Magnetic resonance (MR) angiography of the pulmonary arteries is a rapidly evolving technique with proven clinical usefulness. Multiple-step protocols, such as MR perfusion followed by high–spatial resolution MR angiography, seem to be a good approach for the assessment of different vascular diseases affecting the pulmonary arteries. In combination with other imaging sequences, MR imaging is one of the most comprehensive potential noninvasive imaging techniques available.
Key points
- •
Magnetic resonance (MR) angiography of the pulmonary arteries has proven clinical usefulness.
- •
Contrast-enhanced (CE) and non-CE angiographic techniques are widely available for high spatial and real-time imaging of the pulmonary arteries.
- •
Multiple-step protocols, such as MR perfusion followed by high–spatial resolution contrast-enhanced magnetic resonance angiography (CE-MRA), seem to be an optimal clinical approach for the assessment of different vascular diseases affecting the pulmonary arteries.
Introduction
Given the speed and robustness of modern multidetector computed tomography (MDCT) scanners, this technique has become the noninvasive gold standard for imaging of the pulmonary arteries. The advantages of MDCT are fast examination with high resolution and visualization of even subsegmental pulmonary arteries. The technique is usually available 24/7 and allows exclusion of other causes of chest pain in the same examination. As a drawback, the amount of functional information is usually limited.
Direct visualization of the pulmonary arteries can be done using invasive techniques such as digital subtraction angiography (DSA), which are nowadays reserved for special preoperative settings, such as chronic thromboembolic pulmonary hypertension (CTEPH) and settings in which an invasive mean pulmonary arterial pressure measurement is required. Computed tomography (CT) and DSA have the inherent problem of irradiation and the need for nephrotoxic contrast agents.
MR imaging has evolved as a competitive noninvasive imaging modality. Over the past decades MR imaging has undergone significant technical improvements such as faster acquisitions, larger coverage, and faster reconstruction, leading to substantially improved patient acceptance of this modality. Furthermore, the availability of MR systems has improved over the past years. These factors have positively influenced development of MR applications in the chest. The implementation of MR angiography (MRA), lung perfusion imaging, and the assessment of right heart function seem to be promising techniques for a comprehensive evaluation in patients with either congenital or acquired pulmonary arterial pathologies.
This review highlights the current state of various MRA techniques for the diagnosis of acute and chronic thromboembolic disease.
Technique
Different imaging strategies are available for imaging pulmonary arteries including CE and non-CE acquisitions. The most often used technique is high–spatial resolution CE-MRA.
Contrast-enhanced magnetic resonance angiography with high spatial resolution
CE-MRA consists of heavily T1-weighted gradient echo MR sequences after an intravenous injection of a paramagnetic MR contrast agent. In general, 3-dimensional (3D) techniques with a relaxation time (TR) of less than 5 ms and an echo time (TE) of less than 2 ms are used for CE-MRA of the pulmonary arteries. A short TR allows for short breath-hold acquisitions, and a short TE minimizes background signal and susceptibility artifacts. Nowadays, acquisition time has been shortened further using parallel imaging. In parallel imaging, the image is reconstructed from an undersampled k-space in the phase-encoding direction and thus acquisition time decreases. With a typical acceleration factor of 2, every second line in k-space is skipped, which leads to reduction of scan time of approximately 50%. This reduced acquisition time can also be traded for higher spatial resolution. The trade-off for the reduction of scan time or the higher spatial resolution is a decreased signal to noise ratio (SNR). It is inversely proportional to the square root of the acceleration factor times a geometry factor that is determined mainly by the coil design. In the case of an acceleration factor of 2, the signal is at best 71% of the original signal. Although an acceleration factor of 3 seems to be acceptable for the renal arteries, an acceleration factor of 2 is usually recommended for the pulmonary arteries; this leads to high spatial resolution with a voxel size of 1.2 mm × 1.0 mm × 1.6 mm requiring a breath-hold of 20 to 30 seconds for acquisition. Artifacts in the center of the image can appear if acceleration factors of 2 are used in a coronal acquisition. To overcome this problem, patients are scanned with their arms above their heads, which could, however, cause discomfort in some patients. The use of non-Cartesian, k-space filling techniques, such as radial and spiral image data acquisition, has also been proposed for use in the chest. As breath-hold is crucial for image quality, the scan is generally acquired in the coronal orientation because the number of slices required for full coverage is lower than in other orientations ( Fig. 1 ). This method requires a single injection of contrast. To improve spatial resolution and reduce the duration of the breath-hold, the sequential acquisition of 2 sagittal slabs covering the right and left lung separately has also been used successfully in patients with CTEPH.
