Contrast-enhanced MR angiography (CEMRA) is becoming increasingly competitive with both catheter angiography and CT angiography (CTA) for nearly all vascular territories, with the possible exception of native coronary arteries. Over the past 5 years, steady improvements in gradient technology, pulse sequences, and postprocessing algorithms, together with dramatic improvements in radiofrequency (RF) technology, have laid the foundation for the robust current status of CEMRA applications at 1.5T. Ironically, it may be the very success of 1.5T imaging that has spurred the search for newer, less well explored fields. Recently, whole-body 3T MRI systems have become available, with the promise of greatly improved signal-to-noise ratio (SNR) in comparison to 1.5T. All else being equal (which, of course, it is not), noise diminishes linearly with field strength; 3T imaging might be expected to double the available SNR over 1.5T. Increased SNR is never a burden, but it can be an expensive commodity if it comes with too many strings attached. Going from 1.5T to 3T involves more than increasing SNR; it also awakens sleeping giants from the world of physics, and some of these pack a nasty punch. For some techniques and pulse sequences (many of which are now bread and butter for nonangiographic imaging at 1.5T), there are substantial trade-offs at 3T. The single most troublesome feature seen by 3T users is the so-called dielectric resonance effect, a phenomenon whereby troughs of MRI signal are seen corresponding to the shorter wavelength of the RF signal at 3T. The signal loss occurs at fairly consistent locations in the body, and it is most troublesome for highly RF intensive techniques. Ingenious plans are being pursued to tackle this problem, and for CEMRA, dielectric resonance effects are, at least so far, not very limiting.

Why do we want higher SNR for CEMRA? Is the current SNR inadequate? In most cases, the answer is no, and if it were not for the introduction of parallel acquisition, 3T might be a hard sell for CEMRA. Parallel acquisition speeds up an MRI measurement by using spatial information from surface coil elements to substitute for a certain percentage of the phase-encoding steps (time-consuming repetitions that have been the basis for spatial encoding for more than 20 years). With the appropriate coil arrangement and receiver chain, it is possible to accelerate an acquisition manyfold, with the specific number being called the acceleration factor. The price paid for the increased speed of parallel acquisition is a drop in SNR, and this can be severe. At 3T, there is more SNR available, so one can use higher acceleration factors and still have adequate SNR. During the same amount of imaging time, users can acquire images with higher spatial resolution and/or greater coverage than would be the case at 1.5T (where SNR breakdown is possible).

Another feature of CEMRA at 3T is that, since the longitudinal relaxation time (T1) of background increases with field strength, sensitivity to injected gadolinium agents for CEMRA is heightened, and smaller contrast doses may be used. CEMRA at 3T can often generate spectacular studies, providing three-dimensional (3D) data with high spatial resolution over a large field of view.

CEMRA TECHNIQUES

Clinical experience with 3D CEMRA techniques at 3T, as implemented on the Siemens MAGNETOM Trio scanner, currently equipped with 32-channel multicoil technology called Total imaging matrix (Tim), covers a variety of vascular territories.

Conventional CEMRA relies on the T1-shortening effect of gadolinium-based contrast agents and is performed with a T1-weighted fast spoiled 3D gradient recalled-echo sequence. The sequence parameters and the contrast-administration scheme should be carefully planned to achieve the best compromise between the expendable SNR and the required spatial and temporal resolution. With recent advances in scanner gradient performance, temporal resolutions on the order of 2.5 to 3 milliseconds and echo times on the order of 1.2 milliseconds are achievable for 576 matrix acquisitions. This results in acquisition of a highspatial-resolution 3D data set during a comfortable breath-holding period.

A unique feature of MRI is the ability to generate temporally resolved 3D images that depict the first-pass transit of contrast through the vascular system with subsecond temporal resolution. This type of measurement is impossible with CTA because the radiation dose would be prohibitive. Time-resolved MRA can readily provide supplemental functional information about cardiovascular hemodynamics, with relative insensitivity to motion and the requirement for only very small doses of contrast. For many applications, in-plane resolution can be preserved while through-plane resolution is traded for rapid temporal sampling.

