Computed tomography angiography (CTA)
CT is a quick, non-invasive imaging modality with excellent spatial and temporal resolution. Modern CT scanners can provide sub-millimetre isotropic three-dimensional (3D) datasets within a single breath-hold during the first past of intravenous (IV) iodinated contrast medium (CM). One of the minimum requirements for more advanced CTA applications, such as coronary CTA, is a 64-channel CT; for many of the other less challenging vascular CT protocols, such as abdominal aorta or visceral aneurysm assessment, a 16-channel CT is adequate. The continued evolution of CT technology is based in no small part on the demands that cardiovascular imaging places in terms of speed, temporal resolution and scan volume. To help cope with the demands of cardiovascular imaging, manufacturers have made significant improvements in
z-axis volume coverage, detector and tube technology, with different emphasis depending on the vendor [
1]. State of the art wide-area detector CT scanners, with up to 320-detector rows, can provide up to 16-cm
z-axis detector coverage in a single gantry rotation; this allows for large volume coverage in both helical and axial (step and shoot) acquisition modes [
2]. Dual-source CT scanners provide the maximum temporal resolution available, as the temporal resolution is equal to a quarter of the gantry rotation time; this is as low as 66 milliseconds (ms) in the third-generation scanners. Maximising temporal resolution is advantageous when imaging structures prone to cardiac motion artefact, such as the aortic root [
3], or when imaging patients prone to motion, such as trauma patients or poor breath-holders.
Obtaining satisfactory arterial enhancement is crucial in the assessment of intravascular pathology. One of the important scan parameters that can influence arterial enhancement is scan acquisition time. A short acquisition time is preferable once the scan begins (in most situations) to ensure uniform arterial opacification on the acquired images. For helical scans, the acquisition time is equal to the gantry rotation time multiplied by the number of gantry rotations required to cover the anatomical area. The number of gantry rotations is determined by the scan range divided by the product of the detector bank width and the pitch. Axial scanning acquires multiple volumes, and is used in particular when performing ECG-gated studies such as coronary and aortic CTA, to help reduce radiation dose [
4]. For axial acquisitions of volumes smaller than the width of detectors, the scan time is equal to the gantry rotation time. When axial scanning is used for volumes larger than the detector width, the total scan time is equal to the total number of volumes required to cover the desired anatomical area multiplied by the gantry rotation time, added to the sum of the interscan time intervals required for table repositioning.
Optimising IV contrast medium (CM) administration is important in obtaining strong arterial enhancement during CTA. The degree of enhancement of a system is proportionally related to the concentration of iodine within it. There is variation in the relationship between enhancement and iodine concentration in different CT scanners, but it is the range of approximately 25-30 Hounsfield units (HU) per milligram (mg) of iodine per millilitre (ml) at 120 peak kilovoltage (kV) [
5]. The easily adjustable factors that determine arterial enhancement in CTA are the concentration of iodine in the CM used, the injection rate and the injection duration.
There is an almost linear relationship between enhancement and iodine concentration, which makes CM preparations with a high iodine concentration ideal (preferably 350-400 mg/ml) for CTA when the injection rate is fixed, resulting in a higher iodine delivery rate. The concept of iodine delivery rate (IDR, mg/s) is a method of standardising the rate of iodine delivery across CM with different iodine concentrations, and is calculated from the following formula: IDR = [CM iodine concentration (mg/ml)/1,000] × flow rate (ml/s) [
6]. High iodine delivery rates are important in providing diagnostic image quality in CTA [
7]. If CM preparations with lower concentrations of iodine are used, for example 300 mg/ml, the injection parameters can be adjusted to ensure a similar iodine delivery rate to that of a higher iodine concentration CM preparation[
8]; for example, in a porcine model undergoing CT pulmonary angiography, CM with an iodine concentration of 300 mg/ml injected at a rate of 5 ml/s provided an identical IDR of 1.5 g/s to 370 mg/ml CM injected at a rate of 4.1 ml/s [
9]. CM with lower iodine concentrations have lower viscosity, with reduced injection pressures, which may potentially reduce extravasation risk [
6,
8]; however, for a fixed scan duration, the higher injection rate required to keep the same iodine delivery rate would result in an increase in total CM volume required [
6]. Experimental models suggest that an IDR of 1.5-2 g/s provides adequate arterial opacification (>200 HU) in CTA protocols, regardless of the concentration of CM used [
10,
11].
