Vendors challenge limits of MR speed and resolution

August 1, 2005

Parallel imaging is extending the limits of resolution with anatomic and functional studies of unprecedented clarity and diagnostic value. It is cutting acquisition times by more than half to freeze motion more easily and increase patient throughput.

Parallel imaging is extending the limits of resolution with anatomic and functional studies of unprecedented clarity and diagnostic value. It is cutting acquisition times by more than half to freeze motion more easily and increase patient throughput.

Parallel imaging attacks the cause of poor MR temporal resolution. Compared with other imaging modalities, conventional MRI is slow because it must perform numerous time-consuming spatial encoding steps to form an image. Parallel imaging produces the same benefits as multislice CT, said Dr. Daniel K. Sodickson, director of the Laboratory for Biomedical Imaging Research at Beth Israel Deaconess Medical Center. Vast improvements in acquisition speed or imaging resolution are possible as more channels are added.

Like CT, MRI is gearing up to acquire data over many more channels. The technology has progressed quickly to a 32-channel option as part of Siemens' total imaging matrix (TIM) coils and integrated parallel acquisition technique (iPAT) applications package. GE Healthcare and Philips have developed 32-channel systems. And Lawrence L. Wald, Ph.D., an assistant professor of radiology at Massachusetts General Hospital, demonstrated a prototype 96-channel head coil at the International Society for Magnetic Resonance meeting in May.

All vendors apply variations of SMASH or SENSE, the two main parallel imaging strategies first described in the late 1990s. SMASH, or simultaneous acquisition of spatial harmonics, uses radiofrequency coils to work with the scanner's gradient coils to perform spatial encoding. The strategy, which Sodickson developed, cuts the number of time-consuming gradient steps while enabling the scanner's computer to fill in the blanks of undersampled data by analyzing linear combinations of signal from the RF coil arrays.

Dr. Peter Boesiger, a professor of biomedical engineering at the University of Zurich in Switzerland, invented SENSE (sensitivity encoding). Its impetus was the realization that the induced signals detected by RF receivers vary depending on their distance from the targeted anatomy producing the signal. That finding led Boesiger to conclude that multiple receivers in parallel could reduce scan time.

SPEED AT A PRICE

Higher speeds come at the price of lower signal-to-noise ratio, however. Multiple RF channels may expedite acquisition, and clever data processing strategies may form complete images from only a partial sampling of k-space, but these strategies amplify noise. This feature of parallel imaging creates an incentive to optimize coil performance and to increase field strengths to address the problem.

GE developed Excite HD, an effort to improve the efficiency of all scanner subsystems, to make the most of its Asset system. The use of many small coil elements preserves SNR but generates two to four times more data to tax the processing capability of a scanner's array processor and other components, said Michael L. Wood, general manager of MR research collaborations at GE. RF components were redesigned to avoid data bottlenecks.

The Asset eight-channel body array features 12 parallel coils that cover the chest to the pelvis without repositioning. It produces 40% more SNR than its predecessor four-channel torso array, Wood said. The benefits of Asset are reflected in improved breast, body, and cardiac applications.

Since it became the first vendor to commercialize parallel imaging in 2001, Philips has improved the user's control over SENSE, expanded its line of SENSE-optimized coils, and applied the technology to all its MR products, including the 3T Achieva.

Philips offers a SENSE accelerator with settings from a factor of 1.1 to 8, which gives radiologists the ability to adjust the settings by as little as one-tenth.

Siemens has pursued an aggressive dual strategy. Its engineers have pushed the limits of iPAT with this year's commercial introduction of the first 32-channel system capable of acceleration factors of up to 16 on its 3T Trio scanner. Its TIM surface coil concept optimizes iPAT performance while offering ergonomic advantages that translate into greater comfort for patients, higher productivity and convenience for technologists, and huge fields-of-view involving the use of up to 102 integrated coil elements for radiologists.

The iPAT strategy is pivotal to Siemens' MRI product offerings and its vision of MRI design trends, according to Nancy Gillen, national MRI sales manager. Siemens engineers conceived of TIM as a way to add iPAT to any application while using various combinations of its flexible RF coils.

Hitachi customers have access to parallel imaging through RAPID (rapid acquisition through a parallel imaging design). Introduced on the 0.3T Aris Elite open platform in 2003, RAPID will be available for the 0.7T Altaire later this year, according to Shawn Etheridge, MRI marketing manager. Head, extremity, and body coils will be optimized for parallel imaging. RAPID generates submillimeter resolution when used with a 3D Balanced SARGE sequence for musculoskeletal applications.

Toshiba Medical Systems adopted a SENSE-like technique for SPEEDER, its first parallel imaging product, in 2003. A multichannel head and neurovascular, body, and shoulder coils were developed in four- and eight-channel configurations for Toshiba's 1.5T Vantage scanners.

While parallel imaging can work high-speed wonders at 1.5T, it may work even better at 3T and beyond. Signal loss caused by parallel imaging is less of a problem, because more signal is produced at very high field strengths.

"Parallel imaging makes high-field imaging better, and high field strength makes parallel imaging better. There is a synergy," Sodickson said.

Parallel imaging solves problems associated with susceptibility artifacts and the increased RF power deposition that prevents high-field systems from performing at their full potential. While addressing those issues, it accelerates imaging speeds and improves resolution.

The next step for parallel imaging will be taken in what Sodickson calls "the great receiver race." Siemens is already out of the blocks with a 32-channel system, and GE and Philips are weighing their options with their own 32-channel products. More channels increase the system's capacity for acceleration. Volumetric imaging will replace targeted slab imaging, increasing reconstruction options. Coil arrays will continue to improve, making the SNR trade-off less of an issue, and system integration along the lines of iPAT and TIM will remain an important theme, Sodickson said.

Researchers are even investigating the possibility of multichannel transmit parallel imaging to further extend the limits of MRI.

Mr. Brice is senior editor of Diagnostic Imaging.