The promise of 3T MRI to be a powerful successor to clinical 1.5T platforms is steadily being realized through technical refinements that overcome its shortcomings and exploit its strengths. Problems remain, but progress is apparent, according to reports during the 2006 International Society for Magnetic Resonance in Medicine meeting.
Doubling field strength-and price-has not translated into doubling MRI's signal-to-noise ratio and increasing scanner speed resolution, as equipment vendors promised when they introduced 3T scanners for routine clinical use, according to Axel Haase, Ph.D., a professor of physics at University of Wurzburg in Germany. Unexpected performance issues, such as a higher specific absorption rate (SAR) and susceptibility and chemical shift artifacts, have stood in the way. Doubled field strength has thus far yielded about a 20% increase in signal.
It has been hard to justify paying twice the price of 1.5T MRI when 3T can deliver only one-fifth more imaging power. But Haase argued that the key to capitalizing on 3T has been found in parallel imaging.
Parallel imaging and 3T MR can share a symbiotic relationship. Three-T compensates for the lost SNR that is the inherent trade-off when parallel imaging is applied, while a fourfold parallel imaging acceleration factor fully negates SAR effects at 3T, Haase said. This realization has led the manufacturers to add more receiver channels. Many scanners are equipped with eight channels, and 32 channels may become the standard.
The combination of parallel imaging and 3T opens the door to a doubling of phase encoding from 128 to 256 per breath-hold, more slices per scan, faster 3D imaging, higher patient throughput, and greater patient comfort. A good match with parallel imaging means less noise at 3T than 1.5T, translating into sharper morphologic images of the joints, brain, and vessels, Haase said. As expected, MR spectroscopy is easier to perform because a doubling of the distribution of spectra to 600 to 700 Hz improves the separation of resonance lines.
Problems still appear during body imaging. Shading effect is apparent from T1 power absorption.
"A lot of technical problems still to be solved involve signal inhomogeneity and SAR limitations," Haase said. "We need more coils, better coils, more receiver channels, and better gradient performance.
Several other speakers weighed in on 3T's progress in reports at the ISMRM meeting.
GOLD ON MUSCULOSKELETAL IMAGING
The rewards of 3T, including comprehensive knee studies in 15 minutes, are tangible with musculoskeletal imaging, said Dr. Garry Gold, an assistant professor of radiology at Stanford University (Figure 1).
The added resolution possible from 3T comes in handy for articular cartilage, especially in the talar dome, acetabular labrum, wrist, and hip joint, where cartilage is especially thin. Higher resolution aids the visualization of small structures such as the labrum, the triangular fibrocartilage complex, ligaments, and nerves, Gold said.
Good radiofrequency coils are key to addressing chemical shift and susceptibility artifacts, which are especially common in images of postoperative patients. Chemical shift problems can be addressed by increasing the readout bandwidth to 64 kHz and lowering the flip angle of refocusing pulses to reduce power deposition. But achieving fat suppression is more difficult, leading Gold to recommend setting the bandwidth above 32 kHz for non-fat-saturation imaging.
SCHONBERG ON CV IMAGING
The developers of cardiac MRI are banking on 3T combined with parallel imaging for the added speed needed to compete with 64-slice CT. The combination allows time-resolved and real-time acquisitions, said Dr. Stefan Schonberg, an associate professor of radiology at the University of Munich. The steady-state free precession pulse sequences often used during cardiac imaging, however, cause artifacts from inhomogeneities from the inherent phase-dependent signal of this technique. The solution at 3T is to modulate the frequency of the RF pulse to shift the artifact off the volume interest.
Combining 3T and the parallel imaging strategy TSENSE cuts scanning time by a factor of four, Schonberg said. Six slices can be acquired in a single breath-hold, capturing enough information for 3D modeling without degrading contrast-to-noise. A short-axis cardiac image can be acquired at 3T using TSENSE in 78 seconds, compared with six minutes on a nonaccelerated 1.5T scanner. Adding a 32-element coil can produce a sevenfold acceleration at 3T.
The marriage of the two technologies reaps benefits for delayed-enhancement MRI as well. DE-MR is inherently time-intensive. Determining the transmural extent of infarction depends on the selection of the inversion time. If a poor selection is made, the infarct may not be visible at all, Schonberg said. But phase-sensitive inversion recovery assures users that the infarction can always be viewed independently from the inversion time. This helps reduce dependence on the time-intense optimization approach that is required during 1.5T imaging.
During contrast-enhanced MR angiography elsewhere in the body, isotropic acquisition made possible with 3T allows radiologists to view suspected pathology with equal clarity from any angle. SNR is preserved even with parallel imaging acceleration factors of three.