By combination of the latest technical developments, such as 32-channel chest coil, 3T MR imaging, high relaxivity contrast agent, and acceleration factor of 6, the acquisition of isotropic (1 × 1 × 1 mm 3 ) voxels covering the entire pulmonary circulation in 20 seconds is feasible.
Contrast administration
For T1 shortening of the blood, a gadolinium (Gd) compound is injected in a peripheral vein as a bolus, preferably by an automated power injector. Mostly, standard-strength Gd compounds in a standard dose (0.1 mmol/kg body weight) are used for optimal opacification of the pulmonary arteries. To guarantee an optimal bolus profile, the administration of the contrast agent with flow rates between 2 and 5 mL/s is mandatory. The bolus geometry is mainly determined by injection parameters; cardiac function is of minor importance. An injection scheme with a flow rate of 2 mL/s results in a mean transit time of the contrast agent through the pulmonary circulation of 14 seconds, and it is reduced to 9 seconds using an injection speed of 4 mL/s. The administration of a saline flush, minimum 20 mL injected at the same flow rate, immediately afterward is strongly recommended to achieve a compact bolus profile. The saline chaser ensures that the whole amount of contrast medium contributes to vessel opacification. Adequate timing of the contrast agent together with an adapted flow rate is essential to achieve high contrast between pulmonary artery branches and surrounding structures. The Gd concentration should be optimal at the time of central k-space acquisition, as this determines the vascular signal intensity. Because a significant number of patients referred for MRA of the pulmonary arteries present with right heart compromise and/or pulmonary hypertension, the scan delay should be individually adjusted using a care bolus procedure or a test bolus examination. In general, arterial enhancement should dominate because overlay of pulmonary veins might otherwise impair the assessment of arteries.
Initially, MRA was seen as a promising tool for patients with renal insufficiency avoiding the nephrotoxicity of iodinated CT contrast media. However, a series of cases with nephrogenic systemic fibrosis came up around 2008. It was soon found that the prevalence was different with different Gd compounds. Based on this, the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study that was underway adopted a conservative approach and lowered the Gd dose from 0.2 mmol/kg to 0.1 mmol/kg (single dose). This lower dose of Gd was later even found to have superior image quality.
In an animal model (rabbit) a blood pool contrast agent outperformed a standard Gd-DOTA (gadoteric acid) contrast media for detection of embolic vascular pathologies. Given the primarily intravascular distribution of the contrast media, high-resolution MR angiographies could be obtained up to 15 minutes after injection.
Contrast-enhanced magnetic resonance angiography with high temporal resolution
Another approach is to optimize the MR sequence for high temporal resolution and to apply it as a multiphasic acquisition. In time-resolved MRA, the scan time for the individual 3D data set is reduced to less than 5 seconds. The rationale is 3-fold: first, patients with severe respiratory disease and limited breath-hold capabilities can be examined. Second, the arterial-venous discrimination is improved, allowing for characterization of vascular territories, especially in anomalies and shunts. Third, time-resolved multiphasic CE-MRA is independent from the bolus timing, because the contrast injection and the MR sequence are started simultaneously. Advanced time-resolved imaging techniques integrate a view-sharing approach to achieve a temporal resolution of 3.3 seconds for a 3D data set with a high spatial resolution of, for example, 1.3 × 1.8 × 3 mm 3 . Latest experimental approaches combine view sharing with spiral k-space filling, allowing for even higher–spatial resolution images with shorter acquisition times. If the temporal resolution is substantially reduced further, that is, 1 second per 3D data set, the perfusion of the lung parenchyma can be assessed parallel to the central pulmonary arteries, allowing for easy depiction of perfusion deficits due to vascular obstruction ( Fig. 2 ).
