Parallel imaging spurs performance of MR at 3T

July 1, 2006

Parallel imaging continues to drive MRI development as an enabling technology for clinical 3T and as the catalyst for new data acquisition strategies offering unprecedented improvements in temporal resolution.

Parallel imaging continues to drive MRI development as an enabling technology for clinical 3T and as the catalyst for new data acquisition strategies offering unprecedented improvements in temporal resolution.

Combining parallel imaging and 3T exploits the strengths of both while minimizing their weaknesses, said Axel Haase, Ph.D, director of the biophysics department at the University of Wurzburg in Germany. Parallel imaging cuts image acquisition times, but loses signal-to-noise ratio. Doubling field strength from 1.5T to 3T also theoretically doubles the SNR to compensate for losses from parallel imaging.

Parallel techniques, such as SENSE, GRAPPA, and SMASH, address problems that have kept 3T from delivering promised improvements in resolution and speed. They compensate, for example, for a four-fold increase in the specific absorption rate at 3T. More radiofrequency channels, more sophisticated coil designs, and pulse sequences optimized for longer T1 times of 3T imaging have also reduced absorption rate problems.

Imaging of the brain, spine, heart, vessels, joints, and cartilage have improved, according to Haase. Better spectral separation is possible during MR spectroscopy, and acquisition times for diffusion tensor imaging have been cut as short as five minutes, increasing its clinical practicality. Technical problems with body imaging remain, but clinical 3T is realizing more of its potential, he said.

Undersampling imaging acquisition and reconstruction techniques are redefining the limits of MR temporal resolution. At the 2005 International Society for Magnetic Resonance in Medicine meeting, Dr. Jurgen Hennig, director of medical physics at the University of Freiburg in Germany, introduced vastly undersampled isotropic imaging with projection (VIPR), an undersampling method that reduces the amount of redundant data collected during image acquisition.

Hennig's lecture inspired Charles Mistretta, Ph.D, a professor of biomedical physics at the University of Wisconsin in Milwaukee. He responded with highly constrained back projection (HYPR), an approach that requires even less data. Mistretta summarized his findings at the 2006 ISMRM meeting.

Although applicable only to MR angiography and other sparsely populated data sets, HYPR can acquire a complete image with 2472 pixel-in-plane resolution in 0.5 seconds. That is 96 times faster than conventional methods when combined with a radial data acquisition technique. In the study, an acceleration factor of 228 was achieved by combining VIPR and HYPR to acquire 18 heart images per cardiac cycle. In another demonstration, phased-contrast images examining the dynamics of contrast media passage were acquired in four minutes instead of the 39 hours that would have been required during conventional Cartesian imaging.

"We have many techniques that exploit the redundancy of data in our time-resolved application," Mistretta said. "All these things can be refined with parallel imaging to generate applications that before now were almost unimaginable."