Schonberg predicts that time-resolved 3T MRA will match the performance of steady-state MRA, revealing smaller vessels, aneurysms, and venous malformations. While acquisition times are reduced to less than a second, spatial resolution will approach the level of digital subtraction angiography. He expects that 3T whole-body MR angiography will foster rapid, comprehensive evaluations of suspected vascular disease viewed at about 1.5 mm or less isotropic resolution.
ROFKSY ON BODY IMAGING
Dr. Neil Rofsky, an associate professor of radiology at Beth Israel Deaconess Medical Center, admits to tempering his initial enthusiasm for 3T as he gained experience with it. Power deposition and signal homogeneity problems have dogged body imaging. Optimized RF coils are especially important, as illustrated by superb prostate imaging performed with
a phased-array coil. Inhomogeneity effects still appear sometimes, however, producing a splotchy pattern, especially for patients with ascites.
Vigilance is the name of the game for radiologists performing 3T body imaging, according to Rofsky. The payoff is in the future.
"I look forward to the meaningful functional capabilities that will change our game and really improve results for our patients," he said.
KUHL ON ENTIRE 3T PACKAGE
The radiology department at the University of Bonn has been a leading center for 3T applications development and testing. Vice chair Dr. Christiane Kuhl summarized her group's recent experience, including its work on 3T breast imaging.
Beginning with neuroradiology, the group demonstrated in a study involving 130 patients that eight-second, T2-weighted images acquired at 3T match the diagnostic accuracy at yield of 120-second acquisitions at 1.5T. Three-T outperformed 1.5T in a trial for detecting plaques in early multiple sclerosis. It also predicted whether patients with early symptoms would progress to full-blown MS, Kuhl said.
"Three-T buys us more signal and contrast-to-noise ratio and thus helps us to identify more subtle plaques, in particular in the infratentorial and juxtacortical regions. This has substantial impact on patient management," she said.
Research at Bonn suggests that 3T MRA performs better than vascular imaging at lower field strength for characterizing small arteries at the base of the brain. Adding parallel imaging expands the anatomic coverage.
Equivalence between 1.5T and 3T was established for coronary angiography and delayed enhancement cardiac studies of myocardial infarction. After more development, the Bonn group expects a diagnostic advantage. Other work showed that cardiac perfusion profits from the combination of higher field strength and parallel imaging by improving image quality and the detection of perfusion deficits, particularly during adenosine(Drug information on adenosine) stress perfusion imaging.
In the abdomen and pelvis, the Bonn group established that 3T is at least good as 1.5T for classifying focal liver lesions. Adding 3x acceleration SENSE parallel imaging promises to shift the advantage to 3T, Kuhl said. More clinically relevant information about lymph node metastases appeared in 3T compared with 1.5T pelvic studies performed in a survey of 25 patients. With flip-angle focusing techniques, such as flip-angle sweep, high-resolution pelvic MRI is achieved in less than 35 seconds, while improving on the image quality achieved during a conventional four-minute acquisition.
Kuhl's research suggests that 3T will reinforce MRI's designation as the gold standard for breast cancer imaging. Because of SAR, T2-weighted sequences at 3T, even with SENSE acceleration, took 4.5 minutes. Applying flip-angle sweep cuts the acquisition time to 45 seconds. More important, temporal and spatial resolution improve with the combined use of 3T and parallel imaging, she said.
The higher signal can be used to acquire unilateral fat-suppression images with a noninterpolated voxel size of 0.5 x 0.5 x 1 mm in 100 seconds (Figure 2). In addition, she said, fat suppression is possible over a large field-of-view.
Better image quality at 3T is translating into clinical advantages for breast imaging, Kuhl said. In a study of 52 patients who underwent breast MRI at 3T and 1.5T, lesions were identified at 3T that were not called at 1.5T, and specificity distinguishing between benign and malignant enhancement was improved.
Diffusion-weighted imaging and MR spectroscopy show promise for breast applications, and T2* perfusion imaging can be used to identify capillary leakage. Partial volume average is less of problem for proton MRS performed at 3T. False negatives that appear during 1.5T imaging because of missing choline peaks are more reliably presented at 3T.
Haase called 3T a good idea during the ISMRM plenary session.
"For routine clinical studies, 3T gives the highest SNR for clinical applications. The technology measures more data in less time, improves spectroscopy, enhances efficiency, and increases throughput," he said. "But there are technical problems to be solved."
This special section was compiled by James Brice, senior editor of Diagnostic Imaging.
