27

MR Angiography^*

Introduction

In this chapter, we will discuss the topic of MR angiography (MRA). As in the last chapter, MRA may at first appear very complicated, but we’ll try to present the major concepts in a simplified fashion. There are three main MRA techniques:

1. TOF (time-of-flight) MRA
2. PC (phase contrast) MRA
3. CE (contrast-enhanced) MRA

TOF and PC techniques can be performed using two-dimensional Fourier transform (2DFT) or three-dimensional FT (3DFT). CE MRA is performed with the 3D technique. Thus, there are a total of five different methods:

1. 2D-TOF MRA
2. 2D-PC MRA
3. 3D-TOF MRA
4. 3D-PC MRA
5. 3D-CE MRA

Each of these techniques lends itself to a different type of clinical application.

TOF MRA

TOF MRA is based on flow-related enhancement (FRE; discussed in the previous chapter) in a 2D or 3D gradient-echo (GRE) technique. (Remember that in GRE imaging, TOF losses do not play an important role.) Usually, flow compensation is used perpendicular to the vessel lumen.

2D-TOF MRA.

Figure 27-2 Axial source image from a 2D-TOF demonstrates no flow…

Figure 27-1 depicts a typical pulse sequence for 2D-TOF MRA. A presat (presaturation) pulse is applied above or below each slice to eliminate the signal from vessels flowing in the opposite direction. Usually a short TR (about 50 msec), a moderate flip angle (45° to 60°), and a short TE (a few millisecond) are used. An example is seen in Figure 27-2.

Figure 27-1 A PSD for 2D-TOF MR angiography.

3D-TOF MRA.

Figure 27-3 depicts a pulse sequence diagram (PSD) for a 3D-TOF MRA. Here, a slab of several centimeters (usually about 5 cm) is obtained that contains up to 60 slices.

Figure 27-3 A PSD for 3D-TOF MRA.

Advantages of 2D-TOF MRA

1. Faster scanning
2. Maximized FRE because each slice is an entry slice

Disadvantages of 2D-TOF MRA

1. In plane saturation effects (discussed later)

Advantages of 3D-TOF MRA

1. Higher signal-to-noise ratio (SNR) because signal is acquired from a larger volume
2. Improved spatial resolution

Disadvantages of 3D-TOF MRA

1. 3D techniques more susceptible to saturation effects (see below)
2. Less sensitive to slow flow

PC MRA

PC MRA is based on the fact that the phase gain of flowing blood through a gradient is proportional to its velocity (assuming constant velocity). We saw in the previous chapter that phase (?) and velocity (v) are related by

? = ∫ωdt = ∫(γGvt)dt = 1/2γGvt²

Therefore, knowing the phase at any point in time allows us to calculate the velocity.

The most common method to employ PC MRA is by the use of a bipolar gradient (Fig. 27-4A). This process is called flow encoding. Because the two lobes in this bipolar gradient have equal area, no net phase change is observed by stationary tissues (Fig. 27-4A). However, flowing blood will experience a net phase shift proportional to its velocity (assuming a constant flow velocity) (Fig. 27-4B). This is how flow is distinguished from stationary tissue in PC MRA. Figures 27-5 and 27-6 illustrate the PSDs for 2D-PC and 3D-PC MRA, respectively.

Figure 27-4 A: A bipolar flow-encoding gradient. Because the two lobes have equal areas with opposite polarities, no phase change is observed by stationary spins. However, flowing spins will yield a net phase…

Figure 27-5 A PSD for 2D-PC MRA.

Figure 27-6 A PSD for 3D-PC MRA.

There are several features unique to PC MRA, as the following discussions demonstrate.

• Question: What are the “magnitude” image and the “phase” image?

• Answer: In PC MRA, you not only get an image of the blood vessels (magnitude image); you also get an image that shows you the direction of flow (phase map). The phase image would tell you whether the flow is right-left, superior-inferior, or anterior-posterior. An example is determination of hepatofugal versus hepatopedal flow in the portal vein of a patient with cirrhosis.