The strength of arterial enhancement (peak HU) is proportional to the injection rate, and the duration of enhancement to the injection length [
12]. Increasing the injection rate leads to a faster accumulation of contrast in the aorta, increasing peak aortic enhancement. For a fixed contrast volume, however, this reduces injection duration, and in turn reduces the available time window to acquire the CT within. With modern scanners, an injection rate of 4-5 ml/s is usually sufficient in providing excellent arterial opacification for most vascular studies; venous imaging does not require as high injection rates. The traditional approach to determine injection duration was to match it to scan duration time; however, with modern multi-detector, fast CT scanners, this may result in inadequate opacification due to a lower volume of CM being delivered. One approach for estimating injection duration is to set a minimum duration of 10 s, and to add on the estimated scan duration time, which is available from all vendors after the scan range has been chosen. CTA requires use of a power-injector to allow uniform high injection rate CM bolus delivery. Use of a saline flush should be routine to help push the tail of the CM bolus into the central blood volume, as without it, the bolus tail would remain unused in the peripheral veins [
13]. The saline chaser also helps reduce intravascular CM dispersion and reduces streak artefact from dense contrast in the brachiocephalic veins and superior vena cava, which is especially important in thoracic CTAs [
14].
Obtaining high-quality arterial enhancement depends on various CT scanner, CM and patient-related factors. Even when CM and scanner use remain constant, patient factors such as body size, cardiac output and disease state can influence inter-individual variation in arterial enhancement. For example, large calibre and diseased vessels may take longer to opacify than normal. Reduction in cardiac output means that the CM bolus is slower to arrive and clear, resulting in delayed, but stronger peak arterial enhancement. These differences mean that the same scan timing delay cannot be used for everyone, and it needs to be tailored to the individual. Two methods commonly used to provide accurate CTA scan timing are the test bolus and bolus tracking methods. The test bolus method is based on injecting a small quantity (10-20 ml) of CM, then obtaining multiple low radiation dose images at a fixed time interval. By placing a region of interest (ROI) over the target vessel, a time-enhancement curve can be plotted to determine the time to peak enhancement. This can then be used to estimate the scan delay for the CTA. The bolus tracking technique involves acquiring a pre-contrast image at a reference level with placement of an ROI over a target vessel. After the CM injection is started, a low-dose monitoring scan is performed at a predetermined level after a fixed time delay, usually 5 s, and thereafter every 1-3 s until the enhancement in the ROI reaches a specified level (typically 150 HU). The CTA then begins after a pre-specified adjustable delay to allow peak arterial enhancement (approximately 8 s); this delay must also take into account time for table repositioning. The two methods are comparable in terms of satisfactory CTA timing, with bolus-tracking frequently used due to its reduced examination time and ease of use [
15]. The test bolus method is useful in patients with challenging anatomy, such as congenital heart disease patients post complex surgical repair [
16,
17].
The role of CM as a causative agent in acute kidney injury (AKI) is currently a topic of debate, with recent studies suggesting the risk of contrast-induced nephropathy (CIN) may not be as high as previously thought [
18‐
20]. The use of CM in patients with normal renal function is safe, with no evidence of a significant drop in glomerular filtration rate (GFR) post CM administration [
21]. To mitigate against the possible risk of CIN, CM should only be given to patients with severe renal dysfunction (GFR <30 ml/min) or AKI on a case-by-case basis after a risk-benefit analysis [
22‐
24]. There is no evidence available that reducing CM volume in patients with mild-to-moderate renal impairment (GFR 60-30 ml/min) has an effect on development of CIN.
