Applied Radiology

Dr. Liu is the Section Head for Body and Musculoskeletal MRI, Department of Radiology, Mayo Clinic, Scottsdale, AZ.

Many recent publications have demonstrated that magnetic resonance angiography (MRA) can be an accurate, fast, and safe tool in the workup of patients with lower extremity ischemia. ^1-6 These studies have proven that MRA can reliably and quickly depict significant stenoses of the abdominal aorta and the arteries of the lower extremities (LE). In some medical centers, MRA has even replaced conventional X-ray angiography (XRA) as the initial method of evaluating LE ischemic symptoms.

Setting up an LE-MRA protocol, however, can be a complex task due to the many proposed sequences, injection protocols, and postprocessing techniques. To simplify this process, this article will detail and explain the methods we have developed for use at our institution for performing LE-MRA. This protocol provides a relatively straightforward technique that can be applied to a wide range of patients.

Imaging challenges

The major challenges faced by technologists and radiologists attempting to perform MRA of the abdominal aorta and LE include the need to quickly cover a relatively large region of interest, the desire to achieve high spatial resolution, and the need to make multiple measurements and calculations during scan set-up. In addition, the limitation on the maximum contrast dose and the wide variability of circulation time have made it difficult to optimize MRA parameters for the entire region of interest. To meet these challenges, we have developed a protocol that employs a fast three-dimensional (3D) contrast-enhanced MRA (CE-MRA) sequence with a dual-injection modified stepping-table protocol, standardized fields of view, and an extremity vascular phased array coil.

Choosing the best MRA sequence

The most successful LE-MRA protocols call for the use of a 3D gadolinium-enhanced, T1-weighted spoiled gradient- echo (SPGR) sequence. Studies have shown that the enhanced 3D SPGR sequences outperformed two-dimensional (2D) time-of-flight sequences for both speed and accuracy. ^1,7

As is seemingly customary, the major MR vendors have given this SPGR sequence a variety of unrelated names: enhanced fast gradient-echo (EFGRE) on General Electric (GE Medical Systems, Waukesha, WA), fast low angle shot (FLASH) on Siemens (Siemens Medical Solutions, Iselin, NJ), and T1-weighted fast field echo (T1W-FFE) on Philips (Philips Medical Systems, Bothell, WA). The SPGR sequence bears similarities to the more familiar T2*-weighted gradient-echo sequences, using a low flip angle and gradient refocusing pulse to create an echo, but differs by the application of a radiofrequency spoiling pulse prior to every alpha excitation pulse. This spoiler pulse destroys the residual transverse magnetization, essentially resulting in a T1-weighted sequence.

Administration of IV gadolinium greatly enhances vascular signal due to its T1 shortening effects. Using short TR and TE settings (5 msec and 1.8 msec) minimizes scanning time and background signal. The optimal flip angle for maximizing arterial signal-to-noise ratio (SNR) is 30š to 60š, depending on the intra-arterial gadolinium concentration, with 45š providing a good compromise. ⁸

The choice of a 3D over a 2D acquisition permits the use of thinner slices while maintaining good SNR. Since information for all voxels in a 3D dataset is acquired throughout the entire length of the sequence, better volume reconstructions can be created. Minor patient motion results in less prominent artifacts than is seen with the 2D mode. While the choice of the 3D acquisition mode increases scan time slightly, a high-resolution 3D CE-MRA sequence can still be performed in 20 to 30 seconds. These qualities make this sequence ideal for use in breath-hold and stepping-table MRA.

Optimizing vascular SNR with timing

The addition of small bolus timing tests, power-injectors, automated bolus detection methods, and now MR fluoroscopy has been found to be very helpful at accurately identifying circulation time from the injection site to the imaging slab. Circulation time to the abdominal aorta has been shown to vary from 10 to 60 seconds in one group of subjects. ⁹ If the circulation time of the bolus can be pinpointed, arterial SNR can be optimized in CE-MRA by timing the acquisition of the central points of image k-space with this peak enhancement. These central or low order points of k-space comprise the subset of the images' raw data that determines image contrast.

