Linda Novello, RT, a 3D imaging technologist in the Stanford 3D laboratory, removes the rib cage from a reconstructed CT data set.

Word is spreading about 3D imaging. As the often startlingly beautiful and complex images generated at pioneering laboratories increasingly make their way to conferences, where they are seen by physicians in a host of different specialties, community hospital radiology departments and other diagnostic imaging facilities will soon face—if they are not already—the problem of how to answer their physicians’ demands for such technology back home.

It is no secret that referring physicians are driving the demand for 3D images. “The primary drivers for new and refined clinical 3D protocols are the referring clinicians, who work with the 3D laboratory to tailor the analyses to meet their clinical needs,” notes Geoffrey D. Rubin, MD, director of the 3D laboratory at the Stanford University Department of Radiology (Stanford, Calif).

Adds Laura Logan Pierce, MPA, RT(CT), 3D laboratory manager, “If you want to retain surgeons, you have to invest in your infrastructure.” Pierce has been with the Stanford 3D laboratory since it was created 10 years ago and, as such, has had a front-row seat watching the development of the field. Yet despite the tremendous advances in 3D made possible by increasingly sophisticated technology, the problems related to developing and running a 3D laboratory have remained surprisingly consistent for her. “These challenges are the same as they were 10 years ago,” she says. How the Stanford 3D laboratory approaches these challenges offers lessons for any institution thinking about creating or expanding a 3D laboratory.

Life on “The Farm”

Stanford’s 3D laboratory operates out of space in two modern medical office buildings—the Richard M. Lucas Magnetic Resonance Imaging Center and the James H. Clark Center—in the northwest corner of the sprawling campus fondly referred to as “the farm” by students, staff, and alumni because it sits on the land once used for stock farming. The laboratory is still led by the two directors that founded it: Sandy Napel, PhD, and Rubin. Because both are also professors of radiology at the University, several students are usually working out of the laboratory, too. At any one time, Napel may have approximately 12 students—a mix of graduate students, physicians, and postdoctoral scholars—doing 3D research each year, and Rubin will have two international visiting medical physicians doing research each year. Christopher F. Beaulieu, MD, PhD, associate professor of radiology, also is part of the laboratory’s faculty.

However, the bulk of the 3D rendering done at the laboratory is performed by Pierce and her five 3D imaging staff technologists: Marc Sofilos, RT; Linda Novello, RT; Keshni Kumar, CRT; William Johnsen, RT(CV) RCIS; and its newest technologist, Noe Hinojosa, RT(CT). In addition, an administrative assistant, Lakeesha Winston, and a software engineer, Kala Raman, provide support. Because the 3D laboratory is run out of two different buildings, new and experienced staff rotate between the two buildings so that everyone has the opportunity to share ideas with and learn from everyone else.

It all adds up to a unique environment where research, clinical work, and teaching come together in one place. “Having all that happening together has really helped us grow,” Pierce says.

People Power

However, despite access to researchers and advanced medical resources, the Stanford 3D laboratory is no more immune to the problem of trained technologist shortages than any other institution. When Pierce needs to hire a new technologist, she has a long list of prequalifications that eliminate many potential candidates. First, technologists must have a good basis in the UNIX and Windows operating systems so that they can troubleshoot computer problems with the 3D software programs’ manufacturer support staff over the phone. Candidates also should have a background in CT, MRI, and cardiovascular diagnostic imaging, and an “excellent” knowledge of cross-sectional anatomy. This last qualification is one of the most critical and also one of the hardest to assess prior to hiring. It requires a special way of thinking about the human body in terms of volume to perform 3D imaging. “You can’t just think in the axial plane,” Pierce says.

In addition, she looks for candidates with a good comprehension of disease and disease pathology, because the technologist must be able to read the patient history and think critically about what questions the physician will need to answer to know which images to render in 3D. For example, in the case of a patient with possible pancreatic cancer, the physician will need more than just a picture of the lesion. He or she will want to know how far that lesion has metastasized and what other parts of the body are being invaded. Without knowing what the goal of the procedure is, the technologist cannot know which 3D images he or she will need to create, Pierce says.

Secret Weapon

Many elements of the Stanford 3D laboratory’s success can be reproduced—good equipment, teamwork, experience—but one of the factors is truly unique, and her name is Kala Raman. Originally attracted to Stanford from New Zealand nearly 6 years ago for another project, Raman switched over to the 3D laboratory after she came to know laboratory director Geoffrey Rubin, MD. Raman is a trained mathematician with a master’s degree in software engineering, and her analytical mind quickly saw how she would write software programs that could improve the laboratory operations. “I could automate factors and exclude the manual errors,” she explained.