Many modern scanners automate peak kilovoltage (kV) selection based on the topogram. Reducing the kV, for example from 120 to 100 in patients with a body mass index (BMI) of <25, can help improve image quality, and may potentially reduce radiation dose [
25]. Much of the radiation dose reduction achieved by reducing kV is offset in the presence of automatic exposure control (AEC), which increases the tube current to maintain a user-specified noise level [
26]; for example, to maintain diagnostic image quality, the tube current approximately doubles for a reduction from 120 to 100 kV [
27]. Small radiation dose reductions are still achievable with the use of model-based iterative reconstruction techniques [
28], but most of the benefit from reduced kV scanning in the presence of AEC in CTA comes from improved vessel contrast. Lower kV CTA has higher vessel HU values due to relatively increased attenuation of iodine as the kV nears its k-edge of 33 kV, improving image signal to noise and contrast to noise ratios [
26]. Lower injection rates should be used in reduced kV CTA (for example, 4 ml/s at 100 kV, 3 ml/s at 80 kV) to prevent the vessels appearing too high density, like bone; this has the added benefit of reducing the overall volume of CM required [
10,
29,
30]. There is, however, a cost to low kV CTA: the higher tube current required to reduce noise requires a larger focal spot, reducing spatial resolution [
31]. Blooming artefact from calcified atherosclerotic plaque or metal stents is also exaggerated at lower kV, which can be problematic in CTA interpretation [
4].
Dual-energy CT (DECT) is a state-of-the art technology that can improve image contrast in CTA by providing monoenergetic lower-energy reconstructions closer to the k-edge of iodine, with improved image contrast by a relatively increased contribution of the photoelectric effect [
32‐
35]. This can be accomplished using dual-source dual-energy (DSDE) CT systems that employ two separate X-ray tubes situated 90° apart that can operate at two different voltages, single-source dual-energy (SSDE) CT systems with fast kV switching or with single-source CT systems with a dual-layer of detectors [
36]. Low-energy monoenergetic reconstructions in CTAs with suboptimal vessel opacification can improve iodine attenuation to levels similar to conventional polyenergetic images obtained with higher volumes of contrast, allowing ‘rescuing’ of a suboptimal CTA [
37]; this also has the potential to reduce the iodine load required to obtain a diagnostic CTA, allowing the use of reduced concentration CM preparations and/or a lower volume [
38]. Virtual monoenergetic datasets reconstructed at a high kV can help reduce blooming artefact, allowing improved assessment of vascular stent patency [
39], and of heavily calcified vessels [
40]. DECT allows reconstruction of virtual non-contrast images from post-contrast CT acquisitions by excluding iodine-containing pixels, thus enhancing water attenuation [
36]; this has the potential to reduce radiation dose in multiphase vascular CT protocols, by obviating the need to acquire a separate non-contrast CT [
41,
42]. DECT can provide an assessment of organ perfusion using iodine map imaging. This is often presented using a colour look-up table, and can improve the detection of embolic disease by detecting areas of parenchymal hypoperfusion; this technique has been shown to improve the diagnostic accuracy of CTA in the detection of pulmonary emboli [
43,
44].
Cardiac motion artefact can be problematic when assessing the aortic root, and the use of ECG-gating in thoracic aorta CTA can help to address this. Motion artefact at the aortic root is dependent on several factors; chief among them, the gantry rotation time of the CT scanner and the patient’s heart rate [
4]. ECG-gating can help to reduce the ill-effects of cardiac motion on the aortic root, but it does not eliminate it. Pre-scan beta-blockade is another step that can help to reduce motion artefact, and is commonly used in coronary CTA; in practice, the administration of beta-blockers to patients undergoing routine thoracic aorta CTA is not often necessary to obtain satisfactory image quality, particularly with the improved temporal resolution of modern CTs [
45]. The available ECG-gating techniques include prospective, retrospective or high pitch gating, with the optimum choice largely scanner dependent [
1]. Scanners with large banks of detectors can cover the thorax quickly, making prospective gating ideal. Smaller detector-width scanners are more suited to retrospective gating with tube current modulation. High pitch gated acquisitions are suited to dual-source scanners. The use of ECG-gating does increase radiation dose, primarily determined by the number of cardiac phases, rather than the type of ECG-gating used. The optimum phase (percent of the R-R interval) for image acquisition is heart-rate dependent [
46]. When the heart rate is less than 75 beats per minute (bpm), a diastolic acquisition window of approximately 70-80% is preferred, with a systolic phase acquisition (30-40%) used in patients with higher heart rates.
