increased radiation dose, today there are several ECG-synchronized scanning modes that allow for comparable or even lower doses [9]. Thus, we routinely perform cardiac CTA with ECG gating because it reduces motion artifact and improves image quality [9]. ECG-gated acquisition can be helical (spiral) or axial [10]. In the former, the table moves continuously during acquisition, while in the latter the table is stationary during imaging and moves in between acquisitions. Four types of ECG-gated CTA are currently available [10]:
Retrospective ECG-gated helical scan: In this scanning mode, radiation is given throughout the cardiac cycle and images are retrospectively reconstructed in the desired cardiac phase of the cardiac cycle. It has the advantages of allowing visualization of cardiac motion (cine imaging) and being more robust to arrhythmias; however, it requires higher radiation dose.
Prospective ECG-triggered axial scan: Images are acquired in a single phase of the cardiac cycle (at a time delay from the prior QRS complex) over several heartbeats. It significantly reduces radiation exposure since X-rays are delivered in only one phase, but it is susceptible to arrhythmias and tachycardia. When used for RPT, which phase of the cardiac cycle is preferable should be determined in advance.
Prospective ECG-triggered scan with a wide detector array (volumetric target scan mode): With a wide enough detector array (typically 320 detectors), the heart can be covered in one single heartbeat. Similarly, to the previous mode, the cardiac phase needs to be predetermined and quality is best with slow, regular heart rates.
Prospective ECG-triggered helical scan: In scanners with two X-ray tubes, detection of the QRS can trigger a high-pitch helical scan that allows covering large anatomic areas in a short time. This technique also requires slow, steady heart rates.
Volumetric and prospective helical scans afford the lowest radiation doses and should be used whenever possible; however, both scanning modes exist only in specific scanners. Prospective axial scanning, which is available in most modern scanners, is the alternative of choice. Retrospective gating is generally avoided unless quantification of cardiac function or valvular evaluation is desired.
Radiation Reduction
When performing cardiac CTA in general, and in infants or children in particular, it is imperative to aggressively reduce radiation exposure. As summarized in Table 3.2, a number of tools are available, and ideally as many of them in combination should be employed whenever possible [11]. Today, it is feasible to routinely perform sub-millisievert (mSv) scans in children and few mSv in adults [4, 11].
Table 3.2
Dose reduction tools
All scans | Limit scan length |
Reduce FOV | |
Decrease mA | |
Decrease kV | |
Use iterative reconstruction | |
Use anatomic-based current modulation (if possible) | |
Use thicker slice collimation (if possible) | |
Retrospective helical scanning | Use ECG-based current modulation |
Prospective axial scanning | Narrow acquisition window |
The main determinants of radiation exposure are scanned length and X-ray tube output. As a general principle, scanning should be limited to the region of interest. Some scanners allow additional dose savings if the field of view in the transverse plane is reduced. Both tube current (milliamperes or mA) and voltage (kilovolts or kV) should be adjusted to body size for every acquisition, including localizers and contrast tracking sequences. Reducing kV is the most effective way of decreasing radiation dose, and the minimum value that affords diagnostic signal-to-noise ratio should be used: This is typically 70–80 kV for neonates, infants, and young children, and 80–100 kV for older children and most adults. Milliamperes should be also minimized according to body size. Many scanners currently provide anatomic-based current modulation, by which mA, and in some scanners kV [12], are automatically increased or decreased based on the patient’s specific anatomic information collected from the initial localizers; however, this implementation may not be available in ECG-gated studies. As mentioned before, volumetric [13] or prospective helical scanning [7] should be used whenever feasible. If prospective axial scanning is employed [14], scanning window should be as narrow as possible. Retrospective ECG gating should ideally be avoided; otherwise, ECG-based tube current modulation (an implementation that only gives maximal dose during a certain phase of the cardiac cycle and minimizes it during the remaining) should be employed [7]. Newer iterative reconstruction algorithms result in significant noise reduction and allow additional reduction in tube settings [15]. Finally, while thinner slice collimations are preferred when imaging neonates or smaller structures such as valves or coronary arteries, other structures can be evaluated with thicker slices that result in reduced noise and similarly enable further dose savings. The details of considerations relevant to the performance of pediatric scans are covered in a later chapter.
