3–tesla body imaging offers diagnostic promise

Technology once reserved for brain imaging reveals higher resolution in multiple applications

By Robert E. Lenkinski, Ph.D., and Neil M. Rofsky, M.D.

Signals in MRI are extremely weak, as they emanate from a relatively small number of protons. Signal–to–noise ratio (SNR) is dependent on a number of fundamental parameters such as in–plane resolution (or voxel size), number of excitations (or signal averages), proximity of receiver coils, and field strength.

Improving SNR has been a major goal since the inception of MRI, and we have taken a number of steps in pursuit of this goal. We have routinely increased the number of excitations (NEX) for some applications, since random contributions of noise will cancel each other out over repeated excitations. We sometimes sacrifice spatial resolution by sampling larger volumes. And we have witnessed continual developments in surface coils, which restrict the amount of anatomy contributing to noise.

The need for increased SNR has also been the catalyst for the development of 3–tesla scanners. Theory predicts, and practice confirms, that 3–tesla scanners provide a factor–of–two increase in SNR over 1.5–tesla scanners.

This advantage has historically been applied primarily to imaging of the brain. General consensus is emerging, for example, that functional MRI (fMRI) studies using the blood oxygen level–dependent (BOLD) effect to monitor the effects of externally applied sensory stimuli are optimally performed at 3 tesla. Proton MR spectroscopy of the brain has also been shown to be superior at this field.

In contrast, the application of 3 tesla to whole–body imaging has remained largely unexplored and underappreciated, for technical reasons. Conventional wisdom holds that 3 tesla’s utility for these applications might be severely limited by imaging artifacts and radio–frequency power deposition issues. Furthermore, body imaging presents special challenges. To minimize cardiac and respiratory motion artifacts, for example, we must be able to image very quickly—ideally, in a single breath–hold—which limits the NEX we can prescribe. In addition, we can’t always afford to compromise resolution in body applications by sampling larger volumes.

The technical problems on which these concerns were based have since been overcome, paving the way for a thorough exploration of this technology’s potential. In the summer of 2000, scientists from GE Medical System’s corporate R&D developed a 3–tesla RF body coil, creating opportunities to evaluate high field’s impact on MR imaging outside of the brain. This body coil, which was described at the April 2001 meeting of the International Society for Magnetic Resonance in Medicine (ISMRM) in Glasgow, Scotland, is 56 cm long and has a 55-cm–diameter bore.

Addressing Challenges

The results to date have been outstanding. We have found that we can invest the abundance of signal produced by a 3–tesla scanner in the most appropriate way for the study at hand–in resolution, in scan speed, or in both. We can tailor our parameters to address the challenges of each study to achieve the best balance of image quality and acquisition speed.

The axial liver images in Figure 1 are an example. The spatial resolution and signal to noise are superior to those obtainable at 1.5 tesla. Note the relative signal uniformity in these images.

The availability of an RF body coil also permits the use of phased–array, receive–only surface coils. An example of the image quality achievable at 3 tesla using an external torso array for imaging the prostate is shown in Figure 2. Note the excellent visualization of the peripheral zone of the prostate on the axial view. This image clearly demonstrates the potential of 3–tesla imaging of the prostate.

We have also obtained high–resolution MR images of an ex vivo prostate specimen (Figure 3). The excellent image quality suggests that even higher resolution images are possible.

These results indicate that we can seriously improve our accuracy in staging prostate cancer at 3 tesla. Just as exciting, in these applications we achieve image quality comparable to what was previously available at 1.5 tesla only with the use of an endorectal coil. If we can eliminate the coil, we can improve patient tolerance and compliance, simplify the study, and shorten the overall exam time. By reducing exam time, we can entertain the possibility of using MRI to screen patients for prostate and pelvic disease. If higher resolution is necessary for accurate MR staging, a 3–tesla endorectal coil can provide even higher signal to noise than the images shown here.

Figure 4 shows an example of 3–tesla images of the wrist. Great potential exists for 3–tesla imaging in a number of other body applications, thanks not only to improvements in spatial and temporal resolution, but also to its ability to decrease acquisition time by reducing NEX.

It holds promise for vascular and cardiac applications, for example, including renal and peripheral–vascular (PV) angiography. Renal angiography acquisition speed at 3 tesla may allow us to achieve full quantification, for a much more precise evaluation. This speed also provides us with an excellent opportunity to add physiological adjuncts to our structural analyses. This is an intriguing proposition, since renal angiography is the most accurate way to diagnose the only potentially curable form of renal hypertension, which is probably an underestimated cause of renal insufficiency. If we can detect this disease earlier, before the kidney is severely damaged, we may be able to forestall or eliminate the need for dialysis in certain patients.

The scan–time reductions with 3 tesla may also permit a substantial increase in the use of spectroscopy to characterize lesions and, potentially, to reduce the need for biopsies. A conventional body exam that includes spectroscopy can take up to 80 minutes at 1.5 tesla. At 3 tesla, the same exam could potentially be completed in just 20 minutes.

Nor is scan speed 3 tesla’s only potential contribution to spectroscopy. At 1.5 tesla, differentiating between chemical constituents can be difficult. Higher field strength also results in better peak separation, which makes peak differentiation much more straightforward. At the same time, the increase in SNR allows us to improve spatial resolution. As a result, we’re beginning to characterize smaller lesions than ever before. We are also beginning to explore a number of other avenues for the use of 3 tesla in body imaging: evaluating duct abnormalities in the pancreas, screening for lung cancer, detecting avascular necrosis, and examining patients who come to the ER with nonspecific abdominal pain to rule out appendicitis or bowel obstruction. If we can conduct what were time–consuming studies like these in just 15 minutes, without delivering any radiation or requiring contrast injection, we see the possibility of creating a new diagnostic paradigm.

We also expect 3–tesla body imaging to make major contributions to surgical planning, particularly when combined with three–dimensional reconstructions. In these cases, we may invest some of our extra signal in speed to refine our 3–D imaging techniques, or concentrate on improvements in resolution to more precisely define the margins and borders between normal and pathological tissues.

In some ways, our experience with 3–tesla MRI has given us a strong sense of deja vu. We remember the skepticism with which the medical community awaited the first 1.5–tesla MR scanners in the early ‘80s—and the enthusiasm that the first 1.5–tesla clinical images inspired when they were published in 1982.

We believe we are seeing the same skepticism at work today among those who believe 3–tesla imaging will never be useful for much more than neuro applications. But those who have had an opportunity to see the images produced by this scanner exhibit the same enthusiasm that early 1.5–tesla systems generated, and they have come away convinced that the case for 3–tesla body imaging may be even more compelling than for neuro applications.

We have found that the availability of a whole–body RF coil at 3 tesla creates enormous opportunities for imaging outside the head. These applications, which include prostate, female pelvis, cardiac, and spine imaging as well as MR peripheral angiography, may all benefit from the increase in signal to noise available at 3 tesla. While it is important to evaluate each of these potential applications in patients, it is clear that 3–tesla MR scanners may ultimately provide diagnostic improvements in a wide range of MR studies.

Dr. Lenkinski is a professor of radiology at Harvard Medical School and director of 3–tesla MR at Beth Israel Deaconess Medical Center. Dr. Rofsky is an associate professor of radiology at Harvard Medical School and director of MRI at Beth Israel Deaconess Medical Center in Boston.

Back to Top