• Question: What is VENC?

• Answer: VENC stands for velocity encoding. It is a parameter that is selected by the MR operator when using PC MRA. VENC represents the maximum velocity present in the imaging volume. Any velocity greater than VENC will be aliased according to the following formula:

  Aliased velocity = VENC − actual velocity

  For instance, if VENC = 30 cm/sec, then a vessel with a flow velocity of 40 cm/sec will be represented as a flow of

  v = 30 − 40 = −10 cm/sec

  that is, a flow of 10 cm/sec in the opposite direction (Fig. 27-7). A smaller VENC is more sensitive to slow flow (venous flow) and to smaller branches but causes more rapid (arterial) flow to get aliased. A larger VENC is more appropriate for arterial flow. Sometimes you may image the same thing with two different VENCs—a small VENC and a large VENC—to image all the flow components accurately (an example is imaging an AVM or an aneurysm).

Figure 27-7 Two phase contrast images (phase maps) through the root of the aorta in a patient with aortic stenosis. A: VENC = 150 cm/sec; B: VENC = 300 cm/sec. Note in (A) the bright central signal (arrow…

Advantages of PC MRA

1. The capability to generate magnitude and phase images
2. Superior background suppression
3. Less sensitive to intravoxel dephasing or saturation effects

Disadvantages of PC MRA

1. Takes longer to do
2. More sensitive to signal losses caused by turbulence and by dephasing on vessel turns (e.g., carotid siphon)
3. The need to guess the maximum flow velocity in order to select an optimum VENC

2D- versus 3D-PC MRA

1. 2D techniques are faster
2. 3D techniques have better SNR

Examples of magnitude phase contrast images are seen in Figures 27-8, 27-9, and 27-10.

Figure 27-8 Coronal (A) and axial (B) thick slab (5 cm) 2D-PC MRV (magnetic resonance venogram) of the brain in normal patients.

Figure 27-9 2D sagittal phase contrast MRV (magnetic resonance venogram) (A) shows markedly decreased flow in the superior sagittal and straight sinuses (arrows) in this patient with sagittal sinus…

Figure 27-10 Axial 3D phase contrast MIP (maximum intensity projection) image of the renal vessels shows normal flow without stenosis. This is a magnitude image reflecting higher velocities in the renal…

CE MRA

CE MRA is different than either TOF or PC imaging since CE MRA is primarily dependent on the T1 properties of gadolinium in the vasculature rather than the flow properties per se. This technique has been made possible due to the advent of high performance gradients (more in Chapter 30) that permit very rapid GRE imaging and the use of the paramagnetic contrast agent gadolinium. This allows imaging to be achieved during the transit time and hence peak T1 shortening of the gadolinium. This technique is therefore very dependent on the precise timing of the arrival of the bolus of gadolinium in the vessel of interest. The plane of imaging is usually in the plane of the vessel (usually coronal) as opposed to 2D-TOF technique in which the imaging plane is usually orthogonal to the vessel of interest. Imaging this way increases coverage while maximizing resolution. Since this technique is more reliant on the T1 properties than any flow properties, it is very resistant to dephasing artifacts that are seen in some of the other techniques.

There are two principal CE-MRA techniques: elliptical-centric and multiphase. In the former, acquisition begins as contrast enters the artery of interest, filling the center of k-space (containing most of the SNR) first and the veins later (during filling of the low SNR periphery of k-space). This technique requires automatic bolus detection software (GE SmartPrep, Siemens CARE Bolus, or Philips BolusTrak) or a timing run to determine the time it takes the gadolinium to reach the artery of interest. The former is performed by placing a cursor over the upstream portion of the artery of interest; the latter is performed by injecting 2 cc of gadolinium and noting the time it takes the artery to turn maximally bright. In multiphase techniques, gadolinium is injected and multiple acquisitions are made, one of which is bound to be in the arterial phase. An additional benefit of multiphase acquisitions is the ability to provide pathologically delayed vessel filling of contrast material. But one important limitation to multiphase imaging is the tradeoff of spatial resolution for temporal resolution.