There is no single best ‘one size fits all’ CTA protocol. Depending on the specific indication, it may be useful to obtain a non-contrast phase first; this can be of particular use when assessing calcified plaque, in postoperative patients, and in cases of suspected active haemorrhage. One approach described for a 64-channel CTA is to: (1) fix the scan duration to 10s for all CTAs; (2) adjust the pitch depending on the volume of coverage required; (3) fix the injection duration to 18 s; (4) operate a constant scan delay time of 8 s after CM arrival; (5) adjust the injection rate according to patient weight (5.0 ml/s for a 75-kg patient, ±0.5 ml/s for every 10 kg of body weight) [
27]. With this protocol, the long injection duration, combined with the extra 8 s delay post CM arrival, allows adequate time for arterial filling in nearly every patient. A delayed phase may also be helpful depending on the indication, with the timing measured from the end of the CM injection.
Magnetic resonance angiography (MRA)
MRA is a multiparametric imaging modality, with excellent contrast resolution. Contrast-enhanced MRA (CE-MRA) involves the administration of a gadolinium based contrast agent (GBCA), which shortens blood longitudinal relaxation (T1). A rapid 3D T1-weighted spoiled gradient echo (GRE) pulse sequence with a short repetition time (TR) and echo time (TE) is ideal for CE-MRA. This provides images with high signal-to-noise ratio (SNR), good spatial resolution and is free from flow-related artefacts [
47]. Subtraction techniques improve contrast resolution in CE-MRA. This reduces signal from background tissues by acquiring a mask image prior to GBCA injection, and subtracting it from the post-contrast imaging.
In general, an injection rate of 1.5 ml/s provides arterial imaging with high vessel to background contrast; this can be improved by increasing the injection rate, but similar to CTA, this reduces the available time window to acquire the scan for a fixed volume of contrast. Two methods are commonly used to appropriately time CE-MRA imaging. Similar to the method described for CTA, a test bolus method can be performed, administering 1-2 ml of GBCA and acquiring a series of rapid two-dimensional (2D) images of the vessel in question to determine the optimum time to start imaging post injection. In fluoroscopic triggering, the full bolus of contrast is administered and fluoroscopic-like images of the area of interest are obtained, and when the bolus is detected within the vessel, the technologist can trigger scan acquisition.
Two different acquisition modes are common in CE-MRA, single phase and time-resolved MRA. Single phase MRA captures vascular images at a single point in time. Time-resolved MRA consists of multiple acquisitions of an imaged volume over successive time points post GBCA administration. It is often known under vendor-specific acronyms such as TWIST (Siemens, Erlangen, Germany), TRICKS (General Electric, Chicago, IL, USA), 4D-TRAK (Philips, Best, Netherlands), TRAQ (Hitachi, Tokyo, Japan) and Freeze Frame (Toshiba, Otawara, Japan). This technique is particularly useful in displaying the passage of the contrast bolus through smaller vessels, such as the hands and feet. The core of time-resolved MRA is a 3D–spoiled GRE sequence employing k-space filling tricks to quicken image acquisition, such as non-Cartesian k-space filling, oversampling the centre of k-space (responsible for image contrast) and under sampling of the periphery (responsible for spatial resolution) [
48]. These techniques, aligned with the use of parallel imaging, delivers ultra-fast imaging [
49].
Paramagnetic contrast agents shorten the T1 and T2 relaxation times of water protons in their immediate surroundings, creating a locally increased magnetic field strength. This change in the local magnetic field strength results in increased local field inhomogeneity, driving the shortening of T1 and T2 relaxation. The resultant increased signal intensity (SI) on T1-weighted images provides the basis behind the use of contrast agents in MR. GBCAs are the most commonly used in MRA, and there are currently nine available GBCAs licensed by the European Medicines Agency (EMA) and the Food and Drug Administration (FDA) in the USA for clinical use. GBCAs can be divided into two different groups, linear and macrocyclic, based on how the ligand chelates the gadolinium ion [
50]. In the linear agents, the ligand wraps around the gadolinium (Gd
3+) ion, but does not completely enclose it. The macrocyclic agents consist of a chelator, which completely surrounds the Gd
3+ ion in a cage-like structure. The latter agents demonstrate greater stability in vivo than the linear agents, with little (if any) free Gd
3+ ion dissociation, reducing the risk of nephrogenic systemic fibrosis (NSF) [
51,
52]. The recent discovery of cerebral gadolinium deposition in patients with normal renal function is also of concern, although the clinical significance of this phenomenon is yet to be determined [
53]. Linear GBCAs are thought to confer a higher risk of cerebral deposition, and their use is now discouraged by the EMA [
54], although this guidance has not been reciprocated by the FDA [
55]. Cerebral gadolinium deposition is not only associated with linear GBCA use, however, and has been demonstrated in macrocyclic GBCAs in both animals [
56] and humans, in particular the macrocyclic agent gadobutrol [
57].
