Interpretations fuel vendor debate over performance of 16-slice scanners

May 1, 2002

Siemens and Philips share hardware but claim distinctions What a difference a few hundredths of a second can make. At the last RSNA meeting, leading CT vendors showcased 16-slice versions of their flagship products. The decision to

Siemens and Philips share hardware but claim distinctions

What a difference a few hundredths of a second can make. At the last RSNA meeting, leading CT vendors showcased 16-slice versions of their flagship products. The decision to do so was hardly surprising. The industry had been forecasting 16-slice scanners throughout 2001. The only twist was the acquisition of Marconi Medical, which catapulted Philips Medical Systems into a multislice leadership role along with Siemens, GE, and Toshiba.

With these four vendors neck and neck (and neck and neck) in the race to win customers, questions have arisen about some specifications. Most intriguing have been differing specs on Philips' Mx 8000 and Siemens' Sensation 16, the only two 16-slice scanners now commercially available.

At the RSNA meeting, Siemens staff boasted a 0.4-second rotational speed for their new Sensation 16. That now appears to have been a "rounded" number. Siemens' Web site claims a 0.42-second rotation time. The difference, while seemingly minor, nevertheless gave Siemens the edge over Philips, an edge that Siemens does not deserve, according to Bill Kulp, manager of Philips CT marketing.

"There is an awful lot of 'mine is bigger than yours' in the CT market," Kulp said. "And if theirs really were bigger than ours, we could take our lumps and live with it. It is distressing, however, to see our competitor credited with a bigger number when it is the result of 'marketing rounding' and not better technology."

DI SCAN reported the Sensation 16's rotational speed as 0.4 second rather than 0.42 second and, applying simple division and multiplication, credited Siemens with two more slices per second (SCAN 12/26/01) than, in fact, it's capable of. (The article was summarized in the March issue of Diagnostic Imaging magazine, leading Philips to cry foul.) But Siemens executive Markus Lusser says rotational speed is open to interpretation. The Sensation 16 can actually complete a rotation in 0.38 second. Siemens chose 0.42 second for cardiac scanning. In radiology mode, the scanner rotates more slowly, at 0.5 second.

"There is a correlation between heart rate and temporal resolution when doing cardiac scanning," Lusser said. "The 0.42-second rotation is optimum for hitting the average heart rate. When scanning other body regions we do not have to go at this speed."

The debate over the relative rotational speeds achieved by the Mx 8000 and Sensation 16 is interesting because the gantries onboard these 16-slice CTs are mechanically identical, making it all but impossible for one company to eke out a faster rotational speed than the other, if both decided to go for speed alone. In fact, most of the hardware on these two scanners is exactly the same. They differ only in their computing engines and the software used to reconstruct the acquired data.

This family likeness is the result of a six-year-old agreement that first linked the engineering groups of Siemens and Elscint (SCAN 9/11/96). The Israeli company was the first to commercialize the idea of multislice CT with its release in 1992 of the CT Twin (SCAN 2/12/92). Picker International partnered with Siemens after purchasing Elscint CT assets in 1998 (SCAN 9/16/98). Philips entered the picture last year with its acquisition of Marconi Medical (SCAN 11/31/02).

With each successive corporate move, the collaboration appears to have become less advantageous for Siemens. When the agreement was initially forged, Elscint had a sterling reputation for innovative engineering but lacked the panache in marketing and service to become a serious contender in CT, a weakness underscored by the decision of management to exit the imaging marketplace in the late 1990s. Picker was not encumbered by such weaknesses and availed itself of well-established facilities and headquarters in the U.S. from which it could address the lucrative North American marketplace for high-end CT.

The engineering excellence of the Mx 8000 and its position in the marketplace drove Philips' decision to acquire Marconi Medical (formerly Picker International). Philips has embraced the Mx 8000 product line, which now includes dual-, quad-, and 16-slice versions, as a critical part of its strategy to make the Dutch firm the leading worldwide vendor of radiology equipment.