One approach to reduce loss of spatial resolution in multiphase imaging relies on the fact that much of the information to form an MR image is present in the central region of k-space. A faster time-series of 3D images can be reconstructed by acquiring a multiphase exam in which the central phase encoding values are acquired more often than the outer regions of k-space. This technique, time-resolved imaging of contrast kinetics (TRICKS), divides the 3D Cartesian k-space into several subvolumes located at increasing distance from the k-space center and oversamples the central region of k-space relative to the sampling rate of the outer regions. In this way, TRICKS is able to consistently capture an arterial phase free of venous overlay.

The resampling of the center of k-space results in a reduction in spatial resolution in TRICKS exams compared with single-image acquisitions acquired in the same scan time. Undersampled projection reconstruction has proved to preserve the spatial resolution and speed up the acquisition with limited streak artifacts. The combination of undersampled PR (projection reconstruction) acquisition in the k_x–k_y plane with a TRICKS encoding in the slice direction can significantly increase the temporal resolution without spatial resolution degradation.

Advantages

1. Rapid technique
2. Resistant to dephasing (e.g., from turbulent flow)
3. Large field of view with good resolution
4. Excellent SNR

Disadvantages

1. Dependent on timing (artifacts or venous contamination may occur)
2. Requires intravenous injection for administration of gadolinium
3. No directional information

See Figures 27-11, 27-12, 27-13, and 27-14 for examples.

Figure 27-11 Coronal source image (A) and selected MIP (maximum intensity projection) (B) of the left carotid artery from a CE MRA shows focal high-grade stenosis at the origin of the left, internal carotid…

Figure 27-12 Coronal source images (A and B) show a focal high-grade stenosis of the left proximal main renal artery (arrow in A), whereas the origin of the right main renal artery is normal (arrow in B). Note that…

Figure 27-13 TRICKS imaging of the renal artery.

Figure 27-14 Asymmetric filling of the contrast agent is demonstrated with PR-TRICKS acquisition with a high spatial resolution of 1.1 × 1.1 × 1.5 mm³ and temporal resolution of 3.8 sec per frame.

Table 27-1 contains a summary of some of the major clinical applications of the five methods of MRA discussed above.

Table 27-1

Maximum Intensity Projection

Figure 27-15 Maximum intensity projection (MIP). A channel of similar pixels in every slice is selected, and the…

We can finally explain how we can image just the blood vessels (in a way that looks 3D) and not the stationary tissue. This imaging is accomplished via an algorithm called maximum intensity projection (MIP). MIP can be used as a noun, verb (the raw data are mipped), or adjective (mipped image). Mipping is done as follows: because flowing blood in MRA techniques has high intensity, the intensity of a pixel in a slice is compared with the corresponding pixels in all the other slices (as in a channel), and the one with maximum intensity is selected. For example, pixel (1,1) in slice 1 is compared with pixel (1,1) of all other slices. This process is repeated for all the pixels in the slice. In other words, the high-intensity dots in space are connected to generate an MRA image. Thus, the mipped image represents the highest intensities (hopefully all caused by flowing blood) in the imaging volume. This image is illustrated in Figure 27-15. Obviously, a certain internal threshold is used, below which no pixel in the channel falls. An example of a CE-MRA MIP with a comparison 2D-TOF MIP is seen in Figure 27-16.

Figure 27-16 CE MRA MIP image (A) and comparison 2D-TOF MIP (B) show occlusion of the left, internal carotid artery in the same patient as Figure 27-2.

