By: Catherine Carrington

Predicting the future has always been a tricky business, but charting CT's path requires juggling an especially complex set of variables. Technology has been advancing at near warp speed, and clinical performance is stunning compared with just five years ago. But even as 16-slice scanners make their market debut, agreement is slippery on what the next big innovation will be. How best to overcome continuing challenges in cardiac imaging and information technology is another question with multiple answers.

Industry observers do agree on one point: CT faces the future from a position of strength (see accompanying story). Data from marketing and research firm IMV Medical Information Division indicate that procedure volume has shot up 51% since 1998. Moreover, multidetector technology is finally beginning to account for the majority of CT sales. Clinically, CT's future is wide open, according to industry insiders.

"CT is not anywhere close to its ultimate performance. We look at clinical applications and know there are further distances we can go to improve performance," said William Kulp, product marketing manager for Philips Medical Systems.

Cardiac imaging, for example, continues to test CT's mettle. The 16-slice scanner offers an unparalleled ability to visualize cardiac anatomy and function. To fully meet the demands of CT coronary angiography and myocardial perfusion imaging, however, would require a scanner capable of not only extremely high spatial resolution, but also temporal resolution in the neighborhood of 50 msec or less and the ability to image the entire heart at one time. That's a tall order, even for today's advanced scanners.

"Everything has to get better. You'll see bigger detectors, faster rotation speeds, faster reconstruction speeds," Kulp said. "And you'll see very large area detectors that have the same kinds of image quality specifications as the detectors we're producing today."

SETTING PRIORITIES

Each vendor is setting its own agenda on how to pursue these goals. Until recently, the highest priorities have been to improve spatial and temporal resolution and reduce scan times. The imaging industry has advanced on these priorities at a fast pace. Toshiba Medical Systems' Aquilion 16 scanner, for example, captures slices as thin as 0.5 mm, while GE Medical Systems' LightSpeed 16 comes in at 0.625 mm, followed by both Siemens Medical Systems' Sensation 16 and Philips' Mx8000 IDT at 0.75 mm. Gantry rotation times of 0.5 msec (0.42 sec in Siemens' case), coupled with reconstruction algorithms, put effective temporal resolution close to the 100-msec mark in cardiac mode. Total scan times have dropped dramatically as well, from perhaps 40 seconds to image the entire heart with a four-slice scanner to 20 seconds with a 16-slice scanner.

Although further improvements in temporal and spatial resolution are still goals, the focus of research and development is shifting. For one reason, there are limits on how much faster a scanner can be expected to acquire data using a rotating gantry, and for another, technological hurdles must be overcome before it's possible to further reduce slice thickness while maintaining a desirable signal-to-noise ratio and limiting radiation dose to the patient.

The new focus is on increasing the volume of coverage. In the quest for increased coverage, several vendors have already made modest advances in expanding detector size. Toshiba's detector provides 32 mm of z-axis coverage, while the Philips and Siemens detectors capture 24 mm at a time, and GE's detector captures 20 mm.

Research and development teams have far more ambitious goals, however. Toshiba is testing the imaging capabilities of a prototype 256-slice detector. Among the most tantalizing: the ability to acquire 3D images in real-time. It's a long leap from the R&D lab to commercial release, but such technological advances hint at what the future may bring-and the clinical doors it may open. Once again, cardiac imaging provides a ready example.

"What this technology provides is the ability to view the heart continuously without moving the table. As your gantry rotates, you'll be able to see-in 3D and real-time-what's happening in the patient. It's going to be very exciting," said Bryan Westerman, Ph.D., clinical sciences manager for Toshiba.

In the interim, it's almost certain that detector width will expand incrementally, though with 16-slice scanners just breaking into the market, no one is saying what features are likely to show up next, or when.

MORE SLICES OR NO SLICES?

A long-range option for increasing coverage is the use of flat-panel detectors. By dispensing with multiple detector rows, flat-panel detectors eliminate the need to acquire image data slice by slice. Assuming the panel is large enough, they also enable imaging an entire organ at once, an important characteristic for perfusion imaging.

"Flat-panel displays take you from slices to volumes, and the closer we get to true volume information, the more exciting it gets," said Dr. Lawrence N. Tanenbaum, neuroradiology chief at New Jersey Neuroscience Institute in Edison. "It takes you to the next level, to where you're getting seamless volumetric data that you can slice and dice infinitely. It should radically affect the quality and style of our scanning."

Dr. Thomas Brady, who directs the cardiac imaging program at Massachusetts General Hospital, has been working with GE in its development of a flat-panel detector. The early performance of volume CT, at least in imaging tissue specimens, has been impressive. Research progress has been slow, however.

"The image quality is phenomenal. The contrast is excellent, especially between bone and soft tissue," he said. "But there's not much new on the VCT front."

Some, including Kulp, question flat-panel detectors as a promising approach to improving coverage volume, at least in the near term.

