A Slice of the Future: CT Harnesses New Technologies

Medical Device & Diagnostic Industry Magazine
MDDI Article Index

An MD&DI May 1999 Column

Developments in computed tomography have the potential to dramatically alter current medical practice and reinvigorate a lethargic imaging market.

Computed tomography is making a bid to recapture the high ground of medical imaging. A decade has passed since any significant changes have been made in either the technical or clinical capability of this technology, which some still refer to by the archaic acronym CAT (computerized axial tomography). Very soon, however, a new name—multislice CT—will enter the lexicon.

This next-generation imaging technology creates slices of the body that are similar to ones first created some 25 years ago when CATs were introduced. The computer-generated slices cut across the body from side to side. But the way in which these machines make their slices, the number of slices they produce, and the resolution of the images are very different and promise to dramatically change the practice of medicine.

Each of the new scanners being offered by GE Medical Systems, Siemens Medical Systems, and Picker International acquire four slices simultaneously, using advances in detectors, electronics, computer processing, and mechanical assemblies. A scanner to be produced by Toshiba is expected to offer approximately the same capabilities. Together, these developments could reinvigorate a lethargic marketplace.

Over the past several years, CT sales have shifted decidedly from high-end systems toward competitively priced midrange units, which typically hover around $600,000 to $750,000. The year 1997 was a watershed for the industry. GE executives estimate that whereas 1997 sales in the premium end grew a mere 12%, sales in the low end rose 27% and midrange sales leaped 72%. The introduction of multislice CTs could rekindle interest in premium-performance scanners.

"We've seen the awareness of CT grow exponentially in the entire medical imaging marketplace [since the launch of these multislice scanners]," says John Sandstrom, PhD, director of strategic marketing and CT for Siemens Medical Systems. "People want to know about multislice technology because they want to figure out where it fits in the big picture. They want to position themselves for the future."

The scanners sparking this introspection go by glitzy names—LightSpeed QX/i, from GE; Somatom Plus Volume Zoom, from Siemens; Mx 8000, from Picker; Aquilion, from Toshiba. All of these companies have machines available for sale in the United States, except for Toshiba, which is still developing the final version capable of multislice. Available systems are priced at about $1.25 million per scanner, although Siemens, whose U.S. operations are based in Iselin, NJ, offers as an alternative a $400,000 upgrade for the 1300 installed Somatom Plus 4 scanners now in use worldwide. GE, Siemens, and Toshiba developed their new scanners in-house, whereas Picker acquired the Mx 8000 as part of a $275 million purchase, completed in 1998, of the CT division of the Israel-based company, Elscint Ltd.

The Picker and Siemens systems use many of the same components—the result of a collaborative engineering agreement originally struck between Siemens and Elscint. By purchasing the Elscint CT division, Picker gained access to the hardware and algorithms that were part of the deal.

Regardless of the manufacturer, multislice scanners are remarkably alike in capability. All are designed to generate four slices per rotation of the x-ray source. Rotation speed is less than a second for all scanners—0.5 seconds for Siemens, Picker, and Toshiba; 0.8 seconds for GE.

COMPETING DETECTORS

The cornerstone of multislice CT is the detector. GE relies on its HighLight detector material, a ceramic used in the company's conventional single-slice scanners. As part of their collaborative agreement, Picker and Siemens use essentially the same detector made from a new ceramic, called Lightning UFC (ultrafast ceramic), which allows a very fast response time to x-rays.

The Siemens/Picker detector is 40-mm wide, with eight rows of sensors or elements. The rows vary in width: the two center rows have a width of 1 mm, flanked by two rows at 1.5 mm, two more at 2.5 mm, and two outer rows at 5 mm.

The GE detector extends 20 mm and comprises 16 rows, each one 1.25 mm wide. The Toshiba detector, when completed, is expected to comprise 32 rows, each 1-mm wide, thereby extending a total of 32 mm.

Slice thickness depends primarily on beam collimation and the reconstruction algorithms being used. The beam on the GE LightSpeed, for example, might be collimated to shine on all 16 rows, generating data that are reconstructed into four 5-mm slices (Figure 1). Narrowing the beam to fall on just eight rows would still generate four slices—but slice thickness would be 2.5 mm. Because of the varying widths of rows on the Siemens/Picker detector, operators have more flexibility in choosing slice thickness, but the principle is the same. Using all the rows on the detector produces thicker slices and, therefore, the fastest scan time. Narrowing the beam to use fewer rows creates thinner slices and higher resolution, but reduces the speed and coverage.

