The name of the game is speed. Facilities upgrading to the latest generation of 16-, 32-, 40-, 64-slice and dual-source CT scanners have quickly learned that boosting clinical capacity in CT goes hand-in-hand with escalating network depth and data storage to smooth access to the torrents of data these scanner create. Fine-tuning workflow also is key to manipulating and managing the larger volumes of images generated — and more quickly — the higher the slice count climbs. Adaptive facilities are sending automatically generated multi-planar reformatted images from the scanner directly to PACS — which allows radiologists to get 3D data faster and avoid a trip to the workstation. Expectation is high for quick results throughout the healthcare enterprise, especially from emergency departments dealing with trauma cases.
Avoiding multidetector shock
You don't need to have a 64-slice CT system to see a huge jump in data. That happens as soon as you leap into the multislice arena.
"The reality is that once you've gone from a single- or dual-slice system to 16 slices, you have already gone over your biggest hurdle," says Pedro J. Diaz, PhD, vice chair of the Department of Radiology at Case Western Reserve University/MetroHealth Medical Center in Cleveland, Ohio, a Level 1 trauma center. MetroHealth's department performs nearly 50,000 CT studies yearly, many of which serve the hospital's emergency department. They currently have four Philips Brilliance CT systems, three of which are 16 slice scanners and one 64 slice. Hence, Diaz has a pretty firm understanding of how to deal with load after load of CT images.
In his view, the initial move to multidetector scanners is challenging largely because the older single-slice systems were limited in how small a slice they could acquire. While typical acquisitions with single/dual systems were performed with slice thicknesses between 2 mm and 7 mm, the move to 8- or 16-slice scanners brought acquisition slice thicknesses fell to 0.625 mm or 0.5 mm.
"For a given anatomic protocol, the accompanying increase in the number of slices generated could be dramatic," says Diaz. "However, minimum/typical slice thickness has remained fairly constant as we've moved from 16- to 32- to 64- and higher-slice CT scanners. As such, while advanced applications on 64 and larger scanners will add to the storage burden, if you've solved the basic storage problem with the 16-slice scanner, you've pretty much solved it for larger-slice scanners."
The biggest change is that you're just scanning more quickly. Additionally, when you go from 64 to 128, you add more applications, but the slices remain essentially the same. And generally speaking, the new applications largely pose a workflow obstacle rather than a storage one.
"In more and more instances, we generate thin-slice isotropic data that we will want to store indefinitely, and we also do more post-processing, which requires easy access to this data as well as additional storage," says Diaz.
Now that you've got all of these data and images — including more post-processed images such as 3D — where do you put them, and for how long? The decision at MetroHealth comes down to what radiologists need to see in PACS to review axial images. For a number of the imaging protocols, they will use the thicker slices that are kept in the PACS for five years. The thin slices — used for 3D rendering post-processing — are kept in a temporary processing store for three months.
Ideally, everything — including thin-slice isotropic data — would end up in PACS, says Diaz. Storing that much data isn't the challenge it used to be, as sophisticated storage area network (SAN) and network attached storage (NAS) costs have dropped in price. MetroHealth worked closely with its PACS vendor, GE Healthcare, to deploy mirrored EMC Centera NAS devices, which can be upgraded in capacity easily, for long-term storage. For faster, short-term storage, about a year's worth of data are cached on a RAID (redundant array of independent disks) subsystem.
Additionally, while hammering through tons of CT data, workflow is another weapon to hurry things along. "I'm a big advocate of making it easy for everyone to access thin-slice data and post-processed images from PACS. On our CT scanners, a trained technologist is able to do most of the required post-processing as part of an acquisition protocol," says Diaz. "However, there are instances where a radiologist may want to do some post-processing at the PACS workstation and/or when additional renderings by the technologist are required. Our referring surgeons also access our data for use in 3D surgical planning. Having the required data available in PACS greatly simplifies this process and increases our overall efficiency."
Speed is of the essence
"People are always calling me and asking that I do their CT exams on the 64-slice, and I tell them, 'I'm sorry, I can't do anything else,'" says Bill Shuman, PhD, professor and vice chairman, director of radiology, Department of Radiology at the University of Washington Medical Center (UWMC) in Seattle — a 475-bed hospital with two large outpatient clinics, one dedicated to cancer (Seattle Cancer Care Alliance) and the other dedicated to subspecialties.
The Medical Center is definitely up on the latest and greatest in CT, with three GE LightSpeed VCT 64-slice systems. The initial system was installed about 16 months ago. This was a period of housecleaning for the department as they ushered in the 64s and got rid of older 1-, 4-, and 8-slice systems.
Speed is of the essence with these new scanners that crank out a "sea of data" as Shuman calls it. To help them coordinate the images, each CT scanner is tied in to a GE Centricity PACS 2.1.
All in all, their considerable network handles around 110 CT studies daily in addition to images from PET, PET/CT, and four MR scanners. If that's not enough, they operate three angiography suites, six digital radiography units and six computed radiography systems, and 14 ultrasound rooms. "We're making a lot of data," says Shuman.
Thinking about the costs and logistics of storing so much information could keep even the soundest sleeper awake. But, if they have been paying attention to the costs of fast RAID storage, some of their worries would vanish.
"What we're finding is that the cost of storage is dropping at the same time the use of storage is going up," Shuman says. "It's uncanny, but it's true. The number of slices will go up the same amount the cost of storage goes down. For the last three years, that's been true."
Using what amounts to three years on RAID (backed up by EMC Centera off-site) all of their data are available around the clock, 24/7. "It's all sitting in RAID, and that's it," says Shuman, and they never expect this to change because, again, "the growth of slices is exactly matched by the decrease in cost," he adds like a mantra.
