Technology offers promises of newer, faster applications and challenges of data management

MDCT reshapes imagers' workflow and practice

By: Charles Bankhead

In a span of little more than two years, multidetector CT has become the industry standard. The enhanced image quality and faster scanning afforded by MDCT compared with conventional single-slice technology have attracted a clinical following that extends far beyond the radiology community and into virtually every medical specialty.

"There is no doubt in my mind that multislice CT will soon be the platform that everyone will use," said Dr. R. Brooke Jeffrey, chief of abdominal imaging at Stanford University. "The cost is not that different from a single-slice scanner, and the performance is so much better that you would never think of upgrading and going to any other technology. It will soon be the only game in town as far as CT is concerned."

Single-slice CT machines will remain a useful component of clinical care for several years, until their natural lifespan passes, said Dr. Elliot Fishman, director of diagnostic imaging and body CT at Johns Hopkins University. When an institution has funds to acquire a new CT scanner, however, that machine will be an MDCT.

"When I put on a course at a meeting and ask how many people already have multislice machines, about half raise their hands," he said. "When I ask how many people are getting multislice machines within the next year or two, the rest of them raise their hands."

While the imaging superiority of MDCT is unquestioned, the rapid expansion of multislice technology has raised a number of economic, technological, and clinical issues. The first is how to make the most of a mix of single- and multislice machines during the transition to an MDCT-only environment. Radiologists seem fairly confident that the clinical cases best suited to the strengths of MDCT-such as CT angiography-will migrate in that direction, while more straightforward cases can be served adequately by single-slice scanners until they are phased out.

Less clear is how to deal with the massive amounts of data generated by MDCT. Major unresolved issues relate not only to data storage but to access and transfer across intranets, remote access, and viewing of images that can run into the thousands for a single MDCT study.

"No one has any great solutions. Right now, people are talking more about the problems than the solutions," Fishman said.

Angiography and oncology are currently driving increased use of MDCT, but other applications will soon join the push, according to Jeffrey. Compared with conventional CT, MDCT offers better diagnostic quality and thinner collimation.

"Certainly, multislice CT improves angiography, and the speed of acquisition is critical to that," he said. "In oncology, the thinner collimation leads to much better imaging for pancreas cancer staging and liver tumor evaluation, both of which are clearly improved with multislice studies."

Trauma represents another area ripe for MDCT, Jeffrey said. The new scanners offer the capability to image the head, cervical spine, and thoracoabdominal area with a single contrast injection and in a reasonable amount of time, avoiding the tube cooling problems of single-slice machines.

A huge cardiac market already exists for MDCT, said Dr. Mark Baker, head of abdominal imaging at the Cleveland Clinic Foundation. Applications already in use at his institution include pulmonary vein mapping before patients undergo ablation procedures for cardiac arrhythmias, perfusion studies in stroke patients, and CTA. Cardiovascular and neurovascular applications account for more than 70% of the MDCT volume at the Cleveland Clinic.

"We don't do classic CT angiography because nobody here really cares for just an angiogram," Baker said. "We do a lot of presurgical planning for partial nephrectomies and liver resections. We do all of our living related renal donors with multidetector CT. Really, what we do is CT angiography visualization."

NEW STUDIES

The enhanced imaging capabilities of MDCT afford the potential to markedly improve studies over what was possible with single-slice CT, said Dr. Benjamin Yeh, an assistant professor-in-residence in the abdominal imaging section at the University of California, San Francisco. As an example, he cites CT colonography.

"For the time being, I think the major advantages of MDCT will be most apparent in vascular imaging and for studies such as CT colonography," he said. "The potential to do new types of studies such as perfusion imaging that previously couldn't be done with single-slice CT is especially exciting, although we are also very pleased with the excellent vascular imaging studies we can perform."

Yeh said MDCT offers CTA images that are competitive with conventional angiography in terms of accuracy. And it introduces new diagnostic dimensions not possible with conventional angiography.

"We can look at the thickness of a thrombus in the aorta or in the carotid artery," he said. "Obviously, we cannot do interventions, like we can with conventional angiography, but we can image more accurately the shape of the lumen and thrombus. This would be of great benefit in presurgical planning."

Aside from vascular imaging, conventional CT remains adequate for most types of CT studies.

"For simple clinical things, such as ruling out an abscess, bowel obstruction, appendicitis-where temporal resolution and speed of acquisition are not critical-single-slice CT still works fine," Jeffrey said. "A single-slice machine is fine for something like head CT, which does not have to acquire data in a very rapid mode. Manufacturers have been slow to adapt multidetector technology to CT fluoroscopy, which we use extensively for biopsy work. Because of that, we have delayed upgrading fluoroscopy to a multislice room."

Good imaging triage is key to making the best use of MDCT and single-slice machines during the transition to an all-MDCT environment, Fishman said.

"The important thing to remember is that the single-slice machine is still a pretty good scanner," he said. "During the period when you might have both types of scanners, it's important to triage, which often is a matter of moving patients around. We have a single-detector machine next to a multidetector machine, and that has worked out pretty well."

At UCSF, the older spiral CT machines tend to be located in outpatient settings, where many routine studies without the need for MDCT capabilities are performed, Yeh said.

As institutions make the transition to MDCT, physicians will recognize the superiority of the newer technology even for routine studies, according to Fishman. He said multislice CT improves accuracy across the board in terms of sensitivity and specificity for the classic applications, such as the liver and pancreas.