Noncontrast-enhanced overlapping steady-state free precession sequences
Critically ill patients do not even tolerate a short breath-hold time of 5 to 10 seconds for CE-MRA. The same is true for CT angiography (CTA), and its imaging results might also be suboptimal in these scenarios. For these patients, free-breathing real-time imaging techniques based on steady-state free precession (SSFP), also called balanced fast-field echo or fast imaging using steady-state acquisition, are available. The whole chest can be covered in all 3 orientations in less than 180 seconds with 50% overlapping of the slices due to an acquisition time per image of approximately 0.4 to 0.5 seconds ( Fig. 3 ). This approach allows for a lobar and segmental evaluation of the pulmonary arteries. Real-time MR imaging showed high specificity (98%) and sensitivity (89%) in a study of patients with acute pulmonary embolism (PE), with 16-slice MDCT serving as the reference modality.
Noncontrast-enhanced respiratory-gated steady-state free precession
One technique that may offer an alternative to breath-hold imaging is navigator-gated MR imaging, in which imaging is performed during free breathing. The navigator was first described in 1989 by Ehman and Felmlee, and it has been used primarily to image blood vessels that are subject to respiratory and cardiac motion, particularly the coronary arteries. However, this method has rarely been used for pulmonary imaging. With the advent of faster gradient systems and continued development of SSFP, it has become possible to perform rapid 3D imaging. The faster gradient systems allow for reduced repetition times, making SSFP practical. Further, the refocusing nature of SSFP, as opposed to the standard spoiled gradient echo technique, allows for both a higher flip angle and many more views per segment before significant signal decay. In a study in healthy volunteers, a breath-hold SSFP sequence was compared with a navigator-gated SSFP. Both sequences resulted in comparable image quality with the same SNR level; however, no analysis was performed regarding vessel conspicuity on a segmental level. Image acquisition was approximately 29 seconds for the breath-hold and 180 seconds for the navigator technique. Therefore, in critically ill patients, this technique may be worth trying as an alternative.
Besides respiratory artifacts, pulsation artifacts lead to impaired image quality. This drawback can be overcome by electrocardiography (ECG) and respiratory-triggered SSFP techniques (see Fig. 3 ). In the aforementioned study (21 patients), this double-triggered sequence showed a sensitivity of 67% and a specificity of 100%.
3D gradient echo during recirculation phase
Image acquisition during the first pass of the contrast media through the pulmonary arteries is the mainstay of visualization of the pulmonary arteries. However, this approach is a one shot try and sometimes results in nonoptimal images. Given the excellent enhancement of the vasculature by the current contrast media, even late image acquisitions up to 10 minutes are possible. This method was used in the study by Kalb and colleagues, whereby approximately 3 minutes after the first-pass angiography a late-phase angiography data set was acquired. An axial breath-hold (15 seconds) 3D gradient echo (GRE) sequence was used with a spatial resolution of 1.3 × 1.3 × 2.8 mm 3 . This approach and sequence had the highest accuracy (compared with MRA and triggered SSFP) with a sensitivity of 73% ( Fig. 4 ).
The sequences described earlier can be used in various combinations to provide a stepwise approach to comprehensive assessment of pulmonary arteries and perfusion deficits. A combination of time-resolved angiography followed by high–spatial resolution angiography provides comprehensive assessment of perfusion defects and pulmonary arterial emboli. The authors’ preferred institutional protocol is shown in Table 1 .