With their existing storage, UWMC holds onto body images — axial 2.5s and coronal 2.5s — in the archive as well as their source images. The workstations are loaded with GE's Advantage Windows Suite with dual-processor boxes that can run both the PACS software and the 3D Advantage software.
"So if I'm looking at a case and want 3D, with one mouse click I've got all the 3D tools right there available for me. I need the source images to be on the PACS workstation to accomplish this. As soon as I do that, they are a part of the PACS archive," says Shuman.
All the CT data have to move swiftly, as does the image processing, to keep up with the ever-increasing expectations of physicians and departments waiting for the information. "What's happening now is that we are doing these 500 to 2,000-cut studies every six minutes — and just getting that out of the CT scanner and into the PACS becomes a potential bottleneck," he says, "so you've got to have a really fast communication software and network.
"It takes a few minutes per study for the data to transfer from server to server. If the network gets busy and things start slowing down, that affects our whole departmental productivity because studies aren't available to view in PACS for a period of time. It could take up to 30 minutes. That's just unacceptable in today's world."
UWMC is banking on the speed of multidetector CT as a major component in designing a newly planned wait-free ER. The work-in-progress facility will have no waiting room, and patients will immediately walk directly into an exam area. That's fast. Multidetector CT workflow has got to be a finely tuned part of the process.
Shands Hospital at the University of Florida (UF) is part of the larger Shands Healthcare, a multi-hospital enterprise with locations throughout the Southeast. The Shands at UF facility houses a Toshiba Aquilion 64-slice scanner. The system is now the "workhorse" for the facility, a Level 1 trauma center, says Meryll Frost, president and CEO of Medical Imaging Consultants. Frost oversees connectivity and workflow for the radiology department and designed the PACS integration and the entire information workflow for the department.
Like UWMC, Shands at UF confronts the issue of workflow to aid in trauma care, which includes CT workflow. "The whole problem with these multislice, large-volume scanners is they really impact your workflow. We do everything we can to deliver the images as fast as possible directly to trauma workstations," says Frost.
One major tactic at the facility has been the development of a "slab store," which involves a separate archive for the thin slices needed for the reconstructions for certain CT protocols. The slices are held for a year, after which they are generally purged off the system. This process relieves a lot of the pressure from the system and keeps the thin-slice data from bogging down the main archive. "It's transparent to the user, but that's what happens. It's a ramification of having this many slices," indicates Frost.
Further, the hospital has configured its system so that trauma users don't get any of the pre-fetched priors, for example, so that it doesn't burden the network with a lot of data not necessary in trauma situations. It's optimized to deliver slices directly from the scanner to the viewer so that they can take care of trauma patients more quickly, Frost says.
"Trauma doctors are not interested in looking at 8,000 thin slices; they just want to know 'is the spleen bleeding?,'" says Frost.
CT's future: Tool integration
Some facilities are looking at ways to streamline their CT-related workflow by aggressively pursuing technology that will aid them in integrating their disparate systems. The Department of Radiology at the NYU Medical Center — associated with the 726-bed Tisch Hospital and other hospitals and outpatient facilities in New York City — is one such example.
NYU Medical Center has a strategic alliance with Siemens Medical Solutions through which they have a Siemens Definition Dual Source 64 scanner and two SOMATOM Sensation 64-slice CT systems, one 40-slice system, and three 16-slice systems.
They have squeezed the existing technology and processes for as much efficiency as possible. "All along, our phil≠osophy has been to scan thin but archive thick," says Bernard Birnbaum, MD, professor of radiology, chief of service of the Department of Radiology, and vice-chair clinical affairs and operations.
And along the way, they have tweaked the processes in other ways. "The radiologists were acquiring such phenomenal 3D data sets that we started automatically generating coronal multi-planar reformatted images [or coronal MPRs] right from the scanners and sending them directly to PACS," says Birnbaum. They send the MPRs to the radiologists as well so they do not have to transfer to a workstation to create the 3D sets.
Beginning this month, they are participating in a market entry-phase evaluation of a new Siemens product called Webspace, a part of the company's syngo imaging suite of products. This will involve the implementation of a dedicated CT server where thin-section data will likely reside for several months. A radiologist will be able to use Webspace as a "web client-based server software application that would allow you to construct high-resolution 3D images on any PC, either onsite or off," says Birnbaum.
Webspace is designed to make CT data available for 2D, 3D, and 4D interactive reading with quick rendering at the server level — accessible from anywhere.
"This is going to be state-of-the-art: totally integrated PACS, RIS, 3D post-processing, speech recognition, and even document scanning technology," says Birnbaum. "If we were to roll this product out, a surgeon, for instance, could be sitting in an office and be able to manipulate the data which reside on a client server to generate images."
They are the first U.S. facility to test Webspace, although any full install of the system would not probably be until 2008 if the system is ready for "primetime," Birnbaum says.
Birnbaum believes this brings them closer to more fully utilizing the latest generation of CT scanners, which he regards fundamentally as "3D machines."
You don't need to worry that much about the latest industry buzz about 128- and 256-slice systems looming on the imaging horizon. If you already have a multidetector CT scanner, like an 8- or 16-slice system, for example, you've passed your biggest hurdle in data storage, management and workflow — although you should of course be prepared for new challenges with additional applications coming online.
Beyond that, for departments that serve trauma patients, the expectations are high for extremely quick turnaround. Sixty-four-slice scanners have proven themselves as being capable of churning out the data, but your network must be robust enough to handle it, and your workflow must be fine-tuned to speed critical images from scanner to expert interpretation to caregiver anytime, anywhere.