Figure 1. Early arterial-phase (top) and portal-venous-phase (bottom) 5-mm multislice CT scans of the liver were taken 38 seconds apart. Images courtesy of GE Medical Systems (Milwaukee).

Details regarding the Toshiba detector are sketchy prior to release of the scanner, scheduled for sometime in mid-1999. But sources at Toshiba's U.S. headquarters in Tustin, CA, indicate that the detector will operate in much the same way as competing systems.

One of the most daunting challenges was building the technology to make use of the captured data. GE went from one workstation to two; developed application-specific integrated circuits and special microprocessor controls; wrote 8 million lines of software code; developed an automation index; and created multislice algorithms along with a proprietary data-acquisition system specially suited to the task (Figure 2). These developments were necessary "because nobody out there had the complexity that we needed in terms of switching technology or data sampling," says Vivek Paul, general manger of Global CT at GE Medical, which is based in Milwaukee.

Figure 2. The LightSpeed QX/i from GE Medical Systems features a detector that extends 20 mm and comprises 16 rows, each 1.25-mm wide.

The battle to win customers will likely focus on technological details. Vendors are already trying to differentiate their product by the number of rows (e.g., GE has 16 and Siemens and Picker have 8; Toshiba will have either 32 or 34). Implied by such comparisons is that the company might expand the number of slices to the number of rows, creating "many slice" rather than multislice scanners. While technically possible, doing so would be a retrofitting nightmare, with manufacturers having to add a digital acquisition system for each new slice, an undertaking of extraordinary proportions. Both the detector and electronics for gathering data would have to be replaced, at a cost that, at the present time, is not justified.

Consequently, four-slice scanners will likely be the gold standard at least for the next two to three years. The primary goal will be to build customer acceptance and, ultimately, demand for these products. Manufacturers will do this by framing the four-slice scanner as the harbinger of a new era. During 1999, the industry will ramp up to full production of its four-slice scanners, then take a step backward and fill the gap between this new technological tier and conventional scanners with an economical alternative—the dual-slice scanner. These will be unveiled later this year or in early 2000, probably by GE and Picker.

This trickle-down strategy is nothing new. Since the mid-1990s, manufacturers had been migrating spiral (also called helical) CT technologies from their high-end systems to lower-priced units. These spiral scanners produce slices, but because the patient is moving continuously through the gantry, the data are acquired in a spiral (or helical) pattern. The computer interpolates the data to create a volume, which is then reconstructed into slices depicting a target area of the body.

Spiral scanning changed the way CT was done in the United States, supporting more-accurate diagnoses through improved resolution and vastly reducing the time needed for an examination. Multislice technology promises a similar change.

"We have some great technology with some great technical features and capabilities," says Sandstrom. "The big question is how are we going to translate that into clinical practice?"

POTENTIAL APPLICATIONS

Clinical studies are now getting under way at sites across the United States and Europe to determine the potential of multislice CT. Some opportunities are inherent in the technology. Breath holds are much shorter and, therefore, less stressful to patients. The scan time on the GE LightSpeed to acquire chest/abdomen/pelvis images, for example, is cut from 2 or 3 minutes on a conventional single-slice scanner to 20 seconds. Organs can be completely covered in a single scan, capturing arterial and venous blood flow in CT angiograms.

Thinner slices will allow better image quality and the potential for more accurate imaging, but some physicians may be more than happy to stay with resolution comparable to current scanners to get a much faster scan. Gary Gazer, MD, professor and chairman of the department of radiology at Stanford University, expects the LightSpeed to pave the way for CT to replace more time-consuming or more invasive procedures.

"A physician who deals with trauma patients in an ER setting, where it is difficult to get a full assessment of the body, will find that this scanner should be able to get the answer in a minute or minute-and-a-half," says Gazer, who was among the first physicians to use the GE scanner.

Figure 3. CT angiography of the aorta and peripheral vasculature is becoming possible using fast, new-generation scanners (GE Medical Systems).

Physicians using the new GE scanner hope to harness its speed to allow liver-cancer screening by comparing arterial- and portal-venous-phase images of the liver. They are also examining CT angiography of the aorta and peripherals, including the legs, and techniques for rapidly assessing stroke patients (Figure 3). Another focus is virtual endoscopy, by which CT data are reconstructed into a 3-D view similar to what would be seen with an optically based endoscope. A major challenge has been to control costs for such virtual exams, as imaging time on a CT is more expensive than with an endoscope. GE believes that with its LightSpeed, up to five or six patients could be examined per hour, helping the case for economic justification. Even on routine exams, multislice scanning at the thinner slice levels holds potential for finding lesions that would otherwise be missed.