Centric, elliptical centric, low-high, and spiral k-space acquisition strategies were all recently applied to MRA to acquire the image contrast information rapidly at the beginning of the sequence, so that arterial MRA signal can be optimized and venous signal minimized. If one of these scan options is available, it should be used during acquisition of MRA images of the aorta and LE because of the short time window between the arrival of the bolus and the appearance of venous contrast. With centric phase-encoding techniques, venous filling occurs in the later parts of the MRA sequence, when the peripheral points of k-space are being acquired. These peripheral points contain the image's spatial resolution data, so venous filling does not usually cause significant contrast enhancement on the resulting images. ¹⁰

Stepping-table technique

Optimal enhancement of the abdominal aorta on CE-MRA can be achieved with as little as 0.1 mmol/kg gadolinium, when injected at 2 mL/sec, in patients with normal body habitus and cardiac output. However, for stepping-table CE-MRA of the aorta and lower extremities, a standard dose of 0.3 mmol/kg, up to the regulatory maximum of 40 mL, has been proposed for use in all patients. ⁸ This higher dose is needed to allow for the unavoidable dilution of the contrast bolus in those patients with lower cardiac output, and it better opacifies the smaller vessels in the legs.

In stepping-table MRA, the table automatically translates from one imaging station to another, chasing a contrast bolus as it travels down the aorta and passes into the legs. In most patients, the inflow and outflow vessels, from the suprarenal abdominal aorta down to the pedal arches, can be imaged with three 45-cm imaging stations overlapped by 5 cm. We term these three regions the aorto-iliac, femoral-popliteal, and tibial-pedal stations (Figure 1). While many authors have limited the coverage of their MRA exams to exclude the upper abdominal aorta or the feet, these are regions that frequently need to be evaluated if surgical bypass grafting is planned.

Most stepping-table MRA protocols have called for a single bolus injection of 40 mL of gadolinium for imaging the aorto-iliac station, followed by automated table stepping and scanning of the femoral-popliteal station, then stepping to the tibial-pedal station. Bolus-chase MRA studies performed with this stepping-table, single-injection technique generally provide excellent images of the aorto-iliac and femoral-popliteal stations, but have encountered occasional problems with asymmetric contrast filling in the tibial-pedal arteries. ¹¹ Depending on the circulation time to the lower legs, and the amount of time spent scanning the first two stations, the contrast enhancement peak might not reach the tibial-pedal station arteries coincident with the timing of the third station MRA acquisition.

Not surprisingly, these problems with contrast timing in the lower legs parallel the contrast timing difficulties found in conventional X-ray angiography runoff studies that use a bolus-chase, stepping-table technique. If there are proximal occlusions or low cardiac output, then underfilling of the lower leg arteries will occur. Selective injections with digital subtraction can provide better angiograms in these cases.

With distal arterial underfilling on bolus-chase MRA studies, the tibial-pedal vessels may have underestimated diameters, and stenoses may be overestimated. On the other extreme, if the imaging occurs too late after the injection, then the rapid venous return from the feet will likely obscure some of the ankle and lower leg vessels. Multiphase imaging at one anatomical station can provide more consistent results in these cases.

Two-injection strategy

Recently, MRA research has explored methods of improving imaging of the lower leg arteries. Biphasic contrast injection protocols have been studied with stepping-table MRA in an attempt to prolong the duration of peak arterial enhancement with some success. Other successful solutions used a two- or three-injection stationary table strategy, using smaller doses of contrast for each station. Newer, improved MR imaging sequences and bolus timing techniques have reduced the dose of contrast needed to achieve good arterial SNR. ^2,11-13 It is possible to divide the contrast bolus into two or three smaller boluses, with separate injections for the various imaging stations. Digital subtraction postprocessing is then used to optimize arterial conspicuity. This technique is analogous to a conventional angiographer's use of stationary table digital subtraction angiography (DSA), since the multiple divided bolus MRA image of the lower legs can use more optimized timing and multiphase scanning for delayed unilateral arterial filling. Some authors have termed this an MR-DSA technique.

We have chosen to combine elements of the MRA-DSA and stepping-table techniques by using a two-injection method with automated table stepping to image the aorto-iliac station and the femoral-popliteal station in rapid succession after the first contrast bolus. After a brief 5-minute delay, a single station injection is used with a higher resolution technique for optimized scanning at the tibial-pedal region. A sample case performed with this technique is shown in Figure 2.

Receiver coil choice and positioning

The patient's legs and pelvis are imaged with a lower extremity vascular coil, several of which are available commercially. One of these uses a phased-array coil for the legs and a quadrature coil for the pelvis (Peripheral Vascular Coil 675GE, IGC-Medical Advances, Milwaukee, WI). Both the quadrature and phased-array designs increase SNR compared with the body coil and permit scanning at higher resolution without sacrificing SNR or scan time. This added resolution proves especially helpful when attempting to distinguish severely stenotic vessels from complete occlusions. The length of the coil allows one to acquire high-SNR images from the pelvis to the feet without stopping to reposition the patient or coil.