"We've mounted those on CTs in the past and taken a look at the images, and they're not ready for prime time yet. It sounds cool-very wide coverage, very thin slice-but there are significant drawbacks," he said.

The main problem, according to Kulp, is that so far it has been impossible with the amorphous silicon flat panel to turn the detector on and off and get a signal fast enough to image a moving target, such as the heart.

"It's impossible to get high-quality CT images-to get the resolution and speed that you want-without turning the detector on and off very rapidly," he said. "You can image something that doesn't change much, like someone's hand, but not the heart."

Dr. Jeffrey Carr envisions a cath lab capable of real-time 3D imaging, which could prove useful in percutaneous interventions and minimally invasive surgeries. A CT cath lab would provide the ability to precisely track the location and movement of instruments during placement of stent-grafts or off-pump coronary artery bypass grafting, for example, reducing the risk of complications and making it possible to further reduce the size of surgical incisions. In contrast to MR, CT would permit the use of instruments and equipment routinely used in the operating room.

"Everything we use in the cath lab or the invasive angio suite could be used in the CT suite," said Carr, an associate professor of radiology and public health at Wake Forest University School of Medicine. "As you get to smaller and smaller vessels, in something that's moving like the heart, there may be real utility in that. We don't know yet, but I think the sky's the limit."

INFORMATION OVERLOAD

Impressive advances in CT technology, however, create major data processing headaches, and they are likely to continue doing so well into the future. To appreciate the scope of the problem, Dr. Geoffrey Rubin suggests comparing a full-body CT scan to the human genome. A 6-foot-tall person scanned with 1-mm-thick slices generates about 720 Mb of data, while the genome, well known for its enormity and complexity, takes up about 750 Mb.

"That gives you some indication of the magnitude of data we're talking about with every single CT scan we acquire," said Rubin, cardiovascular imaging chief at Stanford University. "Tools need to be developed to completely change the way we analyze these data."

Short of linking a supercomputer to every CT workstation, the near-term solution to data overload is a combination of better workflow, better computer-aided analysis, better networking, and finely honed coping skills. One option that imagers are more likely to take advantage of as CT scanners acquire larger and larger data sets is to scan thin but read thick. Initial interpretation of images reconstructed as thicker slices may be sufficient in itself, or it may point to specific, limited areas of anatomy where the reconstruction can be redone using thinner slices.

"This is one way of trying to manage the data explosion that has come with multislice scanners," Westerman said. "Otherwise, you're inundated with a tremendous number of images."

Various quantitative analysis tools also limit the need to review every single image. Instead, the radiologist marks regions of interest, and task-specific software performs the analysis and reports its findings. Coronary calcium scoring is one obvious example, while a growing array of tools analyze angiographic images, organ function, and tissue perfusion. All help radiologists sort through enormous bodies of data, extract information that may not be obvious on visual inspection, and arrive at an accurate diagnosis. These tools are likely to become both more sophisticated and more commonplace in the future.

Improvements in workflow are also critical to managing data overload. Reviewing 15 sheets of film on a motorized viewer-as might have been possible in the days of single-slice CT-is a thing of the past. Instead, with CT studies producing 1000 images or more, radiologists must rely on technological advances to improve efficiency. This might mean scrolling through stacked images on a monitor or taking advantage of 3D visualization techniques.

Perhaps even more critical are behind-the-scenes improvements in how scanners and workstations work. Scan protocols have morphed into examination protocols packed with information, not only about the patient, the physician, and the imaging parameters, but also about how the images should be reconstructed and presented, where the data should be routed, and where and how data should be archived.

"If we tell the scanner we want to do a coronary CTA, the scanner automatically knows how to manage the data. The tech doesn't have to do anything manual," said Markus Lusser, Siemens' marketing manager for high-end CT. "The key benefit of integrating postprocessing into the examination protocols is that the time to visualize vessels in 3D and to do the cardiac studies is minimized to minutes rather than hours."

Interfacing with sometimes sluggish hospital networks remains an unpredictable challenge for all vendors. Jonathan Murray, GE's global manager of premium CT, said that's another reason scanners and workstations must continue to improve the way they handle data and manage workflow.

"You automate it so it's happening in the background," he said. "As natural industry trends work to make network transfer times faster, that's just a bonus. In the meantime, you're already making smart use of your time, your bandwidth, and your archive."

PACS are likely to get a boost from advances in CT, as the need to manage and archive large volumes of data becomes increasingly acute. Even in hospitals without full-blown networks, data overload in the CT suite may tip the balance in favor of a miniPACS, according to Kulp. In the meantime, vendors and users alike watch developments in information technology with a keen eye, hoping that advancements in that industry can come close to keeping up with progress in imaging technology.

"You just keep grabbing the newest thing off the shelf," Westerman said. "We're going to take advantage of every data processing development that becomes available, because the need is just growing leaps and bounds."

Ms. Carrington is a freelance medical writer in Vallejo, CA.