Diagnostic Workstations: Radiologys New Workhorses

At 20 years old, 3D is not exactly the new kid on the radiology block. But in the last few years 3D has evolved from a high-end, primarily academic means of producing pretty pictures to an essential clinical tool. “If you have multislice CT you need a 3D workstation,” states Steven Strobbe, MD, CEO, of Gulf Coast Medical Center in Port Richey, Fla. The leap is evident in the array of options from diagnostic modality workstations to third-party solutions and thin-client systems that can put 3D on every desktop.

With a wide range of hardware and software solutions, it’s easy to make a misstep. “Some sites decide what to buy without thinking through the purchase. They don’t buy what they need and then need to buy a second workstation,” notes Strobbe.

With price tags that stretch from the low six-figure range for a single workstation up to $1 million for an enterprise-wide deployment, it’s essential that radiology finds a way to harness 3D technology while meeting both workflow and financial needs.

This month Health Imaging & IT visits with a variety of users to determine the key factors to evaluate when investing in diagnostic workstations. Items to consider include workflow, features and functionality and the practice environment. The right solution can make a world of difference, enabling radiologists to read multidetector CT, MRI and other imaging studies quickly and efficiently, add diagnostic value to an array of studies and share key findings with clinicians.

Workstations, workflow and more

3D is second nature to Elliot K. Fishman, MD, director, diagnostic imaging and body CT at The Johns Hopkins University School of Medicine in Baltimore, Md. The radiologist has used a Siemens Medical Solutions Leonardo workstation in conjunction with Siemens 16 and 64-slice scanners for several years and completes 30 CT angiographies daily with the system. That type of workflow is no longer limited to the academic medical center; large hospitals, community hospitals and diagnostic imaging centers are entering the 3D realm en masse. “Transitioning from 2D to 3D is a whole new story. The challenge is building 3D into the workflow,” explains Fishman.

That challenge has held back the workstation market, says Strobbe, who relies on Vital Images Vitrea 2 workstation and ViTALConnect web solution for interpreting coronary CT angiograms. “All of the vendors are trying to make workstations simpler and more automated. Studies should be presented in a format that is quickly readable,” explains Strobbe.

While a single case can gobble up 45 to 60 minutes, the ideal is a four to five minute process. If it takes several mouse clicks or keyboarding to complete a task, the system may not be as fast or as interactive as it should be. Strobbe’s practice relies on 3D technologists to complete reconstructions in about 10 to 15 minutes; the radiologist or cardiologist reads and completes the case online via ViTAL Connect in five to 10 minutes. Fishman urges radiologists to avoid thinking about what the workstation can do and instead focus on what they can do with the workstation. “Bells and whistles are great, but radiologists should be able to do 99 percent of their work in a timely and efficient fashion,” he continues. For example, if the practice load is heavy on CT runoffs and it takes an hour to edit out the bones with one workstation, it isn’t the right solution.

“It’s important to remember that no two practices are the same. Think about what your business is, and make sure the system does it,” says Fishman. Cardiac studies require a different solution than 3D bones, says Strobbe. Another key angle is the future of the business. Essential questions to ask and answer include prior to investing in a solution include: what is your roadmap and where does 3D fit in?

Speed is essential, says James Gruden, MD, senior associate consultant radiologist, Mayo Clinic Arizona in Phoenix. Gruden uses GE Healthcare’s Advantage Workstation as his primary reading station. “Images load very quickly, and I can interact in real-time with the data,” notes Gruden. Processing power is critical with large datasets; a 3D CT angiography of the chest may comprise more than 3,000 images. The workstation must be capable of holding all of the images in its memory, and additional power is needed to display images in a cine loop.

The Advantage Workstation is essential to optimizing the site’s VCT 64-slice CT scanner. “It’s impossible to harness the benefits of VCT with a diagnostic workstation,” claims Gruden. The workstation enables the volumetric navigation and interpretation that is the hallmark of multidetector CT scanning. Volumetric reading is not only efficient but also helps radiologists make better interpretations for referring physicians, says Gruden. Improved interpretations are a function of the image quality of the 3D workstation, so 3D buyers need to evaluate image quality in addition to user-friendliness and efficiency. Radiologists in the market for a diagnostic workstation should be sure to load and review cases that aren’t ideal to get a feel for how the system handles these studies, says Fishman.

Gruden points to another plus of the workstation. “The repository is large enough to hold source data for a few weeks.” That means radiologists can return to the 500 or so 0.6 mm slices if a clinical questions arises. The practice saves 2.5 mm reconstructions on PACS.

Technical specifications can be persuasive, but they aren’t the final answer, says Fishman, who claims 3D is like driving a car. Fishman encourages a ‘test drive’ prior to purchase, where users determine whether or not a solution feels right. The final question in the analysis is service and training, says Fishman. Buyers should make sure that they will receive the needed support and training from the vendor.

Going places: thin clients and distributed 3D

Most, but not all, insiders predict that thin-clients are the wave of the future. “The client server approach is a good model because users aren’t tied to a single workstation,” explains Fishman, who uses Siemens WebSpace client server solution. Unlike conventional diagnostic workstations, the thin-client model reduces the demand on the workstation because processing is completed at the server.

