Extra signal power at 3T gives users a choice

March 1, 2004

With nearly two times more signal than 1.5T MR, 3T scanners give radiologists the luxury of choosing to improve spatial resolution or to apply imaging techniques that are not otherwise clinically feasible, due to long scanning times. Because they do not have to fight motion, neuroimagers tend to invest the extra signal of 3T in increased resolution to obtain more definitive cognitive activation maps, conduct high-resolution inflow MR angiography, and enhance the sensitivity of diffusion studies. Constrained by the time frame of a breath-hold, body imagers look to capitalize on 3T's speed to monitor the response to antiangiogenesis agents through arterial spin labeling and to interrogate the thorax and abdomen with spectroscopy.

When it comes to brain functional imaging, the trade-off between speed and resolution at 3T MRI is clear: Go for improved spatial resolution. With an intrinsically high-speed technique such as echo-planar imaging, there is not much difference in image acquisition time between 1.5 and 3T in the brain. So 3T's bonus signal-to-noise ratio can be targeted at improving spatial resolution to find discrete areas of function as well as to add anatomic specificity to the localization of activation, said Dr. Bruce Rosen, director of the Athinoula A. Martinos Center for Biomedical Imaging at Massachusetts General Hospital in Boston.

Because radiologists always seem to want more spatial resolution for MR angiography, both brain and body imagers aim 3T's signal gain at image quality in blood flow investigations. The result in the brain is better characterization of cardiovascular disease, according to the authors of a study that compared high-spatial-resolution 3D time-of-flight MRA with a standard 3D TOF protocol at 3T and at 1.5T.

Experienced readers said the high-resolution 3D protocol provided better visualization of small vessel segments and vascular pathologies within the circle of Willis than the standard protocol at either field strength. On a scale of 1 to 5, readers rated the high-resolution protocol an average 4.5 in image quality, compared with 3.1 for standard TOF at 3T and 2.2 for TOF at 1.5T. In the study of eight patients and seven volunteers, conducted at the University of Bonn, Germany, and presented at the 2003 meeting of the RSNA, the high-resolution protocol had a matrix of 832 x 568 on a 250-mm field-of-view. The standard protocol had a matrix of 336 x 213 on a 160-mm FOV.

The finer spatial resolution of 3T may raise MRA to the level of more invasive investigations. The extra signal at 3T could bring MRA in the body close to the resolution of digital subtraction angiography, according to Dr. Robert Edelman, chair of radiology at Evanston Northwestern Medical Center in Evanston, IL.

Diffusion imaging is so signal-hungry at 1.5T that the SNR is only borderline. But now that parallel imaging has reduced echo train length in single-shot diffusion imaging, 3T's improved SNR can find changes in the apparent diffusion coefficient, small ischemic areas, and microemboli, said Dr. Christiane Kuhl, head of the MRI section at the University of Bonn.

Because diffusion studies rely on echo-planar imaging, all phase-encoding steps are acquired with one radio-frequency excitation, which accumulates phase errors and causes susceptibility effects and image distortions, Kuhl said. But parallel imaging decreases phase errors by reducing the number of phase-encoding steps.

In a trial involving 55 patients and presented at the 2003 RSNA meeting, Kuhl found that blurring and image distortions were reduced, particularly in areas close to the skull base, on diffusion studies with parallel imaging at 3T. Ischemic lesions located near the skull base or vertex were more conspicuous at 3T than at 1.5T, and microemboli in the frontodorsal and temporal cortex were detected only at 3T in four patients.


The bump in speed that accompanies the increased SNR at 3T MR should move advanced imaging techniques out of the research realm and into clinical practice.

"When you get a burst in signal to noise, you can take a 20-minute exam, which is not practical for a single study, and turn it into a 12-minute exam, which is feasible for most patients," Rosen said.

Two beneficiaries are diffusion tensor and diffusion spectral imaging for viewing brain tissue in greater detail and for mapping connections within the white matter. These examinations are prohibitively long when performed with conventional techniques at 1.5T, he said.

Time is also a factor when using arterial spin labeling (ASL) to measure blood flow in the brain or heart. The magnitude of the signal changes in ASL is small, on the order of a few percentage points. Getting enough SNR at 1.5T to generate a clinically useful image takes 12 to 15 minutes. But because of the boost in SNR at 3T, small changes in signal on ASL appear in only about four minutes.

For evaluating the cerebral circulation in patients with suspected stroke or for conducting preoperative workup of brain tumors, high-resolution ASL at 3T may replace T2*-weighted dynamic susceptibility contrast (DSC) perfusion pulse sequencing with continuous ASL, Kuhl said.

By tracking the change in signal intensity loss in the small capillaries with the passage of gadolinium through the circulation, DSC perfusion imaging provides a relative measurement of the perfused volume of brain tissue at the microvessel level.

Continuous ASL (CASL) at 3T provides the same information as DSC perfusion imaging without contrast enhancement. And it quantifies the milliliters of perfusion per 100 mL of brain volume to generate an absolute perfusion value. But it has not been used clinically at 1.5T because acquisition is time-consuming.

The longer T1 relaxation time and higher SNR at 3T may make it possible for CASL to identify globally or homogeneously decreased perfusion in the brain. An example is a patient with bilateral carotid stenosis who has insufficient perfusion in both hemispheres. A contrast-enhanced perfusion study would appear to be normal because it would not find any differences in perfusion; both hemispheres would have the same level of reduced perfusion. CASL, however, would demonstrate that there was too little perfusion on both sides of the brain. It also would quantify the amount of blood flowing to a tumor and assess the effects of chemotherapy by documenting reductions in perfusion, Kuhl said.

With 3T MR, Dr. Neil Rofsky, director of MRI at Beth Israel Deaconess Medical Center, can perform ASL in the body and monitor response to antiangiogenesis drugs in renal cell cancers. And he has found that changes in tumor perfusion patterns, as shown by the passage of magnetically labeled protons in the blood, occur six to eight months earlier than changes in tumor size, providing a more timely assessment of treatment efficacy than standard anatomical imaging.

Speed at 3T is particularly helpful in cardiac imaging, and the most promising application is in myocardial perfusion, said Dr. Scott Flamm, director of MRI and cardiovascular MRI research at St. Luke's Episcopal Hospital/Texas Heart Institute in Houston.

"One of the limitations at 1.5T is that we cannot push enough slices to get complete coverage of the heart during every beat. We are usually limited to three or four slices. If we go to 3T and use parallel imaging, we may be able to increase the number of slices we get through the myocardium because of the inherent increased signal intensity," Flamm said.


Choosing between resolution and speed is a nice distinction to make on paper, but clinically, radiologists often apply the increased signal intensity of 3T for both. In breast imaging, extremely high spatial resolution is needed to depict the small morphological details of cancers. Imaging also has to be fast to capture differences in enhancing tumor and adjacent breast tissue in the early postcontrast phase. MR at 3T seems to be ideally suited to breast imaging because it has enough signal to use a higher imaging matrix and obtain contrast from lesions before normal breast tissue enhances, Kuhl said.

"You can go twice as fast as 1.5T with the same types of signal to noise, or you can get twice as much resolution, or you can go with a combination of the two," Rofsky said. "It's the user's choice."

MS. SANDRICK is a freelance writer in Chicago.