For all CEMRA applications, precise control of contrast administration is essential, especially for highspatial-resolution MR angiography, where the center of k-space should be aligned with the peak vascular enhancement. Timing can be easily optimized through use of a test bolus, but real-time triggering algorithms provide an alternative, if less flexible, method.

RECENT ADVANCES

Figure 1. Coronal and sagittal oblique maximum-intensity projection from contrast-enhanced MR angiography shows tight stenosis of the left carotid bifurcation (arrow) and mild irregularity along the right common and internal carotid arteries (voxel dimensions: 0.7x0.7x0.7 mm3 in a 20-second breath hold over a 440-mm field of view).

The introduction of parallel imaging is one of the most significant recent advances in MRI, and it has improved the performance of MRA applications by changing the way that data are spatially encoded. Component coil signals in an RF coil array are used for partial encoding of spatial information by substituting for phase-encoding gradient steps that have been omitted. Therefore only a subset of the k-space data, defined by the acceleration factor, is sampled; the whole data set is reconstructed afterward. The major drawback to parallel acquisition is that SNR is diminished, and this represents a fundamental challenge as acceleration factors are increased. Among the strategies to counteract the SNR loss of parallel imaging are the use of stronger magnetic fields and improvement and adjustment of array coil geometry and sensitivity. The recently-introduced Trio with Tim system has the potential to improve CEMRA applications by providing a firm base to support aggressive acceleration factors.

The integrated multicoil arrays have better sensitivity profiles for parallel acquisition than those available previously, with the promise of increased CEMRA performance in a 32-channel receiver chain. Multiple arrays of RF receiver coils, with associated multiple RF receiver-channel electronics, combine for more effective parallel acquisition strategies. The appropriate use of parallel acquisition can result in improved CEMRA performance to increase coverage, speed, or spatial resolution, in any combination.

CLINICAL APPLICATIONS

3D CEMRA has been established as the method of choice in evaluation of the craniocervical vasculature for a variety of conditions (including atherosclerotic arterial disease, aneurysms, and arteriovenous malformations) and for the presurgical assessment of tumors. At UCLA Medical Center, we have used both time-resolved and highspatial-resolution techniques for craniocervical CEMRA. Figures 1 and 2 show examples of head and neck CEMRA acquired using the Trio with Tim system.

Figure 2. Coronal maximum-intensity projection from time-resolved contrast-enhanced MR angiography shows early arterial enhancement of the right thyroid lobe (data were acquired with in-plane resolution of 1x1 mm and temporal resolution of 1 second, after injection of only 5 mL intravenous contrast). Coronal turbo spin echo and postcontrast images using longitudinal relaxation time fat suppression through presaturation show nodular enlargement of the right thyroid lobe.

Clinical applications in the pulmonary circulation include evaluation of pulmonary embolism, pulmonary hypertension, and congenital heart disease, along with pulmonary venous mapping. In the abdominal vasculature, clinical applications are assessment of atherosclerotic arterial disease, aneurysms, and aortic dissection; preoperational assessment of tumors; and evaluation of abdominal veins. In our practice, MR venography is a rapidly growing application.

At 3T, multistation whole-body CEMRA can be performed with highspatial-resolution data sets (submillimeter voxels). By integrating parallel imaging, an appropriate contrast-injection protocol, and flexible table movement, venous contamination can be minimized or avoided. The procedure is feasible and convenient for both patient and technologist, eliminating the need for coils or patient repositioning. Figure 3 shows an example of whole-body MRA acquired using the Trio with Tim system.

Figure 3. Coronal maximum-intensity projection images from four-station whole-body contrast-enhanced MR angiography at 3T acquired using a Trio with Tim system showing almost the entire vascular tree with submillimeter voxel size. The data were acquired in four separate image-acquisition steps following separate two-step contrast-injection phases.