The majority of GBCAs in routine clinical use are extracellular fluid (ECF) agents. After injection, they initially distribute in the intravascular space, before rapidly diffusing across the vascular membranes into the interstitial space, eventually establishing an equilibrium between the intravascular and interstitial compartments after approximately 10 min. ECF GBCA agents that demonstrate weak plasma protein binding, such as gadobenate dimeglumine (Gd-BOPTA), help increase relaxivity compared to the other ECF GBCAs [
58]. Gadofosveset trisodium (Gd-DTPA-DO3A/MS-325) is currently the only intravascular GBCA licensed by the FDA. It was licensed by the European Medicines Agency for distribution in the European Union (EU) in 2005, but was voluntarily withdrawn from commercial use in the EU by the manufacturer in 2011. It is a linear ionic agent, which binds strongly to albumin, limiting its diffusion into the extravascular space [
59].
GBCAs with high relaxivity that remain within the blood-pool are the most attractive from an image quality point of view, however, safety considerations should be taken into account when choosing an agent. Gadobenate dimeglumine has the highest relaxivity of the ECF GBCAs, and gadofosveset trisodium is the only intravascular GBCA. These are both linear ionic agents, placing them in the intermediate risk category for NSF in susceptible patients, and at potentially higher risk for cerebral deposition. The current commercially available GBCA with the most favourable risk profile in terms of NSF and cerebral deposition is the macrocyclic ionic agent gadoterate meglumine. This agent has inferior protein binding and relaxivity compared with some of the other GBCAs, but this is offset by its safety profile. There have to-date been no reported cases of NSF with this agent, even in patients with severe renal dysfunction [
51,
52], and it is also the only GBCA in which cerebral deposition has not been demonstrated [
56].
For routine MRAs, a straightforward protocol is to begin with localisers of the anatomical area in question, followed by coronal and axial T2-weighted single-shot fast spin echo sequences, which allow a global anatomic assessment. This is then followed with 3D CE-MRA with two successive arterial phase acquisitions, providing anisotropic images, which allows reconstruction of the dataset on a 3D workstation. Finally, an axial T1-weighted 3D spoiled GRE sequence with a fat selective prepulse of the anatomical area can be acquired, allowing for an assessment for significant incidental findings. This basic MRA protocol can then be modified/added to according to the clinical question, as outlined in the anatomical site-specific sections below. It is our practice to administer weight-based GBCA dosing to all patients with GFR >30 ml/min, followed by a saline chaser (Table
2). There is no evidence available to support GBCA dose reduction in patients with mild to moderate renal impairment (GFR 60-30 ml/min). The risk of NSF remains in patients with severe renal dysfunction (GFR <30 ml/min); in these patients the decision to administer GBCA should be made on a case-by-case basis [
60].
For patients who cannot have gadolinium, such as those with severe renal impairment (GFR <30 ml/min), non-contrast imaging of the aorta and larger vessels can be performed. In these circumstances, a bright blood imaging, such as a balanced steady state free precession (SSFP) sequence can be used; this is a coherent gradient echo sequence that provides bright blood imaging without gadolinium. In the setting of acute aortic syndromes, this can detect the presence of an aortic dissection with high accuracy when compared with CE-MRA [
61,
62]. The major disadvantage of this sequence, however, is the presence of off-resonance artefact; this artefact is more pronounced at higher magnetic field strengths [
63]. To overcome this, where possible we perform non-contrast MRAs on 1.5 T rather than at 3.0 T.