Rationale | Sequences |
---|---|
Initial overview of morphology | T2w HASTE transverse & coronal |
Vascular imaging (noncontrast) | SSFP transverse & coronal |
Contrast-enhanced vascular imaging and perfusion | FLASH 3D (time resolved), 0.1 mmol/kg |
Postcontrast assessment (between time-resolved and high–spatial resolution angiography) | VIBE transverse & coronal |
Contrast-enhanced vascular imaging | FLASH 3D (high spatial resolution), 0.1 mmol/kg |
Introduction
Given the speed and robustness of modern multidetector computed tomography (MDCT) scanners, this technique has become the noninvasive gold standard for imaging of the pulmonary arteries. The advantages of MDCT are fast examination with high resolution and visualization of even subsegmental pulmonary arteries. The technique is usually available 24/7 and allows exclusion of other causes of chest pain in the same examination. As a drawback, the amount of functional information is usually limited.
Direct visualization of the pulmonary arteries can be done using invasive techniques such as digital subtraction angiography (DSA), which are nowadays reserved for special preoperative settings, such as chronic thromboembolic pulmonary hypertension (CTEPH) and settings in which an invasive mean pulmonary arterial pressure measurement is required. Computed tomography (CT) and DSA have the inherent problem of irradiation and the need for nephrotoxic contrast agents.
MR imaging has evolved as a competitive noninvasive imaging modality. Over the past decades MR imaging has undergone significant technical improvements such as faster acquisitions, larger coverage, and faster reconstruction, leading to substantially improved patient acceptance of this modality. Furthermore, the availability of MR systems has improved over the past years. These factors have positively influenced development of MR applications in the chest. The implementation of MR angiography (MRA), lung perfusion imaging, and the assessment of right heart function seem to be promising techniques for a comprehensive evaluation in patients with either congenital or acquired pulmonary arterial pathologies.
This review highlights the current state of various MRA techniques for the diagnosis of acute and chronic thromboembolic disease.
Technique
Different imaging strategies are available for imaging pulmonary arteries including CE and non-CE acquisitions. The most often used technique is high–spatial resolution CE-MRA.
Contrast-enhanced magnetic resonance angiography with high spatial resolution
CE-MRA consists of heavily T1-weighted gradient echo MR sequences after an intravenous injection of a paramagnetic MR contrast agent. In general, 3-dimensional (3D) techniques with a relaxation time (TR) of less than 5 ms and an echo time (TE) of less than 2 ms are used for CE-MRA of the pulmonary arteries. A short TR allows for short breath-hold acquisitions, and a short TE minimizes background signal and susceptibility artifacts. Nowadays, acquisition time has been shortened further using parallel imaging. In parallel imaging, the image is reconstructed from an undersampled k-space in the phase-encoding direction and thus acquisition time decreases. With a typical acceleration factor of 2, every second line in k-space is skipped, which leads to reduction of scan time of approximately 50%. This reduced acquisition time can also be traded for higher spatial resolution. The trade-off for the reduction of scan time or the higher spatial resolution is a decreased signal to noise ratio (SNR). It is inversely proportional to the square root of the acceleration factor times a geometry factor that is determined mainly by the coil design. In the case of an acceleration factor of 2, the signal is at best 71% of the original signal. Although an acceleration factor of 3 seems to be acceptable for the renal arteries, an acceleration factor of 2 is usually recommended for the pulmonary arteries; this leads to high spatial resolution with a voxel size of 1.2 mm × 1.0 mm × 1.6 mm requiring a breath-hold of 20 to 30 seconds for acquisition. Artifacts in the center of the image can appear if acceleration factors of 2 are used in a coronal acquisition. To overcome this problem, patients are scanned with their arms above their heads, which could, however, cause discomfort in some patients. The use of non-Cartesian, k-space filling techniques, such as radial and spiral image data acquisition, has also been proposed for use in the chest. As breath-hold is crucial for image quality, the scan is generally acquired in the coronal orientation because the number of slices required for full coverage is lower than in other orientations ( Fig. 1 ). This method requires a single injection of contrast. To improve spatial resolution and reduce the duration of the breath-hold, the sequential acquisition of 2 sagittal slabs covering the right and left lung separately has also been used successfully in patients with CTEPH.
