CT vendors promise more than they can deliver

R&D into computed tomography is racing forward at breakneck speed. GE, Siemens, Philips, Marconi, and Toshiba are developing CT scanners that will generate 16 slices with each turn of the gantry, an enormous leap from the current quad-slice capability. Getting these advanced scanners out the factory door, however, is something else again.

Philips has set 2004 as the year in which its first 16-slice products will ship. Siemens expects to have a 16-slice scanner on its loading docks by the end of 2002. Toshiba has targeted 2003.

GE has not yet announced a date for its launch of such a scanner and is probably feeling little pressure to do so. The company went into production with its LightSpeed scanner ahead of all competitors. With that advantage, and with the largest CT installed base in the world, the company has sold more multislice CTs than anyone else: about 700 as of the end of 2000. As competitors were hawking 16-slice detectors on the RSNA show floor, GE pointed to FDA clearance for its eight-slice detector.

"There is lots of talk from competitors about new detector systems, but it's all a reaction to what GE's going to do in the market," said Bill Radaj, GE Americas marketing manager for CT.

But even GE is playing the vaporware game. Radaj said the company will not release the eight-slice version, unveiled at the 2000 RSNA meeting, until the end of 2001. Its FDA clearance was based on a prototype. The production model still needs work and interpretive tools must be developed to assist physicians in the diagnostic process, he said.

"We have lots of engineering milestones that have to be cleared," Radaj said. "We have to bring the product up to spec, and we want to make sure this product covers all the bases."

Vendors are clearly trying to build demand for more advanced scanners as a way to keep their installed base from going to a competitor, as well as to position themselves at the forefront of R&D. The possibility that their efforts to focus customers on the future might stall present sales will be mitigated, say vendors, by the upgradability of their platforms. Toshiba's flagship Aquilion is designed to support 16-slice scanning and beyond. The detector currently shipping with the unit has 34 sensor arrays. Theoretically, each one could be outfitted with a data acquisition system (DAS), turning the product shipping right now into a 34-slice scanner, although it is unlikely that Toshiba would push the product that hard. Siemens' Volume Zoom and Marconi's Mx8000 are each outfitted with detectors capable of supporting at least eight slices per rotation. GE's LightSpeed has a 16-row detector and might be able to support 16 simultaneous slices.

Even if advanced detectors must be built to support 16 or more slices, they will likely be compatible with the CTs now being built. The installed base, therefore, should be able to upgrade to these advanced versions as they become available.

"The reason we wanted to announce (the 16-slice version now) is that we don't want our customers to be surprised," said Bill Kulp, manager of CT marketing for Marconi. "We want them to be able to budget for the new technology and plan for it on their road map, and we want to reassure them that there is a growth plan for the Mx8000 they are buying today."

The further away the release date, however, the greater the chance that something will go wrong, causing the schedule to slip. Vendors are admittedly struggling with the underlying technology. Data acquisition systems are expensive. Adding these data components necessarily

raises the price of the scanner and creates logistical problems, as they must be integrated into the gantry, which is already tight for space. Image reconstruction is especially vexing.

"The biggest challenge is the data pipeline, getting the data off the gantry and then having a computing system that can catch, process, and feed the data into a proper workflow," said Bernd Montag, Ph.D., market segment manager for advanced radiology in Siemens' CT division.

The data from 16-slice scanners will probably have to be reconstructed in near-real-time, or the scanners will be swamped with a backlog of data processing. The computing engines necessary to accomplish such reconstructions are nowhere in sight, which has led vendors to begin studying the development of their own computing platforms.

A more fundamental problem is just figuring out how to reconstruct the data. Vendors today reconstruct image data as slices. This works only up to a point. After eight slices, artifacts begin to creep into the images.

"According to (our) manufacturing division, we will have to go to a Feldkamp (algorithm) when we introduce the eight- or 16-slice version," said Yusuke Toki, CT product manager for Toshiba.

Most vendors are working on a modified Feldkamp algorithm, which is designed to address problems accompanying "cone-beam" reconstruction, simply defined as the means for processing a volume of data. Philips purports to have developed its own solution, called 3 pi, but has provided no details about the algorithm.

If these challenges can be met quickly, enormous benefits could be reaped. A company now at the back of the pack could leap to the front. Philips subscribes to this idea.

The company was left in the dust when multislice scanning came along. Its platform for multislice scanning, CT Secura, was unveiled in 1999 and the multislice detector, called TrueView, was discussed at the 2000 RSNA show (SCAN 12/20/00). Philips is expected to begin shipping a quad-slice scanner in the second half of 2001, but its real focus is on the 16-slice version, due for release in 2004.

"Our customers are not discussing four slices anymore but asking when they will be able to go to eight and to 16," said Gerry Winkels, Philips global marketing director for CT. "The real question is, will it ever end?"

In deciding, as an industry, to make machines with more slices, vendors have committed to the same path with the same milestones. Consequently, their efforts to distinguish themselves through future technologies have only underscored the impressive homogeneity of their current and promised offerings.

Flagship products from all vendors now generate no more than four slices per rotation. Speed of rotation is a half-second. Data matrices for input and output are 5122 for GE, Siemens, Toshiba, and Philips. (Marconi claims the ability to go from 512 to 1024, if the protocol demands it.) All vendors rely on software corrections to deal with artifact created by pitch, which is the speed at which a patient is pushed through the gantry during a scan. The common vernacular is slice-based, with each vendor setting its sights on creating scanners capable of achieving 16 simultaneous slices in three to four years. In the more distant future, the consensus calls for development of area detectors, which will generate volumes of data.

As for near-term strategies, vendors are still reading from the same page. At the RSNA show, GE unveiled a two-slice scanner as part of its HiSpeed line, allowing the owners of single-slice versions to step up to multislice. Siemens introduced a dual-slice version of its mid-tier CT scanner, Emotion. (Siemens already showcased a dual-slice version of the Volume Zoom in 1999.) Toshiba went the extra mile at the RSNA show, describing plans to offer both a dual- and a quad-slice version of its mid-tier Asteion. Even Marconi, which has had a dual-slice version in its lineup since acquiring the Twin from Elscint more than two years ago, is coming out with a new dual-slice scanner. This one will be built into the Mx8000, thereby affording customers the option of buying into the company's premier CT platform, with its inherent upgradability, at a budget price.

The proliferation of low-cost dual-slice scanners means the obituary for single-slice scanners is already written, at least in mainstream radiology. Only in nontraditional markets, such as hospital departments outside radiology, small clinics, and physicians in private practice, will there be demand for this older-generation technology.

Consequently, multislice imaging will be the focus of vendors going forward. Upgradability will be an essential element. As vendors promise more powerful machines with many more slices than they can currently supply, meeting self-imposed expectations will be their greatest challenge.