Her software programs have not only made daily operations much easier by automating much of the workflow and billing, but they also have made the laboratory more research friendly by standardizing the way data is stored so that it can be searched more easily. Her programs also allow the radiologists and referring physicians to access the laboratory’s images and results more easily. Without sacrificing any Health Insurance Portability and Accountability Act (HIPAA) requirements, Raman’s software allows physicians to view their patients’ 3D quantitative results on any hospital PACS station and, in some cases, even through a special-access Web page or as an addendum to the patients’ radiology reports. Raman’s software is not integrated with the PACS, only with the radiology information system.

“She has enabled a lot of our progress in being able to run the lab efficiently and in being able to pursue the industrial contracts that we have for the clinical trials, because the basis of all that we do is accurate and effective record keeping that enables us to learn from how we are practicing,” Rubin says.

While Raman works hard to make the system very easy for anyone to use and adapt without her, she is still an integral part of keeping it going because she is constantly adapting and growing it as new technology becomes available and physicians’ preferences change. “I’m constantly changing and upgrading the software all the time,” she explains. “It is totally specific to this lab.”

Raman also makes sure the data backup is simple and secure. Plus, her work helps patients. In the old manual system, it was too much work to transcribe every measurement from a study into the patient record. With automation software, this is not a concern and it makes comparing two or more studies done on a patient at different times much easier. Raman’s databases store prior measurements and import them for comparison with present data instantly, with the click of a mouse. “That way, when a patient comes back for another study, we have those numbers there for comparison, so you can assess for any changes,” says Laura Logan Pierce, MPA, RT(CT), 3D laboratory manager . “It is just nice to have the numbers printed out on a piece of paper.”

—L. Kauffman

There is no 3D-technologist certification at this point, and Pierce knows of only one or two programs in the country that even include 3D imaging in radiology technologist training. On top of that, the field is so new that the recognition—and associated financial reward—for 3D imaging skills is not yet there to drive demand for 3D training programs. The pay, Pierce says, is about equal to what a senior MRI or CT technologist will make, so she understands why someone might not want to leave the comfort zone of these two established modalities and put in a lot of extra effort to learn new 3D skills.

Without organized education programs for 3D, Pierce has to train new hires from scratch—a process that can take 6 to 8 months of having the technologist work with trained staff, read educational documents, take skills-assessment exams on different procedures, and sit in on the radiologists’ meetings. Pierce is especially keen to have new hires learn from not just the other technologists but also the radiologists. One of the things that makes Stanford’s 3D laboratory unique, she says, is the close relationship between the technologists and the physicians. “The technologist needs to feel comfortable asking questions, and the radiologist needs to feel comfortable giving direction, and then it works. That is why we have been so successful here,” she says. “The technologist must be motivated and not be afraid to get in there and make mistakes and ask questions. I think that’s the hard part, because they leave their comfort zone in CT and MR and go into the unknown where, all of a sudden, they maybe aren’t so confident.”

Rubin adds, “One of the basic premises of our 3D lab is that it does have close integration and oversight by radiologists.”

Equipment Challenges

When it comes to equipment, buy the best you can, Pierce says. A 3D laboratory is no place to save a few dollars by choosing lower-resolution monitors or foregoing service agreements and software updates offered by the equipment manufacturers. In addition, you might need several different systems to be capable of processing any type of clinical exam, because some systems are better at certain exams than others, she says.

At Stanford, the 3D workstations include four Advantage Windows from GE Healthcare (Waukesha, Wis), two Vitrea from Vital Images Inc (Minnetonka, Minn), one Aquarius from TeraRecon Inc (San Mateo, Calif), two systems from AccuImage (San Francisco), and one Leonardo from Siemens Medical Solutions (Malvern, Pa). “We will sit down at whatever workstation will work the best,” Pierce says, although she will not say which one she personally prefers for which procedures.

Rendering techniques include curved planar, oblique, and multiplanar reconstruction; maximum- and minimum-intensity projections (MIP and minIP); virtual endoscopy; and volume rendering. Variable thick slabs are available in any plane in MIP, minIP average, and volume renderings.

The laboratory also houses three TeraRecon thin client servers for remote 3D imaging and two research servers for storage of data and images. In addition, it uses a variety of personal computers and some legacy SGI Unix workstations for software development and support. An up-to-date list of equipment used in the laboratory is online at 3dradiology.stanford.edu/tour.html.

Finally, some quality printers are useful even though today’s 3D medical images are really designed to be reviewed on a computer. The reason, Pierce says, is that some physicians like to keep printouts of 3D images in the patients’ files for later reference and also to show patients the study results. However, because high-resolution color printing quickly can become costly, Pierce provides these printouts only when they are specifically requested.

Preventing Data Overload

Generating 3D scans can require thousands of submillimeter 2D slices, which add up to massive amounts of data that must be transferred and stored. At Stanford, the 3D laboratory stores most of the image data in the hospital’s PACS. “We network the CT and MRI images for postprocessing directly from the hospital’s GE PACS system,” Pierce explains. “We also network the 3D secondary screen captures directly back to PACS for radiologist interpretation with the source images.”