To image the abdominal aorta and include the renal artery origins, however, we need to use the body coil. The peripheral vascular coil does not reach above the pelvis, and the large calibers of the aorta and iliac arteries do not require the extra SNR afforded by the peripheral vascular coil.

If a dedicated peripheral vascular coil is not available for imaging the legs, the body coil can also be used to image the femoral-popliteal station during the first injection bolus, and the torso phased-array coil used for imaging the tibial-pedal station. If the coil is rotated to orient its long axis in the craniocaudal direction, then the field of view can reach 40 to 45 cm in the craniocaudal dimension.

The ankles are partially plantarflexed and elevated with approximately 4 inches of padding to bring the lower leg vessels parallel to the imaging table. These maneuvers decrease the number of slices needed for MRA coverage of the lower leg and midfoot arteries.

Scout series prescription

Because aorto-bifemoral or aorta-iliac bypass procedures sometimes involve reimplantation of the renal arteries, we attempt to image the aorta from the renal artery origins to the deep plantar arches of the feet. Distally, visualizing the plantar arches is important because delivery of blood to these arches is essential for adequate perfusion of the forefeet and toes.

By starting the scan set-up at the feet, we can ensure that the deep plantar arches will be included in all examinations. Standardized fields of view and coil placement help to simplify the sometimes tedious set-up of the multiple imaging stations. The inferior margin of the vascular coil is positioned at the MTP joints of the feet in order to include the deep plantar arches.

A mark placed on our coil at a level 22.5 cm cephalad to its inferior margin is used to "landmark," or "zero," the position of the imaging table. Therefore, a 45-cm field of view coronal slab centered at this zero level will reach from the deep plantar arch to the popliteal arteries in most patients. The femoral-popliteal and aorto-iliac station 3D slabs, which are also imaged with 45-cm FOV, will be centered 400 and 800 cm cephalad to the first landmark to insure that the three slabs each overlap by 5 cm (Figure 1). The use of these standardized imaging stations will ensure easy set-up for imaging up to the renal arteries in all but the tallest patients.

A coronal T1-weighted spin-echo scout series of the abdomen and pelvis is performed to obtain a black-blood image of the abdominal aorta. These images will be needed to align the timing test sequence or bolus-tracking scan through the often-tortuous aorta. Axial bright-blood, gradient-echo scouts are then obtained through all three imaging stations, and the anterior and posterior locations of the pertinent arteries determined.

Contrast bolus timing

We use a sagittal or axial multiphase, spoiled gradient-echo timing test with a gadolinium mini-bolus to determine circulation time from the antecubital vein to the distal abdominal aorta. A 1 mL bolus of gadolinium is injected at 1 mL/sec, followed by 20 mL of saline flush at the same rate. MR-fluoroscopy techniques, now commercially available, can also be used to provide accurate, real-time visualization of the arrival of the contrast bolus in the aorta.

Scanning

For each 3D coronal slab, 30 to 36 slices are prescribed with a maximum thickness of 3.0 mm. Zero-fill interpolation is applied in the slice dimension to create a second stack of 3.0 mm thick slices, offset from the original stack by 1.5 mm in the slice dimension. This interpolation effectively increases the resolution of MIP images reconstructed in a perpendicular plane. Scan matrix is 256 x 160 in the aorto-iliac and femoral-popliteal stations and 256 x 192 at the tibial-pedal station. Zero-fill interpolation is also applied in the matrix dimension to double the effective in-plane resolution of the source images, with no added time and only a small loss of SNR. Field of view is 45 * 36 cm for the two upper stations and 45 * 27 cm for the lower station. Receiver bandwidth is set to 62 kHz, which strikes a good balance between imaging speed and SNR.

At the aorto-iliac and the femoral-popliteal stations, mask precontrast series are obtained and reviewed to confirm appropriate slab prescription. The mask series of the second station will be used during postprocessing for image subtraction. We instruct the patient to remain motionless to decrease subtraction artifacts. Fortunately, the securely fastened peripheral vascular coil serves the added function of immobilizing the patient's legs. The first 20-mL gadolinium bolus is injected at 1 mL/sec, followed by a 20 mL saline flush at 1 mL/sec. At a predetermined delay time (circulation time minus 5 seconds) the patient is instructed to hold his or her breath. The arterial phase is scanned, followed immediately by automated table stepping to the second station. Immediately, the femoral-popliteal station is scanned; breath holding is not necessary at this point. In case there is delayed filling of aneurysmal or collateral vessels, these scans are repeated automatically at both imaging stations to acquire venous phase images.

