3D whole heart MR angiography image obtained in 67 seconds. Image source: Toshiba Medical Systems (Tochigi, Japan)

MRI provides excellent soft-tissue contrast, but the drawback is its temporal resolution. Whereas CT scanning of the coronary arteries takes mere seconds, MRI can take up to 20 minutes or even longer. The latest innovation to increase the speed of MRI, however, is to employ multi-channel radiofrequency (RF) coils with parallel acquisition techniques.

Speed vs. resolution

Initially, researchers developed detector coil arrays, which enabled imaging of extended regions with increased signal-to-noise ratio (SNR). Next, parallel imaging techniques emerged, the earliest of which were known as “SMASH” and “SENSE,” which are enabled by multi-channel RF coil arrays.

“We want try to accelerate the scan to capture an even motion of the heart within a shorter period of time. We are not there yet compared with CT, but because of parallel imaging techniques being available now, cardiac MR imaging is becoming more competent,” says Hiroyuki Fujita, founder and CEO of the Cleveland-based Quality Electrodynamics (QED), which develops RF coil technology.

To support parallel imaging in cardiac applications, one also needs a high-channel count RF coil. Initially, eight-channel coils were routine, then 16, then 32. In the last few years, researchers have experimented with 128-channel RF coils. Having more receiver coils potentially allows more information to be acquired simultaneously, thereby shortening the scan time considerably.

Hardy and colleagues from GE Global Research experimented with a 128-channel body MRI with a flexible high-density receiver-coil array (J Magn Reson Imaging 2008;28:1219-1225). The array comprised two “clamshells” containing 64 coils each, with the posterior array built to maximize SNR and the anterior array design incorporating considerations of weight and flexibility as well. Researchers found that the larger number of receiver channels—compared with 32 channels—enabled significantly higher acceleration factors for parallel imaging and improved SNR, which theoretically would lead to improved sensitivity.

Schmitt and colleagues from Massachusetts General Hospital in Boston tested a 128-channel receive-only cardiac coil for highly accelerated cardiac MRI at 3T (Magn Reson Med 2008;59(6):1431-9). The in vivo measurements with the 128-channel coil resulted in SNR gains compared with a 24-channel coil (up to 2.2-fold in the apex, closest to the coronary arteries). The 128- and 32-channel coils showed similar SNR in the heart. The ability of the 128-channel coil to facilitate accelerated cardiac imaging, however, was demonstrated in four volunteers using acceleration factors up to seven-fold—within a single breath-hold.

While the increase in speed may be minimal between 32- and 128-channel coils, the 128-channel model allows acceleration over larger volumes and multiple different body areas, according to Daniel K. Sodickson, MD, PhD, vice chair for research in the department of radiology at New York University Langone Medical Center in New York City. “Without changing out of the coil, you can get ultrafast acquisition and comprehensive imaging of the heart and then ultrafast comprehensive imaging of the peripheral arterial system,” says Sodickson.

The next step is to maximize coil fittings to the particular anatomy. “Well-designed coil arrays are critical for parallel acquisition,” says Stephen Riederer, PhD, director of the MR lab at the Mayo Clinic in Rochester, Minn. “They have the potential to allow better superior and inferior coverage.”

Riederer and colleagues found that the conventional coil arrays used in the lower limbs had uniform elements that, upon closer scrutiny, were found to be too wide for optimal anterior-posterior image acquisition. They designed a coil array with longer, narrower elements in the anterior-posterior positions. “We increased our spatial resolution dramatically,” he says. The team is developing other element modules for all regions of the anatomy that better conform to a patient’s diameters.

Speed & simplicity

With MRI, it’s not just about speed; it’s also about simplicity, Sodickson says. Today, the technologist needs to plan each scan plane separately, which can take up to an hour for a full cardiac MRI exam. Using a grant from the National Institutes of Health, Sodickson and colleagues are developing coils and image reconstruction techniques that can accelerate imaging by at least a factor of eight compared with conventional cardiac MRI.

“We realized that if you have enough speed, you can simplify the whole process as well,” he says. “Rather than image each coronary artery one after the other, with enough speed we can essentially press a ‘go’ button and scan from the top of heart to the bottom of heart in one breath-hold. In the next breath-hold, wall motion would be imaged for functional data, then perfusion in a breath-hold, then delayed enhancement in a breath-hold. The entire exam would last five minutes and you’d get a comprehensive volumetric survey with all the different rich information that MRI can deliver.”

Sodickson and colleagues will soon be recruiting subjects to compare the ultrafast scanning technique with conventional MRI. They want to ensure that the ultrafast scanning delivers all the required information.

Vendors, such as Siemens Healthcare, with whom Sodickson and colleagues partner (but not for research funds), have begun to address MRI workflow issues. Instead of technologists having to input technical information to guide each part of the scan, they merely input patient data, such as heart rate and breath-hold capacity. The software optimizes the patient exam strategy, automating the steps needed to get a high-quality scan and adapting to the patient’s condition.

Siemens calls this software Dot (day optimizing throughput). The Royal Bournemouth Hospital in the U.K., after testing a prototype of the cardiac Dot (it’s not yet available in the U.S.) reported that throughput and ease of use increased. Whereas only three technologists could routinely handle the complexity of cardiac MR, the software-based workflow automation now allow all 16 scanning staff to perform cardiac exams..

“The goal is to connect ultrafast acquisitions with workflow engines,” says Sodickson. “We are at least approaching the simplicity of CT, while maintaining the rich content of MR, and all without exposure to ionizing radiation.”

1.5T vs. 3T

When parallel imaging is used with a 1.5T scanner, the scan time shortens but the images are often poor for cardiovascular applications in terms of SNR. At 3T, the scanner has more SNR to begin with, so the SNR sacrificed to speed doesn’t influence image quality as much as on a 1.5T system. At 3T, the scan time is faster and the image quality is better, says Fujita.

The next step is to combine high-Tesla scanners with multi-channel coils and optimized protocols into one package with workflow optimization in mind, says Fujita, adding that this reality is a few years away. “If we can match the temporal resolution of CT with MRI, than MRI has a distinct advantage in providing soft tissue contrast. There will be no modalities matching MRI.”

At some point, however, adding more coils will not show a benefit in speed. To that end, Sodickson has taken a page out of applied mathematics. Rather than acquire enormous amounts of data, which can then be compressed without losing the most important pieces, researchers are experimenting with acquiring pre-compressed images. Called “compressed sensing,” this technique has the potential to increase speed without adding more coils.

“I’m not a believer in the either/or philosophy of MR or CT,” says Sodickson. “MR is not necessarily best positioned to replace CT entirely, but it is getting faster and simpler and has a rich information content, which is no longer going to be accessible just to the experts.”