Disadvantages of MIP

The major drawback of MIP is that bright structures other than flowing blood may potentially be included in the mipped image. Examples are fat, subacute hemorrhage, and the posterior pituitary gland (Fig. 27-17). This problem is mainly with TOF MRA and not with PC MRA. (The latter is a subtraction technique based on velocity-induced phase shifts rather than on tissue T1 and T2.)

Figure 27-17 Sagittal MIP (A) image shows worrisome signal for aneurysm near the anterior communicating artery and the internal carotid terminus. Additional sagittal T1 (B) shows the high signal to…

Saturation Effects

Saturation effects refer to the gradual loss of longitudinal magnetization caused by repeated excitation radio frequency (RF) pulses. This, in turn, leads to loss of signal (and thus reduced SNR). This problem usually arises in a 2D acquisition in which flowing blood has to travel within (rather than through) a slice or in a 3D acquisition in which the blood travels through a thick imaging volume (or slab). In such a situation, saturation effects may cause the distal portion of a vessel not to be included in the image.

There are two main causes of saturation effects:

1. Decreasing TR
2. Increasing α

Decreasing TR.

As shown in Figure 27-18, a shorter TR causes less recovery of longitudinal magnetization from one cycle to the next, causing gradual loss of the M_z component. This effect is less pronounced with a longer TR.

Figure 27-18 Saturation effects. A longer TR (A, for a fixed flip angle α) causes better recovery of longitudinal magnetization than a shorter TR (B) thus reducing saturation effects.

Increasing α.

Next consider the case of a large α. A larger α causes more loss of longitudinal magnetization. Therefore, for a given TR, there is more gradual loss of M_α with a smaller α than with a larger one (Fig. 27-19).

Figure 27-19 Saturation effects. A smaller flip angle α (A, for a fixed TR) also causes more recovery of longitudinal magnetization than a larger flip angle α (B) thus leading to reduced saturation effects.

In GRE technique, saturation effects become problematic because very short TRs are used. The use of small flip angles counteracts this effect. These saturation effects become especially important in 2D or 3D in-plane flow or in 3D imaging in which volume imaging is performed over a slab, and signal losses might be significant from one end of the slab to the other. Multislice GRE techniques that use longer TRs decrease these saturation effects and allow for larger flip angles (which improves the SNR).

We have already discussed another mechanism for reducing these saturation effects: using a paramagnetic contrast agent, such as gadolinium (CE MRA). The use of this agent causes T1 shortening of blood. Consequently, the T1 recovery is much faster with fewer saturation effects (Fig. 27-20).

Figure 27-20 Saturation effects. Use of gadolinium (Gad) causes T1 shortening, which reduces saturation effects. A: Without Gad. B: With Gad.

Two other techniques are available to reduce saturation effects: MOTSA (multiple overlapping thin-slab acquisition) and TONE (tilted optimized nonsaturating excitation).

MOTSA.

Figure 27-22 MOTSA MRA of the circle of Willis shows a left parietal lobe AVM.

Motsa is a combination of 2D-TOF and 3D-TOF techniques for the purpose of reducing the saturation effects associated with a thick slab. In this method, multiple thin slabs are used, which are overlapping by 25% to 50% (Fig. 27-21). The final imaging volume is created by extracting the central slices of each slab and discarding the peripheral slices (which are more affected by saturation effects). The main drawback of this technique is the potential for Venetian blind artifact at the points where the slabs overlap. See Figures 27-22 and 27-23 for examples.

Figure 27-21 MOTSA (multiple overlapping thin-slab acquisition). To reduce saturation effects in deeper slices, multiple thinner slices are processed (in this example, 7). Next, the peripheral slices in each slab are…

Figure 27-23 The MOTSA technique can also be used for extracranial carotid evaluation, as seen in this normal…

TONE.