SUMMARY

A wide spectrum of vascular disease processes can be depicted using 3D CEMRA at 3T. Our experience, to date, suggests that CEMRA at 3T is robust and frequently yields spectacular images using lower contrast doses than are typical at 1.5T. We now use 3T routinely for CEMRA of the carotids, thorax, abdomen, and pelvis, and have found, so far, that patient acceptance is similar at 1.5T and 3T.

It is still too soon to comment definitively on patient outcomes because experience is far more limited at 3T than at 1.5T. To date, however, we know of no cases at our institution in which CEMRA at 3T has provided misleading or false information. We also know of many cases in which vascular findings, evaluated in very fine detail, were confirmed by conventional angiography or during surgery.

It is also too early to comment on cost-benefit issues at 3T. There is no link between field strength and reimbursement level; unless this changes, it will remain more expensive to run a service at 3T than at 1.5T.

Q&A with Dieter Enzmann, MD

Just more than one year ago, the University of California Los Angeles Medical Center's department of radiology acquired a 3T Trio MR from Siemens. To discover what the university has accomplished with it since then, Decisions in Imaging Economics interviewed Department Chairman Dieter Enzmann, MD. Imaging Economics: Has the introduction of 3T impacted practice patterns at UCLA? Enzmann: Allow me to clarify up front that our 3T machine is at this juncture used only for research and not for clinical purposes. There is clear evidence that 3T offers a higher signal-to-noise (SNR) ratio. However, because there are magnetic susceptibility effects with 3T, modifications are necessary in the pulse sequences to take advantage of the increased signal-to-noise ratio. Our research is attempting to determine how best to adapt the pulse sequences, contrast injection, and various other parameters. We are evaluating whether new approaches to RF and coil design would be helpful. With regard to your question, we are also exploring at what point the SNR becomes a key determinant of image quality and whether that will make a clinical difference. Much of this has already been shown for neuro applications; our research is primarily geared to see how well 3T works in cardiac imaging and vascular imaging. Imaging Economics: Within your department, has the introduction of 3T resulted in rethinking the diagnostic imaging protocols for any particular disease states or conditions? Enzmann: We are finding image quality with 3T is much improved compared to 1.5T, and that is perhaps most evident in MR angiograms where we are able to obtain never-before-seen vessel detail. As a result, when high-end angiography is called for, when we really need to have an accurate rendition of small vessels, we are now convinced that patients should if possible be steered to 3T. Imaging Economics: Are there any circumstances in which cardiac MRA should displace catheterization? Enzmann: It has already replaced a lot of what I will call conventional diagnostic angiograms; 3T will probably complete the progression. Imaging Economics: Faster scan times are reportedly possible with 3T. What are the economic implications of such speed? Enzmann: It potentially means improved throughput, but by how much will depend on multiple factors, including patient handling. This is an aspect we have yet to investigate. Imaging Economics: How is your radiology department handling the rollout of 3T? Enzmann: It's a slow, methodical process. We test a new protocol, and then, if it looks like it might be of clinical benefit, we offer it to patients. As we began to move out of the research stage, the first clinical uses were in cardiovascular applications. Right now, we are continuing to develop and evaluate new pulse sequences, and we recently installed a new set of coils for testing [as part of the Tim upgrade, Siemens matrix coils were recently delivered to UCLA]. We are constructing two new hospitals, and each will have access to at least one 3T machine how rapidly that occurs will depend on how soon those new buildings are completed. Imaging Economics: Are there any obvious turf implications associated with the greater field strength? Enzmann: [wryly] Nothing that wasn't there already with 1.5T.

—R. Smith

Kambiz Nael, MD, is radiologist; Gerhard Laub, PhD, is physicist; and J. Paul Finn, MD, is professor of radiology and medicine, chief, diagnostic cardiovascular imaging, and director, magnetic resonance research, Department of Radiology, University of California, Los Angeles, Medical Center.