An interval of 10 minutes between the first and second contrast injections is imposed to allow for clearing of the gadolinium bolus from the vascular bed and muscles before imaging the final tibial-pedal station. During this waiting period the multiphase timing test is repeated in the coronal plane, centered at the popliteal arteries, using a 2-mL mini bolus for the smaller vessels.

Since no breath-holding is required in the lower leg station, scan speed is not as important and can be sacrificed for the purpose of increased spatial resolution. We choose to prescribe an increased number of thinner slices than we used in the more proximal stations, in order to increase spatial resolution of the smaller distal vessels. Subtraction of a mask scan series from the arterial phase scan will eliminate any persistent venous or soft-tissue enhancement. For the single station imaging of the lower legs, the second 20-mL gadolinium bolus can be injected at 2 mL/sec, for a higher peak vascular contrast level.

Non­stepping­table MRA

If an automated stepping table MRA is not available, then a three-station, three-injection protocol can be performed with a few small changes to this protocol. The dose for each contrast bolus will need to be decreased to 0.1 mmol/L for each of the three bolus injections and delay periods will have to be planned between each station for contrast clearance.

Post-processing

MRA postprocessing software is now more user friendly and faster. From a 3D MRA data set, our technologists routinely perform image set subtraction and the creation of maximum intensity projection (MIP) images in approximately 10 minutes.

Subtraction of a precontrast mask series from the arterial phase source images improves visualization of small caliber vessels (Figure 3) and can also reduce the dose of contrast needed to obtain good arterial SNR. ^14-17 If the patient's breath-holding is consistent from the mask series to the arterial phase images, mask subtraction may be helpful in the aorto-iliac station to eliminate soft tissues, any bright bowel contents, and aliasing (phase wraparound artifact).

In the femoral-popliteal and tibial-pedal stations, we routinely perform mask subtraction in all cases to decrease soft-tissue and venous signal. If a phased-array coil is used in these lower stations, then subtraction becomes essential for reduction of high near-field signal at the skin and subcutaneous fat, where fat saturation prepulses are often not completely effective. This bright signal can greatly decrease image quality by obscuring underlying vessels on the reconstructed MIP images.

MIP reconstructions have become the most accepted mode for presentation of 3D MRA data sets. With the routine use of mask subtraction and subvolume coronal MIP reconstructions, there is little background signal on our routine MIP images. We have found that there is virtually no need for tedious manual tracing or cutting of unwanted tissues from the 3D MRA datasets since image subtraction is so effective at removing background signal.

Our technologists now perform a standardized postprocessing routine at the scanner console, almost entirely obviating the need for physician postprocessing. For each arterial phase 3D data set, whole-volume rotating MIPs are reconstructed at 15š intervals around a 180š arc, and a set of subvolume coronal MIP reconstructions is prescribed with 10-mm slice thickness, overlapping at 5-mm intervals. These whole-volume and overlapping subvolume MIPs are viewed on a workstation in a cine-mode to take advantage of the 3D properties of the data set. The use of a 10-mm subvolume MIP slab thickness removes any overlapping soft tissues while being thick enough to display moderately long segments of vessels (Figures 4 and 5).

Interpretation pitfalls

Because any overlapping structure with moderate or high signal intensity can result in a false narrowing of an artery on a MIP reconstruction, the radiologist should be acutely aware of this possibility for creation of a pseudostenosis. As stated above, subtraction removes many of these overlapping artifacts, but examination of subvolume MIPs and source images is necessary when a stenosis or occlusion is suspected in order to exclude such pseudostenosis artifacts.

In addition, metal prostheses, stents, and surgical clips can create signal void artifacts that can cause false stenoses on MIP images. Review of previous angiograms, plain X-rays, and the MRI source images should reveal these metal devices as the sources of the signal loss.

Conclusion

Recent advances in commercially available MR hardware and software have made peripheral MRA faster and more automated, but the many variables that must be determined by the operator can make setting up the procedure a complicated task. When one takes into account the anatomic and disease variability from patient to patient, it becomes obvious that a single, robust strategy is needed if the exam is to run efficiently. The MRA protocol presented here uses generous fields of view and relatively large contrast doses, in order to provide allowances for many of these variables without sacrificing image quality. With practice a complete MRA study of the abdominal aorta and lower extremities can be performed in approximately 30 minutes using this protocol, with an additional ten minutes allowed for postprocessing time.

Aknowledgment

Thanks to Joel Brower, RT, RMR for assistance with creating and optimizing this protocol.