PRACTICE ECONOMICS

Direct acquisition and imaging costs of MDCT do not differ strikingly from those of single-slice technology.

"Depending on what you buy and whom you buy it from, the price of a multislice scanner is probably the same as what you paid for a single-detector scanner when it was new and what you paid 20 years ago for a 10-second scanner," Fishman said.

Reduced personnel requirements for certain applications might even make MDCT a cost-savings investment in some situations, according to Yeh. Conventional angiography, for example, requires direct involvement by a nurse and an attending radiologist and may be time-consuming. In contrast, CTA does not require direct monitoring by a nurse or radiologist and can be performed in minutes without the need for anesthesia.

Faster scanning creates the potential for increased throughput for at least some applications, according to Fishman.

"Throughput can increase by 40% without doing a whole lot of work and without changing people or workflow, just the scanner itself. We have seen that in the hospital and the outpatient setting," he said.

A study at Massachusetts General Hospital showed a 51.9% increase in the number of CT studies performed with an MDCT machine versus a single-detector machine during the peak imaging period (9 a.m. to 5 p.m.). Productivity and throughput increased significantly despite the fact that multislice studies tended to involve more complicated imaging protocols (AJR 2001;177:769-772).

The ability to improve productivity with MDCT depends on factors other than the speed of the machine itself, Baker said. The type of patient and the type of study, for example, will influence throughput.

"If you can do patients faster, you probably have a better process," he said. "Some things about the process won't change that much just because you have a multirow scanner. If you're doing a perfusion study on a stroke patient, you don't get those patients on and off the scanner very fast. There is no question that a four-slice scanner is faster than a single-slice scanner in terms of how many patients you can do. But it's doubtful that a 16-row scanner is faster than a four-row scanner. You can do more slices and reconstruct more images and get better data sets, but I'm not sure you're talking about more than a few minutes faster than a four-row scanner."

Yeh goes even further in discounting gains in productivity with MDCT.

"It's a misconception that you can scan substantially more patients," he said. "We have noticed little difference between multidetector and single-detector CT. Throughput doesn't increase that much. Throughput depends to a great extent on how quickly you can get patients on and off the scanner, and we have not noticed much of a change in that area."

STORAGE NIGHTMARE

Although radiologists may quibble about the impact of MDCT on productivity, no one disagrees that along with superior imaging quality have come major challenges with respect to storage and use of data and images generated. Jeffrey said the need for a PACS is a foregone conclusion for successful transition to MDCT. Organizations that do not already have PACS in place can no longer avoid taking that step and assuming the associated costs. The volume of images generated by MDCT makes film totally impractical.

PACS is just part of the solution, according to Fishman. The transition to MDCT introduces larger issues involving an institution's infrastructure. A typical spiral CT study generates 80 to 100 slices. A four-slice CT machine generates 150 to 500 slices. Eight-slice scanners generate twice as many slices, and the new 16-slice scanners increase the output by a factor of four over the four-slice machines. Johns Hopkins will install four 16-slice scanners over the next several months.

"With a 16-slice scanner, you generate 500 to 2000 slices," Fishman said. "You're talking about between 800 MB and 1 GB per case. Where do you store all that data?"

At the Cleveland Clinic, the radiology department already generates 15 TB of data annually, and the department is not yet fully digital, Baker said. If operated to full capacity, MDCT scanners at the clinic can generate an additional 12 to 13 TB. Even when compressed to a 2:1 ratio, the data output is 6 to 7 TB annually, which he calls "not small change." Moreover, compression involves inherent loss of information.

Johns Hopkins performs 60,000 CT studies a year, creating the potential for 600 TB of data, Fishman said. The evolution of MDCT into the standard for CT imaging will necessitate a change in an institution's entire workflow.

"In some ways, CT has always been the same: You have a patient whom you scan with the scanner, then you reconstruct the data, film, and read," Fishman said. "Now you have postprocessing and large data sets, and the whole paradigm will change a lot. I think that's very exciting. You never think about storage as being a problem, yet that's one of the biggest challenges we face."

At UCSF, the major storage problem relates to the raw data generated by MDCT studies. The solution has been to keep raw data in temporary storage for two or three days, then jettison the information to free up storage space. In the process, radiologists lose some capability to manipulate and refine images.

"We simply don't have the capacity to keep all the raw data," Yeh said. "We end up keeping just the image data, which we store electronically. Obviously, we can still perform rudimentary manipulations on this electronic data, but we can't, for example, change the reconstruction algorithm or decrease the slice thickness."

Access to the data and images generated by MDCT poses another set of challenges. Institutions have just begun to grapple with what Baker calls a traffic jam of data, whether accessing images internally or remotely.

Transferring data across a network requires substantial bandwidth, which many institutions' intranets lack. Remote access poses additional problems, as downloading images via modem or even DSL will prove impractical. Viewing images generated by MDCT represents still another challenge. With conventional CT, scrolling through images on a computer screen has been the common approach to viewing. With MDCT, scrolling becomes impractical.

"If you have 200 images, scrolling hurts. If you have 2000 images, you can't scroll," Fishman said.

Though no one has come up with an ideal solution for storage and access problems, conversion to 3D volume data sets offers potential to solve problems and to change the very nature of CT imaging.

"I think the data coming off the scanner will be 3D, like volume data sets or 3D maps, and we will then be able to interact with the 3D volume and get around a lot of problems with slow loading and storage," Fishman said. "This will change CT down to its core. It will reflect and change how we look at and use data.