“Radiologists can use a thin-client from almost anywhere, and all rendering is done on the server, so you don’t need a big memory on the client. You can complete very sophisticated post-processing on a very cheap computer,” adds Dominik Fleischmann, MD, assistant professor of radiology at Stanford University Medical Center in Stanford, Calif. Fleischmann relies on Terarecon’s AquariusNET server for 3D reading.

A host of physicians benefit from thin clients and web distribution. Radiologists can read 3D studies from nearly anywhere. In fact, Strobbe regularly uses ViTALConnect to interpret studies from a cabin in Kodiak Island, Alaska. The solution also allows the practice to distribute images to referring physicians, radiologists and cardiologists. Multi-site practices benefit from the thin-client model, too. Terarecon’s AquariusNET provides enterprise capabilities for Main Street Radiology, a three imaging center practice in Flushing, N.Y. Finally, on-call specialists such as neurosurgeons or orthopedic surgeons can rely on the thin-client arrangement to make expedient clinical decisions from home.

The thin-client model seems to carry obvious workflow, clinical and financial benefits, but it may not be necessary in every case. “Sites need to consider how many places they want 3D available. If there’s only one 3D post-processing location, it might be smarter to invest in a single workstation. If 3D is desired in three or four places, it’s probably wise to think about a server,” explains Fleischmann.

Swedish Medical Center in Denver, Colo., implemented Barco’s VOXAR 3D four years ago and plans to transition to VOXAR 3D Enterprise this fall. “The advantage of the enterprise approach is the integration with PACS. Radiologists won’t need to close out of PACS to launch and import images into a separate 3D application,” explains PACS Administrator Brian Phillips. Instead the radiologist can click on a VOXAR button in the Fujifilm Medical Systems Synapse PACS to access 3D tools. The goals for the implementation are simple; the department aims to improve workflow and increase the use of 3D.

The new system employs a server to temporarily hold images for 48 to 72 hours and distribute 3D power anywhere PACS is available. Swedish won’t be limited by the number or power of workstations, says Donald Davidson, director of medical imaging. “VOXAR 3D Enterprise is as efficient on my laptop as it is on a PACS workstation because the resources come from the server not the PC,” explains Davidson.

Another advantage of the central server approach is that it lowers the overall network load; images stay on the server, which means radiologists aren’t gobbling up bandwidth by importing and exporting 3D data. Phillips identifies a final benefit of the new configuration. “A radiologist can be interrupted in the middle of 3D and pick up right where he left off, or another radiologist can take that dataset and modify it.”

“There are challenges to the thin-client server approach,” asserts Strobbe. Wide distribution entails IT involvement. Firewall ports must be opened, and there are likely different rules for each firewall. What’s more, server-based architecture typically means the purchase of additional servers. Conventional wisdom might recommend five to 10 users per server, but it might be more efficient to have fewer than five users on a server. The ratio varies depending on applications and network speed. The network can be another thin-client challenge; a larger pipe correlates with faster distribution. A robust WAN that provides 6 megabits per second bandwidth can yield a fairly rapid interactive experience; slower bandwidth in the 3 to 4 megabits per second range requires compression, but the degree and speed of interaction is workable. In addition, images are not diagnostic quality; however, they do suffice in cases where an on-call specialist must make a rapid decision.

Another piece of the thin-client puzzle is the availability of advanced software modules; the most advanced 3D functions may not be available on standard desktops or laptops in the thin-client distribution model. Main Street Radiology uses Terarecon’s AquariusNET for thin-slice reconstructions of CT scans acquired on its Philips Medical Systems Brilliance 16-slice CT scanner. Terarecon’s Aquarius workstation equipped with clinical software provides the muscle for advanced applications including lung nodule analysis and CT colonography.

Stanford also employs a two-pronged approach to its 3D program. Radiologists use AquariusNET at the time of readout to answer clinical questions. Depending on the nature of the study, the radiologist completes volume rendering, maximum intensity projections and multiplanar reformations and saves snapshot images and movie clips to document findings to PACS. Readout findings also may be emailed to referring physicians. The second arm of the 3D program is the 3D lab, which is equipped with multiple workstations including Aquarius. Radiologists and techs use pre-set protocols to complete an official set of 3D images to send to PACS and referring physician. 

The software question

In the not-too-distant past diagnostic workstations came in a one-size-fits-all model. Today there is an array of software available to fine-tune the systems for every niche and nuance. Application-specific software streamlines and facilitates functions like virtual colonography, lung nodule analysis, cardiac CTA and more. But not every workstation needs every piece of sophisticated software, says Gruden. One station could be designated as the virtual colonography machine and another as the CTA workstation. The only hitch? Such an arrangement could complicate workflow.

Looking into the crystal ball

The 3D boom is sure to continue, and new options are on the horizon. The increasing use of higher-slice CT scanners translates into more CT angiography and coronary studies, says Fleischmann. “We’ll need more tools to interrogate these large datasets,” notes Fleishmann.

Fishman foresees additional application-specific software hitting the market to facilitate and expedite specialized applications. What’s more, artificial intelligence and second reading will be increasingly integrated into the diagnostic workstation. The final growth area is no surprise. The client-server model will continue to take off. In fact, the line separating thin-client solutions and diagnostic workstations will blur as vendors merge the two systems into a single platform.