The staff employs a high-speed network within the laboratory; however, to get the data to the laboratory, they must rely on an older and slower 1,000-megabit/second switched network from the hospital. This can slow data transfer to 45 minutes to an hour. Their solution is to begin importing the data for the next patient while they work on the first patient’s 3D images.

Another solution might be to move the technologists over to the hospital, but Pierce says having the 3D laboratory off-site carries an important advantage: peace and quiet. Radiology is a cerebral discipline often requiring high levels of critical thinking, and this is especially true for 3D, she says. Therefore, she recommends placing 3D laboratories as far away from the hustle and bustle of the daily work of medicine as possible. “You want the technologists not to be interrupted,” she says.

Managing Workflow

Communicating which patients should have 3D imaging and in what order the procedures are done is a challenge for any laboratory. At Stanford, an in-house software engineer (see “Secret Weapon“) allowed the laboratory to create a system for reviewing all of the 200 to 250 exams performed daily at Stanford Hospital and selecting those it will slate for 3D imaging. On average, this ends up being about 10% to 12% of the total number of scans performed at the hospital.

Want to Study at Stanford?

The Stanford 3D Laboratory offers two types of visiting fellowships: a 3D postprocessing fellowship and a 3D laboratory-operations fellowship. Each can be individualized for specific areas of interest. One-on-one instruction runs around $100 per hour, and observation runs about $50 per hour.

“[3D] shouldn’t just be limited to academic centers, such as Stanford,” says Laura Logan Pierce, MPA, RT(CT), 3D laboratory manager. “We want to make it available to everyone.”

The laboratory typically trains one to three fellows per month.

Details are available online at 3dradiology.stanford.edu/education/ fellowships.html.

The procedures are selected by a trained technologist who comes in at 6:30 am to review all of the scans scheduled for that day and choose the ones that will need to be sent to the laboratory for 3D rendering. “We are very careful about not choosing and doing 3D on a patient that does not need it,” Pierce says. “We do not want to be charging someone for something that is not necessary.”

The 3D imaging is done in order of selection, unless the laboratory has received a special request to prioritize the work on one particular patient. This happens most frequently when a patient has been referred to Stanford from another regional hospital and is waiting at the hospital for the results.

The laboratory aims for a 2- to 3-hour turnaround time, which means that keeping an eye on the clock is key. With 3D imaging, there is a great temptation to keep working on an image to create a really beautiful picture even if that extra detail in the picture is not required for the physician to diagnose and treat the patient. “Our job is really to answer the questions the clinician is asking for the purpose of the exam,” Pierce says.

To help guide the technologists, the laboratory uses a procedure manual that Pierce is responsible for updating. Currently, it contains about 35 3D protocols, including MRI and CT, as well as quantitative data instructions.

The radiologists, Rubin says, also help facilitate communication between the referring physicians and the laboratory. “When we develop visualization protocols, they are going to be useful to referring physicians as well as to us as radiologists,” he explains.

Keeping the Drive Going

With the growth of 3D imaging, it is not unusual for people doing 3D to feel somewhat overwhelmed—and the case is no different at Stanford. Since September 1, 1996, when the laboratory was founded, the number of 3D imaging exams performed each year has kept growing 25% to 30% per year. “I keep expecting it to plateau, and it hasn’t,” Pierce says. Much of the growth is linked to the growth of CT, but MRI also composes about 10% of the volume. The Stanford laboratory will do a little more than 8,000 scans in 2006, Pierce estimates. The largest segment of these will be for cardiovascular-related imaging.

More exams means a need for more technologists, more equipment, and, hardest of all, more time, and when a 3D technologist feels overwhelmed, it can be hard to remain enthusiastic about learning new protocols and keeping the laboratory’s services up to date. This is where having a good medical director that keeps the focus on employing the best science for the benefit of the patient is key, Pierce says. “You need good direction,” she explains. “You need someone pushing you.”

For Pierce, Rubin and Napel perfectly embody this role. The laboratory’s vision is multifaceted, according to Rubin. It is focused not only on the clinical side, but also on research and education. The types of studies the laboratory is involved with is largely driven by the interests of the researchers at Stanford and the availability of funding. “It is really a broad spectrum of activity,” Rubin says. Studies being done include work on novel graphics approaches as well as computer analysis of quantitative data and computer-assisted detection.

Because of the laboratory’s drive to push the field of 3D forward, its protocols and procedures are constantly evolving. “If we simply continued to provide the same service we did in 1996, then we would stagnate,” Rubin explains. “Our technologists would probably lose interest, and our referring physicians would lose interest in the results. I think it is very important on the clinical side to make sure we are always providing a value added with whatever we are doing.”

Lena Kauffman is a contributing writer for Medical Imaging.