GE began working with prominent radiologists early in the design process, says Ken Denison, manager of Americas CT marketing at GE. "That allowed us to identify issues important to the end-users and to determine what the scanner had to do—up front—and get that built into the software," Denison says.

The company relied on simulations to help define requirements for processing power, data flow, data pathways, and software, says Denison. "We do all that well in advance of starting to build anything, because that way you get the flexibility you need," he says.

Siemens expects its collaborating radiologists to focus on applications in trauma, including stroke, pain management, and cardiac imaging. Most of the work so far has been done at research sites in Germany. In addition to cardiac images, the company, through its clinical investigators, is looking into quantitative measures such as ejection fraction. Siemens also expects Volume Zoom to be a factor in bringing CT angiography into the clinical mainstream and is focusing specifically on the potential of imaging the coronary arteries in 3-D.

Figure 4. The Mx8000 multislice CT scanner from Picker International can acquire up to four spiral slices simultaneously in a mere 0.5 seconds, enabling head-to-toe scanning in less than 30 seconds.

Radiologists using the Mx 8000 by Picker, which is based in Cleveland, are focusing on trauma and ER applications, pulmonary embolism, and cardiac imaging, as well as a range of CT angiography applications (Figure 4). Opportunities in cancer treatment may be found regarding the staging of the disease. The Mx 8000 holds promise for pediatric exams, according to the company, in that its high-speed scans might eliminate the need for sedation. Brain perfusion for stroke management and hepatic perfusion to assess liver health are also being explored.

"We have learned to collaborate with clinicians and have them help us design the products," says Gary A. Kaufmann, director of marketing and sales for the Picker CT division. "We then follow up with them through development stages."

Interventional CT has enormous potential and will be examined by radiologists for each of the major vendors. In this application, physicians use minimally invasive techniques to biopsy tissues or treat patients. The primary reason for exploring this area with multislice CT is the speed of these new scanners and the potential of increased resolution.

TECHNICAL CHALLENGES

Opening the door to new opportunities will create enormous new challenges. By creating four times as many slices as a conventional scanner, these new CTs could literally bury physicians in slices—up to 800 might be produced in a typical exam. Radiologists shudder at the thought of having to plow through so many images to make a diagnosis.

One answer being proposed is the use of artificial intelligence, which might identify anomalies and direct the physician's attention to specific images or even areas within these images. There may be a psycho-political barrier, however, in that intelligent pattern-matching routines can diminish the role of physicians. The sensitivity of the issue is indicated in the insistence of companies now developing such algorithms for use in breast exams that their pattern-matching algorithms be used only as part of a "second-reading" methodology.

A more likely answer to the data overload coming from multislice technology may be found in a technique that, until now, has been mostly a medical curiosity—3-D imaging. Although this technique is used routinely at some institutions for surgical planning, its diagnostic impact has been virtually nil. In multislice CT, however, 3-D reconstruction might allow the physician to comprehend the entire volume of data and then focus just on specific segments that look pathologic. Physicians are not likely, however, to make diagnoses directly from these reconstructions, at least not for a while.

"I think our minds are open to new ways of looking at images, and 3-D certainly appeals to me," says Kenyon Kopecky, MD, professor of radiology at Indiana University (Indianapolis) and an early user of the Picker Mx 8000. "But I am not willing to give up two-dimensional images." Vendors are now working on processing and display technologies that will allow diagnosticians to extract 2-D slices from the 3-D volume for closer inspection.

Three-dimensional reconstruction might also be the answer to handling images produced while doing interventional procedures. Future techniques might involve automatic tracking systems for visualizing the area immediately surrounding the interventional probe as it moves through tissue. This technique might also be used to highlight target tissue and critical anatomy en route to the target.

As multislice scanners usher in a new era of clinical applications and image display, the stage will be set for the next great leap forward—cone beam reconstruction. This technique, already being researched by the major vendors and some universities, could expand the width of the detectors from the current 20 to 40 mm by a factor of 10 or more. Algorithms will interpret the data, not as slices, but as whole volumes. Ultimately, an entire segment of the patient will be constructed in a single rotation.

These area detectors will evolve from the solid-state arrays being built now for use in x-ray imaging, namely radiography and fluoroscopy. The day when they will be able to deliver the necessary resolution or data acquisition speed, however, is still far off—at least five to seven years.

When cone beam reconstruction capability does arrive, the technological demands, especially for data processing, will be unprecedented. Reconstruction speeds will need to be in the millisecond range to keep up with the data stream. But, just as the challenge will be great, so will the potential for clinical benefit in a medical realm that will have been reshaped initially by the introduction of multislice scanners.

Copyright ©1999 Medical Device & Diagnostic Industry