Magnetic Resonance Imaging
Patient Selection and Preparation
When deciding whether CMR is the appropriate image modality to be utilized, MRI compatibility must be assessed of any implants, i.e., cardiac pacemakers and implanted cardiac defibrillators, or presence of metallic or ferromagnetic foreign bodies that can be subject to both thermal and mechanical forces during the scan. Other metallic devices such as stents, coils, or sternal wires may not be a contraindication to performance of the CMR, but may cause significant image artifact, rendering the dataset unusable to create a 3D virtual model. In patients such as infants or young children who are unable to breath-hold for a good quality MRA image dataset, i.v. placement with general anesthesia and intubation can be utilized. In working with the anesthesia team, an adequate breath-hold is used to allow lack of movement artifact during image acquisition. Patients who have acutely deteriorating renal function, have had nephrogenic systemic fibrosis or a previous anaphylactic reaction to a Gadolinium based contrast agent are not candidates to receive Gadolinium contrast and other imaging modalities must be considered.
CMR Technique and Image Analysis
The ability to create 3D cardiac models from CMR allows direct visualization of complex anatomy prior to entering the operating room [16–18]. Compared to cardiac CT, CMR offers the advantages of lack of radiation exposure, better temporal resolution, and good blood to myocardium differentiation without the necessary use of intravenous contrast. Both 3D balanced steady-state free precession (bSSFP) and magnetic resonance angiography (MRA) are commonly utilized CMR 3D sequences and either may be used to create a 3D model. Our group compared the quality of models created by these sequences, given that there was no published data on the optimal CMR sequence for 3D printed cardiac models [19].
The image datasets used for this study were retrospectively collected. The settings used for both MRA and post-contrast bSSFP at our institution are detailed. CMR was performed on a 1.5-tesla General Electric scanner (GE Signa HD®, GE Medical Systems, Waukesha, Wisconsin). Gadolinium-enhanced MRA was performed during respiratory suspension after administration of 0.2 mmol/kg Magnevist® (Berlex, Montville, New Jersey) at an injection rate of 1.5–2.0 ml/s followed by a 10–20 ml saline flush. Two acquisitions were performed using a non-electrocardiogram (ECG)-gated, 3D spoiled fast gradient-echo sequence, also known as fast low-angle shot (FLASH), with the following parameters: echo time (TE) 1–2 ms, repetition time (TR) 3–5 ms, flip angle 40°, receiver bandwidth 62.5 kHz/s, rectangular field of view, coronal orientation, acquired slice thickness 2.4–3.0 mm interpolated to 1.2–1.5 mm, and matrix size adjusted to produce near-isotropic voxels with spatial resolution of ~1.6–2.8 mm. The typical breath-hold time was 15–30 s. The image acquisition was done after a delay of approximately 10 s after contrast injection, with the goal of having the contrast present in both sides of the heart.
After the contrast-enhanced MRA, ECG-gated and respiratory navigated isotropic 3D bSSFP images were acquired in the following manner: sagittal acquisition, 224 − 192 × 224 − 192 (frequency × phase) matrix, slice thickness 2.4–3.0 mm interpolated to 1.2–1.5 mm, frequency field of view 240–300 mm with 100% FOV in the phase direction, TE 1.09–1.69 ms, TR 3.14–3.6 ms, flip angle 60°, and receiver bandwidth 125 kHz/s. Acquisition was triggered to mid-diastole, and temporal resolution was adjusted for faster heart rates. The navigator was set to acquire during end-expiration, with a tracker length of 5–10 cm and an acceptance window of 1–2 cm.
Comparison of Models by Image Acquisition Sequence
Both blood pool and myocardial segmentation were used to create a 3D model in each of our patients from each set of either bSSFP or MRA source images resulting in 76 models categorized in four groups: Group 1—bSSFP/MS, Group 2—bSSFP/BP, Group 3—MRA/MS, and Group 4—MRA/BP (Fig. 3.3). In BP segmentation, a 1-mm-thick layer was created onto the 3D object which represented the blood pool. The 3D object was then “hollowed” internally, excluding the 1-mm layer, allowing for the intracardiac anatomy to be represented. For MS, the threshold cutoffs of gray values were set to isolate the myocardium. Once the models were created, the quality of the models using bSSFP and MRA was compared.