By combination of the latest technical developments, such as 32-channel chest coil, 3T MR imaging, high relaxivity contrast agent, and acceleration factor of 6, the acquisition of isotropic (1 × 1 × 1 mm 3 ) voxels covering the entire pulmonary circulation in 20 seconds is feasible.
Contrast administration
For T1 shortening of the blood, a gadolinium (Gd) compound is injected in a peripheral vein as a bolus, preferably by an automated power injector. Mostly, standard-strength Gd compounds in a standard dose (0.1 mmol/kg body weight) are used for optimal opacification of the pulmonary arteries. To guarantee an optimal bolus profile, the administration of the contrast agent with flow rates between 2 and 5 mL/s is mandatory. The bolus geometry is mainly determined by injection parameters; cardiac function is of minor importance. An injection scheme with a flow rate of 2 mL/s results in a mean transit time of the contrast agent through the pulmonary circulation of 14 seconds, and it is reduced to 9 seconds using an injection speed of 4 mL/s. The administration of a saline flush, minimum 20 mL injected at the same flow rate, immediately afterward is strongly recommended to achieve a compact bolus profile. The saline chaser ensures that the whole amount of contrast medium contributes to vessel opacification. Adequate timing of the contrast agent together with an adapted flow rate is essential to achieve high contrast between pulmonary artery branches and surrounding structures. The Gd concentration should be optimal at the time of central k-space acquisition, as this determines the vascular signal intensity. Because a significant number of patients referred for MRA of the pulmonary arteries present with right heart compromise and/or pulmonary hypertension, the scan delay should be individually adjusted using a care bolus procedure or a test bolus examination. In general, arterial enhancement should dominate because overlay of pulmonary veins might otherwise impair the assessment of arteries.
Initially, MRA was seen as a promising tool for patients with renal insufficiency avoiding the nephrotoxicity of iodinated CT contrast media. However, a series of cases with nephrogenic systemic fibrosis came up around 2008. It was soon found that the prevalence was different with different Gd compounds. Based on this, the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study that was underway adopted a conservative approach and lowered the Gd dose from 0.2 mmol/kg to 0.1 mmol/kg (single dose). This lower dose of Gd was later even found to have superior image quality.
In an animal model (rabbit) a blood pool contrast agent outperformed a standard Gd-DOTA (gadoteric acid) contrast media for detection of embolic vascular pathologies. Given the primarily intravascular distribution of the contrast media, high-resolution MR angiographies could be obtained up to 15 minutes after injection.
Contrast-enhanced magnetic resonance angiography with high temporal resolution
Another approach is to optimize the MR sequence for high temporal resolution and to apply it as a multiphasic acquisition. In time-resolved MRA, the scan time for the individual 3D data set is reduced to less than 5 seconds. The rationale is 3-fold: first, patients with severe respiratory disease and limited breath-hold capabilities can be examined. Second, the arterial-venous discrimination is improved, allowing for characterization of vascular territories, especially in anomalies and shunts. Third, time-resolved multiphasic CE-MRA is independent from the bolus timing, because the contrast injection and the MR sequence are started simultaneously. Advanced time-resolved imaging techniques integrate a view-sharing approach to achieve a temporal resolution of 3.3 seconds for a 3D data set with a high spatial resolution of, for example, 1.3 × 1.8 × 3 mm 3 . Latest experimental approaches combine view sharing with spiral k-space filling, allowing for even higher–spatial resolution images with shorter acquisition times. If the temporal resolution is substantially reduced further, that is, 1 second per 3D data set, the perfusion of the lung parenchyma can be assessed parallel to the central pulmonary arteries, allowing for easy depiction of perfusion deficits due to vascular obstruction ( Fig. 2 ).