More is better and all but inevitable in medical imaging. In CT, more means slices. In MR, it’s channels for receiving radiofrequency signals. These currently number 32 on the most advanced commercially available systems. But a replacement for that benchmark is in the works.
More is better and all but inevitable in medical imaging. In CT, more means slices. In MR, it's channels for receiving radiofrequency signals. These currently number 32 on the most advanced commercially available systems. But a replacement for that benchmark is in the works.
In collaboration with Massachusetts General Hospital, Siemens Medical Solutions has redesigned its high-end 3T Magnetom to simultaneously acquire and reconstruct data from up to 128 channels. A presentation Sunday at the International Society for Magnetic Resonance in Medicine meeting in Berlin described how the addition of RF switch matrices, receiver boards, splitters, clock and fiber-optic control signals, and other electronics quadrupled the number of channels on a Siemens Trio. The move could change the future of MR.
"In going from 32 to 128, we hope to benefit in sensitivity and the ability to accelerate image encoding with parallel imaging methods, such as GRAPPA and SENSE," said Lawrence L. Wald, Ph.D., director of the NMR Core at the Martinos Center for Biomedical Imaging at MGH, where the 128-channel Trio is being put through its early paces.
Sensitivity rises because of the increased signal-to-noise ratio that occurs when a given patient area is scanned using more, ever smaller coils, Wald said. The downside is that signal reception is nonuniform over this area, but this is less of a problem when doing modern reconstructions.
"GRAPPA and SENSE can accommodate nonuniform reception nicely," he said. "So in our view, added sensitivity is a good thing, even if it is added nonuniformly."
The advantages were clear when the Siemens prototype was tested with a 128-channel cardiac coil. The matched coil and scanner allowed whole-heart coverage in a single breath-hold with a sevenfold acceleration, producing TrueFISP cine images of diagnostic quality and capturing data that supported cardiac function analyses such as ejection fraction, Wald said. The greatest SNR gains were at the apex of the heart, exactly where enhanced SNR is needed if the coronaries are to be visualized in their entirety.
The Siemens/MGH team hopes to achieve between 12- and 16-times acceleration on the system to allow fast 3D coronary and whole-brain imaging for functional MRI at an ease-of-use level akin to CT, Wald said.
"I think it's a new world for MRI, where the burden of image encoding shifts from the gradients to the coil array," he said.
The cardiac images were reconstructed using software provided routinely on the commercially available 32-channel system. But more horsepower will be needed as the scanner expands to other applications. The researchers are planning to test a high-performance reconstruction engine to support challenging clinical applications associated with parallel imaging at higher acceleration factors.
"While the current generation of reconstruction computers can keep up with parallel imaging algorithms for 32 channels, 128 channels taxes them a bit," he said. "It also taxes raw data storage, since the raw data (from an exam) can be 100 GB."
Whether these issues become a concern for the general imaging community any time soon will depend on how Siemens' 128-channel prototype fares in applications now in the works. Clinical applications will determine when 128-channel systems become commercially available, Wald said.