GE Healthcare reaches out to MR community for new ideas

Ideas for tomorrow's products are bouncing around the medical community right now. Some are wild imaginings, hardly tested, if tested at all. Some have preliminary results based on animal or cursory human tests. Others are well thought out but lack the technological engine to push them forward. GE Healthcare has told Michael Wood, Ph.D., to snatch up as many in MR as he can.

In a few days, GE's new general manager of research collaborations for global MR will be looking for those ideas at the International Society for Magnetic Resonance in Medicine meeting in Miami.

"Frankly, anything that will happen in MRI will happen there," Wood said.

He'll be attending the sessions, but the most interesting information is likely to come from encounters in the hallways.

"There will be a buzz about something," he said. "I don't know what that will be yet, but every year there is a buzz about something. We need to assess that and talk to a core group of people we can count on to steer us toward the new areas of MR."

One idea that has already caught his attention is the use of diffusion-weighted imaging (DWI) as a prognostic tool in brain cancer. Results obtained at the University of Michigan Comprehensive Cancer Center on 20 patients indicate that changes happening in the movement pattern of water molecules in the brain several weeks after the start of chemo- and/or radiotherapy correlate strongly with a positive response to the treatment.

Early assessment of patient response promises to help physicians decide whether the patient should continue a therapy regimen or switch to a more beneficial one. The first application may be in brain cancer, where it is currently being studied. But other forms of cancer, such as those invading the breast or prostate, might also be addressed.

Clinical applications, however, must await the careful documentation of DWI data and interpretation of their significance. Wood predicts that more detailed results with larger groups than the 20 patients at Michigan will appear with increasing frequency at scientific meetings.

Other technologies may follow a different path toward adoption. An example is the one taken by GE's TRICKS (time-resolved imaging of contrast kinetics). This technology, initially developed at the University of Wisconsin in Madison, makes possible dynamic MR angiography of the peripherals, automatically producing maximum intensity projections through each 3D volume. The technique is proving helpful in imaging complex vascular anatomy, such as the visualization of collateral flow and retrograde flow below the knee.

The technique had been around for awhile. It just wasn't practical.

"TRICKS languished because it took hours on a host computer at the university to make the images," Wood said.

GE's advanced data pipeline, High Definition MR, cut the processing time of TRICKS to a matter of seconds or minutes, depending on the volume of data being computed.

HDMR had a similarly pragmatic effect on a technique called Propeller, whose radial acquisition of data "sprays" artifacts out of the volume rather than through it. This makes the images relatively insensitive to motion. Propeller is, therefore, ideally suited for use with patients prone to movement: unsedated children and adults who can't cooperate because of tremor, such as Parkinson's disease patients.

Processing these radially acquired data had been the stumbling block, but this problem was overcome by HDMR. Other such computing-intensive techniques are out there, just waiting for technology to catch up. An example is VIPR (voxel isotropic projection imaging), an MR angiographic technique with the potential to increase the volume of coverage and spatial resolution by a factor of 30. Unfortunately, the computational requirements are commensurate.

"In the early days of MR, we were frustrated at how long it took to get a standard image," Wood said. "We progressed to where that wasn't a problem, but now we are back to the beginning again. Some of the exotic techniques that we want to do take too long to make images."

The answer will come from technologies that process images faster. They will make possible many of the ideas that Wood hopes to find brewing in the imaging community.

"There are examples of wild ideas that are very speculative, and there are ideas of technologies that are essentially finished and all we would have to do is license them," he said.

To find and assess these ideas, Wood is personally reaching out to the MR community. He attends meetings, listening to presentations and clinical teaching sessions. He brainstorms with presenters and luminaries.

"Meetings give us an opportunity to get together and recalibrate and to find out what might be promising in the future," he said. "The key is talking to people who are going to play a role in shaping the future."

GE has various approaches to the development of ideas. The company might get directly involved. The preferred route to product development of new ideas has GE playing matchmaker between the academicians who invent a technology and a small company with expertise in the area that can further develop it. GE would then work with the small company to ensure that the final product can be integrated, or at least made compatible, with GE equipment.

"What I like about this approach is that the innovation and the development are done with the group that knows about it," Wood said. "Then we step in and tie it to our MR scanner, so we are not trying to reinvent it."

Regardless of the path, ideas should advance to the product stage only after their clinical benefits have been well documented, he said. Temptations to jump the gun may arise, especially when the technological barriers to productization are low, as in the case of using DWI for tumor assessment. A radiologist, for example, can create a functional diffusion map of the brain using the hardware and software already in hand, according to Wood.

Its use in assessing patient response to cancer treatment starts with a fairly standard acquisition of diffusion-weighted images. These are then sent to a workstation for coregistration with the basic imaging data set, segmentation, and finally statistical analysis. The output is a presentation that helps radiologists and radiation oncologists assess whether there is a change in the diffusion-weighted coefficient of the tumor.

"The next step is to extend the scope of the study to make sure the promise is borne out," Wood said. "Parallel to that, we need to look at how to integrate image analysis with acquisition. The key is integrating the acquisition and analysis to make it convenient for people to use."