In this scheme, the flip angle α is increased progressively as the flowing spins move into the imaging volume by using increasing RF pulses. Recall that a larger α yields a higher SNR. Thus, increasing α counteracts saturation effects of slowly flowing blood in deeper slices. This allows better visualization of distal vessels and slow-flowing vessels. This scheme is illustrated in Figure 27-24 in which a ramped flip angle excitation pulse is used. In this example, the center flip angle is 30° and the flip angle at each end varies by 30% (i.e., 20° at the entry slice and 40° at the exit slice).

Figure 27-24 TONE (tilted optimized nonsaturating excitation). In this scheme, a ramped flip angle is used, which is larger in deeper slices. Because a larger α yields a larger transverse component in the x…

1. There are five main ways of reducing saturation effects:
   1. Decreasing the flip angle α
   2. Increasing the repetition time TR
   3. CE MRA
   4. MOTSA
   5. TONE

• Question: How does magnetization transfer (MT) affect MRA?

• Answer: Magnetization transfer (MT) saturation was discussed in Chapter 25. MT is based on suppression of the off-resonant protein-bound water protons (e.g., brain tissue). This technique, combined with TOF MRA, helps to suppress the background signal (e.g., it can reduce the signal from brain parenchyma by about 30%), increasing conspicuity of small and distal branches, vessels with slow flow, and aneurysms. MT can also be combined with TONE for further visualization of small vessels.

• Question: Why do MRA techniques overestimate the degree of stenosis?

• Answer: Because accelerated flow through the stenotic area leads to dephasing during TE. To reduce this effect, use a shorter TE. Also, turbulent flow and vortex flow (flow eddies) as well as stream separation distal to stenosis and at vessel turns (e.g., carotid siphon) may cause dephasing and flow void, overestimating the length of stenosis (in the case of post stenosis) or mimicking stenosis (in the case of vessel turns). CE MRA minimizes these effects.

Black-Blood MRA

Black-blood MRA is a subset of TOF techniques, which accentuates the TOF losses resulting in flowing blood appearing dark rather than bright. Rapidly flowing blood (arterial flow), as discussed in the previous chapter, demonstrates TOF signal losses. Slowly flowing blood (venous flow) has higher intensity. Various flow presat pulses and dephasing methods via gradients are employed in this technique to render flowing blood black. Note that the maximum intensity projection is replaced by a minimum intensity projection algorithm.

Advantages

1. This technique does not overestimate the degree of stenosis as much as bright-blood TOF MRA.
2. Dephasing in vessel turns that mimic stenosis with bright-blood TOF MRA is not a problem here.

Disadvantages

1. Calcified plaques may also be dark and therefore invisible. Thus, this technique may underestimate the degree of stenosis.
2. Other black materials (such as air or cortical bone or calcification) may mimic flow.

Fresh-Blood Imaging

Figure 27-25 Coronal FBI MIP image of the lower extremity shows abrupt…

3D-fresh blood imaging (FBI) MRA is a new technique based on the fact that vessel signal intensity is dependent on blood flow (or cardiac phase) in T2-weighted images. The early systolic phases (0 ~ 200 msec after the R wave) show low signal intensity for arteries (high velocity with TOF losses) and high signal intensity for veins (lower velocities with no significant TOF loss), whereas the diastolic phases (400 ~ 600 msec after the R wave) show high signal intensity for both arteries and veins (no significant TOF loss). Bright-blood MRA is achieved by subtracting systolic images from diastolic images. The 3D-FBI method employs an electrocardiography (ECG)-gated 3D half-Fourier fast spin-echo sequence triggered for systolic and diastolic acquisitions. The ECG triggering time is an important factor influencing the blood signal intensity in the vessel of interest. An “ECG-prep scan” is typically used to produce 2D half-Fourier FSE single-slice images at various triggering delay times. An appropriate ECG delay time is determined for the vessel of interest and later applied in the 3D half-Fourier FSE acquisition synchronized by ECG gating for every slice encoding. An example is seen in Figure 27-25.

Advantages

1. Non-contrast technique; gadolinium not required
2. Arterial and venous imaging in all areas of the body
3. Insensitive to off-resonance artifacts