Lusser and Kulp agree that the partnership has given them a boost in their battle to win market share from the leading radiology vendor in North America, GE Medical Systems. But the advantages are less clear-cut on the other side of the Atlantic.

"In Europe, Philips and Siemens are pretty evenly matched and GE is kind of the third-place guy," Kulp said. "Philips and Siemens are big rivals."

The acquisition of Marconi Medical has intensified that rivalry. CT had been the weak link in Philips' chain of imaging products. For years, the company had depended on scanners built by Hitachi. The decision by GE, Siemens, Toshiba, and Marconi to launch multislice scanners caught Philips flatfooted. Its purchase of Marconi Medical made Philips a contender in head-to-head CT competitions, while providing the company with the opportunity to win orders that require CTs to be bundled with other equipment, such as MR, nuclear medicine, and vascular x-ray devices.

"It has been a huge change coming from Marconi," said Kulp, who served as CT marketing manager at the Cleveland-based company before Philips bought it. "When we were there, we had to explain why people should take a chance with us before we could even start talking about the scanner. Now people come to us and say, 'I am a Philips customer and I want to buy your CT.'"

Neither Siemens nor Philips can formally comment on the particulars of the collaborative agreement that guides the codevelopment of their flagship CT products. (The contract forbids public disclosure by the two companies, according to Lusser and Kulp.) They can, however, discuss the ramifications and challenges inherent in the agreement.

Lusser contends that there is enough difference between the CT offerings of Siemens and Philips to give the German company the upper hand. Siemens has developed a very fast reconstruction algorithm that creates a true 3D reconstruction of the volume, he said. The algorithm records the data as 2D slices, then reformats them to account for artifacts and factors such as pitch (the speed at which the patient moves through the gantry).

"Actually, if we invented the CT today, the definition for 'pitch' would not exist, because in our scanners today you no longer adjust the pitch," he said. "You adjust the acquisition time, depending often on the patient's ability to hold his breath. Everything else is done automatically."

Kulp focuses on Philips' algorithm, which is more time- and computing-intensive but, he said, creates a true "cone beam" reconstruction. This type of reconstruction, which is based on mathematical theory pioneered some 30 years ago, accounts for differences in the angles between the detectors and the x-ray tube when the beam passes through different parts of the body.

"We have images that show the differences," Kulp said.

The trade-off is time. The Siemens algorithm reconstructs about six images per second; the Philips algorithm reconstructs four per second. The time difference adds up when doing hundreds of slices. And, when cardiac scanning enters the mainstream, which could occur in the next several years, the scanners will have to generate thousands‹not hundreds‹of slices.

Clinical results, Kulp and Lusser agree, are what customers want and what both Siemens and Philips strive to deliver. But the differences between multislice scanners at the current level of performance are so slight that they may be inconsequential. When two pieces of equipment both deliver excellent results, marketing staff must get creative, if they are to give their sales staff ammunition. And that may happen even with the hard numbers that constitute product data sheets.

"If you look at the product data sheet on the Sensation 4 (Siemens' quad-slice scanner), you will see 30 line pairs per centimeter. We say the Mx 8000 (quad-slice) delivers 24," Kulp said. "Once again: same geometry, same scanner. How do they do that? One number is theoretical; the other is what you see with a phantom. It all comes down to how you talk about the product."

Both companies take poetic license in narratives about their equipment. On its Web site, Siemens reports "unmatched" volume acquisition and "unmatched" temporal resolution with its Sensation 16, even though these capabilities are mechanically derived‹and are shared by Philips' superpremium Mx 8000 system. On its Web site, Philips claims the Mx 8000 delivers an "unprecedented" 16 simultaneous slices when the scanner is outfitted with the 16-slice Infinite detector, even though the same detector is available on the Sensation 16.

Perhaps the word the two companies are looking for is "unsurpassed," which will work at least as long as they are the only two vendors shipping 16-slice CTs.